Tech Guides

Learning something by rote i.e., repeating it many times, perfecting a skill by practising it over and over again or building something by making minor adjustments progressively to a prototype are things that comes to us naturally as human beings. Machines can also learn this way and this is called ‘Iterative machine learning’. In most cases, iteration is an efficient learning approach that helps reach the desired end results faster and accurately without becoming a resource crunch nightmare. Now, you might wonder, isn’t iteration inherently part of any kind of machine learning? In other words, modern day machine learning techniques across the spectrum from basic regression analysis, decision trees, Bayesian networks, to advanced neural nets and deep learning algorithms have some inherent iterative component built into them. What is the need, then, for discussing iterative learning as a standalone topic? This is simply because introducing iteration externally to an algorithm can minimize the error margin and therefore help in accurate modelling.  How Iterative Learning works Let’s understand how iteration works by looking closely at what happens during a single iteration flow within a machine learning algorithm. A pre-processed training dataset is first introduced into the model. After processing and model building with the given data, the model is tested, and then the results are matched with the desired result/expected output. The feedback is then returned back to the system for the algorithm to further learn and fine tune its results. This clearly shows that two iteration processes take place here: Data Iteration - Inherent to the algorithm Model Training Iteration - Introduced externally Now, what if we did not feedback the results into the system i.e. did not allow the algorithm to learn iteratively but instead adopted a sequential approach? Would the algorithm work and would it provide the right results? Yes, the algorithm would definitely work. However, the quality of the results it produces is going to vary vastly based on a number of factors. The quality and quantity of the training dataset, the feature definition and extraction techniques employed, the robustness of the algorithm itself are among many other factors. Even if all of the above were done perfectly, there is still no guarantee that the results produced by a sequential approach will be highly accurate. In short, the results will neither be accurate nor reproducible. Iterative learning thus allows algorithms to improve model accuracy. Certain algorithms have iteration central to their design and can be scaled as per the data size. These algorithms are at the forefront of machine learning implementations because of their ability to perform faster and better. In the following sections we will discuss iteration in different sets of algorithms each from the three main machine learning approaches - supervised ML, unsupervised ML and reinforcement learning. The Boosting algorithms: Iteration in supervised ML The boosting algorithms, inherently iterative in nature, are a brilliant way to improve results by minimizing errors. They are primarily designed to reduce bias in results and transform a particular set of weak learning classifier algorithms to strong learners and to enable them to reduce errors. Some examples are: AdaBoost (Adaptive Boosting) Gradient Tree Boosting XGBoost How they work All boosting algorithms have a common classifiers which are iteratively modified to reach the desired result. Let’s take the example of finding cases of plagiarism in a certain article. The first classifier here would be to find a group of words that appear somewhere else or in another article which would result in a red flag. If we create 10 separate group of words and term them as classifiers 1 to 10, then our article will be checked on the basis of this classifier and any possible matches will be red flagged. But no red flags with these 10 classifiers would not mean a definite 100% original article. Thus, we would need to update the classifiers, create shorter groups perhaps based on the first pass and improve the accuracy with which the classifiers can find similarity with other articles. This iteration process in Boosting algorithms eventually leads us to a fairly high rate of accuracy. The reason being after each iteration, the classifiers are updated based on their performance. The ones which have close similarity with other content are updated and tweaked so that we can get a better match. This process of improving the algorithm inherently, is termed as boosting and is currently one of the most popular methods in Supervised Machine Learning. Strengths & weaknesses The obvious advantage of this approach is that it allows minimal errors in the final model as the iteration enables the model to correct itself every time there is an error. The downside is the higher processing time and the overall memory requirement for a large number of iterations. Another important aspect is that the error fed back to train the model is done externally, which means the supervisor has control over the model and how it modifies. This in turn has a downside that the model doesn’t learn to eliminate error on its own. Hence, the model is not reusable with another set of data. In other words, the model does not learn how to become error-free by itself and hence cannot be ported to another dataset as it would need to start the learning process from scratch. Artificial Neural Networks: Iteration in unsupervised ML Neural Networks have become the poster child for unsupervised machine learning because of their accuracy in predicting data models. Some well known neural networks are: Convolutional Neural Networks   Boltzmann Machines Recurrent Neural Networks Deep Neural Networks Memory Networks How they work Artificial neural networks are highly accurate in simulating data models mainly because of their iterative process of learning. But this process is different from the one we explored earlier for Boosting algorithms. Here the process is seamless and natural and in a way it paves the way for reinforcement learning in AI systems. Neural Networks consist of electronic networks simulating the way the human brain is works. Every network has an input and output node and in-between hidden layers that consist of algorithms. The input node is given the initial data set to perform a set of actions and each iteration creates a result that is output as a string of data. This output is then matched with the actual result dataset and the error is then fed back to the input node. This error then enables the algorithms to correct themselves and reach closer and closer to the actual dataset. This process is called training the Neural Networks and each iteration improve the accuracy. The key difference between the iteration performed here as compared to how it is performed by Boosting algorithms is that here we don’t have to update the classifiers manually, the algorithms change themselves based on the error feedback. Strengths & weaknesses The main advantage of this process is obviously the level of accuracy that it can achieve on its own. The model is also reusable because it learns the means to achieve accuracy and not just gives you a direct result. The flip side of this approach is that the models can go wrong heavily and deviate completely in a different direction. This is because the induced iteration takes its own course and doesn’t need human supervision. The facebook chat-bots deviating from their original goal and communicating within themselves in a language of their own is a case in point. But as is the saying, smart things come with their own baggage. It’s a risk we would have to be ready to tackle if we want to create more accurate models and smarter systems.    Reinforcement Learning Reinforcement learning is a interesting case of machine learning where the simple neural networks are connected and together they interact with the environment to learn from their mistakes and rewards. The iteration introduced here happens in a complex way. The iteration happens in the form of reward or punishment for arriving at the correct or wrong results respectively. After each interaction of this kind, the multilayered neural networks incorporate the feedback, and then recreate the models for better accuracy. The typical type of reward and punishment method somewhat puts it in a space where it is neither supervised nor unsupervised, but exhibits traits of both and also has the added advantage of producing more accurate results. The con here is that the models are complex by design. Multilayered neural networks are difficult to handle in case of multiple iterations because each layer might respond differently to a certain reward or punishment. As such it may create inner conflict that might lead to a stalled system - one that can’t decide which direction to move next. Some Practical Implementations of Iteration Many modern day machine learning platforms and frameworks have implemented the iteration process on their own to create better data models, Apache Spark and MapR are two such examples. The way the two implement iteration is technically different and they have their merits and limitations. Let’s look at MapReduce. It reads and writes data directly onto HDFS filesystem present on the disk. Note that for every iteration to be read and written from the disk needs significant time. This in a way creates a more robust and fault tolerant system but compromises on the speed. On the other hand, Apache Spark stores the data in memory (Resilient Distributed DataSet) i.e. in the RAM. As a result, each iteration takes much less time which enables Spark to perform lightning fast data processing. But the primary problem with the Spark way of doing iteration is that dynamic memory or RAM is much less reliable than disk memory to store iteration data and perform complex operations. Hence it’s much less fault tolerant that MapR.   Bringing it together To sum up the discussion, we can look at the process of iteration and its stages in implementing machine learning models roughly as follows: Parameter Iteration: This is the first and inherent stage of iteration for any algorithm. The parameters involved in a certain algorithm are run multiple times and the best fitting parameters for the model are finalized in this process. Data Iteration: Once the model parameters are finalized, the data is put into the system and the model is simulated. Multiple sets of data are put into the system to check the parameters’ effectiveness in bringing out the desired result. Hence, if data iteration stage suggests that some of the parameters are not well suited for the model, then they are taken back to the parameter iteration stage and parameters are added or modified. Model Iteration: After the initial parameters and data sets are finalized, the model testing/ training happens. The iteration in model testing phase is all about running the same model simulation multiple times with the same parameters and data set, and then checking the amount of error, if the error varies significantly in every iteration, then there is something wrong with either the data or the parameter or both. Iterations are done to data and parameters until the model achieves accuracy.   Human Iteration: This step involves the human induced iteration where different models are put together to create a fully functional smart system. Here, multiple levels of fitting and refitting happens to achieve a coherent overall goal such as creating a driverless car system or a fully functional AI. Iteration is pivotal to creating smarter AI systems in the near future. The enormous memory requirements for performing multiple iterations on complex data sets continue to pose major challenges. But with increasingly better AI chips, storage options and data transfer techniques, these challenges are getting easier to handle. We believe iterative machine learning techniques will continue to lead the transformation of the AI landscape in the near future.  

The Transport layer is responsible for end-to-end data communication and acts as an interface for network applications to access the network. This layer also takes care of error checking, flow control, and verification in the TCP/IP  protocol suite. The Application Layer handles the details of a particular application and performs 3 main tasks- formatting data, presenting data and transporting data.  In this tutorial, we will explore the different types of vulnerabilities in the Application and Transport Layer. This article is an excerpt from a book written by Glen D. Singh, Rishi Latchmepersad titled CompTIA Network+ Certification Guide This book covers all CompTIA certification exam topics in an easy-to-understand manner along with plenty of self-assessment scenarios for better preparation. This book will not only prepare you conceptually but will also help you pass the N10-007 exam. Vulnerabilities in the Application Layer The following are some of the application layer protocols which we should pay close attention to in our network: File Transfer Protocol (FTP) Telnet Secure Shell (SSH) Simple Mail Transfer Protocol (SMTP) Domain Name System (DNS) Dynamic Host Configuration Protocol (DHCP) Hypertext Transfer Protocol (HTTP) Each of these protocols was designed to provide the function it was built to do and with a lesser focus on security. Malicious users and hackers are able to compromise both the application that utilizes these protocols and the network protocols themselves. Cross Site Scripting (XSS) XSS focuses on exploiting a weakness in websites. In an XSS attack, the malicious user or hacker injects client-side scripts into a web page/site that a potential victim would trust. The scripts can be JavaScript, VBScript, ActiveX, and HTML, or even Flash (ActiveX), which will be executed on the victim's system. These scripts will be masked as legitimate requests between the web server and the client's browser. XSS focuses on the following: Redirecting a victim to a malicious website/server Using hidden Iframes and pop-up messages on the victim's browser Data manipulation Data theft Session hijacking Let's take a deeper look at what happens in an XSS attack: An attacker injects malicious code into a web page/site that a potential victim trusts. A trusted site can be a favorite shopping website, social media platform, or school or university web portal. A potential victim visits the trusted site. The malicious code interacts with the victim's web browser and executes. The web browser is usually unable to determine whether the scripts are malicious or not and therefore still executes the commands. The malicious scripts can be used obtain cookie information, tokens, session information, and so on about other websites that the browser has stored information about. The acquired details (cookies, tokens, sessions ID, and so on) are sent back to the hacker, who in turn uses them to log in to the sites that the victim's browser has visited: There are two types of XSS attacks: Stored XSS (persistent) Reflected (non-persistent) Stored XSS (persistent): In this attack, the attacker injects a malicious script directly into the web application or a website. The script is stored permanently on the page, so when a potential victim visits the compromised page, the victim's web browser will parse all the code of the web page/application fine. Afterward, the script is executed in the background without the victim's knowledge. At this point, the script is able to retrieve session cookies, passwords, and any other sensitive information stored in the user's web browser, and sends the loot back to the attacker in the background. Reflective XSS (non-persistent): In this attack, the attacker usually sends an email with the malicious link to the victim. When the victim clicks the link, it is opened in the victim's web browser (reflected), and at this point, the malicious script is invoked and begins to retrieve the loot (passwords, credit card numbers, and so on) stored in the victim's web browser. SQL injection (SQLi) SQLi attacks focus on parsing SQL commands into an SQL database that does not validate the user input. The attacker attempts to gain unauthorized access to a database either by creating or retrieving information stored in the database application. Nowadays, attackers are not only interested in gaining access, but also in retrieving (stealing) information and selling it to others for financial gain. SQLi can be used to perform: Authentication bypass: Allows the attacker to log in to a system without a valid user credential Information disclosure: Retrieves confidential information from the database Compromise data integrity: The attacker is able to manipulate information stored in the database Lightweight Directory Access Protocol (LDAP) injection LDAP is designed to query and update directory services, such as a database like Microsoft Active Directory. LDAP uses both TCP and UDP port 389 and LDAP uses port 636. In an LDAP injection attack, the attacker exploits the vulnerabilities within a web application that constructs LDAP messages or statements, which are based on the user input. If the receiving application does not validate or sanitize the user input, this increases the possibility of manipulating LDAP messages. Cross-Site Request Forgery (CSRF) This attack is a bit similar to the previously mentioned XSS attack. In a CSRF attack, the victim machine/browser is forced to execute malicious actions against a website with which the victim has been authenticated (a website that trusts the actions of the user). To have a better understanding of how this attack works, let's visualize a potential victim, Bob. On a regular day, Bob visits some of his favorite websites, such as various blogs, social media platforms, and so on, where he usually logs in automatically to view the content. Once Bob logs in to a particular website, the website would automatically trust the transactions between itself and the authenticated user, Bob. One day, he receives an email from the attacker but unfortunately Bob does not realize the email is a phishing/spam message and clicks on the link within the body of the message. His web browser opens the malicious URL in a new tab: The attack would cause Bob's machine/web browser to invoke malicious actions on the trusted website; the website would see all the requests are originating from Bob. The return traffic such as the loot (passwords, credit card details, user account, and so on) would be returned to the attacker. Session hijacking When a user visits a website, a cookie is stored in the user's web browser. Cookies are used to track the user's preferences and manage the session while the user is on the site. While the user is on the website, a session ID is also set within the cookie, and this information may be persistent, which allows a user to close the web browser and then later revisit the same website and automatically log in. However, the web developer can set how long the information is persistent for, whether it expires after an hour or a week, depending on the developer's preference. In a session hijacking attack, the attacker can attempt to obtain the session ID while it is being exchanged between the potential victim and the website. The attacker can then use this session ID of the victim on the website, and this would allow the attacker to gain access to the victim's session, further allowing access to the victim's user account and so on. Cookie poisoning A cookie stores information about a user's preferences while he/she is visiting a website. Cookie poisoning is when an attacker has modified a victim's cookie, which will then be used to gain confidential information about the victim such as his/her identity. DNS Distributed Denial-of-Service (DDoS) A DDoS attack can occur against a DNS server. Attacker sometimes target Internet Service Providers (ISPs) networks, public and private Domain Name System (DNS) servers, and so on to prevent other legitimate users from accessing the service. If a DNS server is unable to handle the amount of requests coming into the server, its performance will eventually begin to degrade gradually, until it either stops responding or crashes. This would result in a Denial-of-Service (DoS) attack. Registrar hijacking Whenever a person wants to purchase a domain, the person has to complete the registration process at a domain registrar. Attackers do try to compromise users accounts on various domain registrar websites in the hope of taking control of the victim's domain names. With a domain name, multiple DNS records can be created or modified to direct incoming requests to a specific device. If a hacker modifies the A record on a domain to redirect all traffic to a compromised or malicious server, anyone who visits the compromised domain will be redirected to the malicious website. Cache poisoning Whenever a user visits a website, there's the process of resolving a host name to an IP address which occurs in the background. The resolved data is stored within the local system in a cache area. The attacker can compromise this temporary storage area and manipulate any further resolution done by the local system. Typosquatting McAfee outlined typosquatting, also known as URL hijacking, as a type of cyber-attack that allows an attacker to create a domain name very close to a company's legitimate domain name in the hope of tricking victims into visiting the fake website to either steal their personal information or distribute a malicious payload to the victim's system. Let's take a look at a simple example of this type of attack. In this scenario, we have a user, Bob, who frequently uses the Google search engine to find his way around the internet. Since Bob uses the www.google.com website often, he sets it as his homepage on the web browser so each time he opens the application or clicks the Home icon, www.google.com is loaded onto the screen. One day Bob decides to use another computer, and the first thing he does is set his favorite search engine URL as his home page. However, he typed www.gooogle.com and didn't realize it. Whenever Bob visits this website, it looks like the real website. Since the domain was able to be resolved to a website, this is an example of how typosquatting works. It's always recommended to use a trusted search engine to find a URL for the website you want to visit. Trusted internet search engine companies focus on blacklisting malicious and fake URLs in their search results to help protect internet users such as yourself. Vulnerabilities at the Transport Layer In this section, we are going to discuss various weaknesses that exist within the underlying protocols of the Transport Layer. Fingerprinting In the cybersecurity world, fingerprinting is used to discover open ports and services that are running open on the target system. From a hacker's point of view, fingerprinting is done before the exploitation phase, as the more information a hacker can obtain about a target, the hacker can then narrow its attack scope and use specific tools to increase the chances of successfully compromising the target machine. This technique is also used by system/network administrators, network security engineers, and cybersecurity professionals alike. Imagine you're a network administrator assigned to secure a server; apart from applying system hardening techniques such as patching and configuring access controls, you would also need to check for any open ports that are not being used. Let's take a look at a more practical approach to fingerprinting in the computing world. We have a target machine, 10.10.10.100, on our network. As a hacker or a network security professional, we would like to know which TCP and UDP ports are open, the services that use the open ports, and the service daemon running on the target system. In the following screenshot, we've used nmap to help us discover the information we are seeking. The NMap tools delivers specially crafted probes to a target machine: Enumeration In a cyber attack, the hacker uses enumeration techniques to extract information about the target system or network. This information will aid the attacker in identifying system attack points. The following are the various network services and ports that stand out for a hacker: Port 53: DNS zone transfer and DNS enumeration Port 135: Microsoft RPC Endpoint Mapper Port 25: Simple Mail Transfer Protocol (SMTP) DNS enumeration DNS enumeration is where an attacker is attempting to determine whether there are other servers or devices that carry the domain name of an organization. Let's take a look at how DNS enumeration works. Imagine we are trying to find out all the publicly available servers Google has on the internet. Using the host utility in Linux and specifying a hostname, host www.google.com, we can see the IP address 172.217.6.196 has been resolved successfully. This means there's a device with a host name of www.google.com active. Furthermore, if we attempt to resolve the host name, gmail.google.com, another IP address is presented but when we attempt to resolve mx.google.com, no IP address is given. This is an indication that there isn't an active device with the mx.google.com host name: DNS zone transfer DNS zone transfer allows the copying of the master file from a DNS server to another DNS server. There are times when administrators do not configure the security settings on their DNS server properly, which allows an attacker to retrieve the master file containing a list of the names and addresses of a corporate network. Microsoft RPC Endpoint Mapper Not too long ago, CVE-2015-2370 was recorded on the CVE database. This vulnerability took advantage of the authentication implementation of the Remote Procedure Call (RPC) protocol in various versions of the Microsoft Windows platform, both desktop and server operating systems. A successful exploit would allow an attacker to gain local privileges on a vulnerable system. SMTP SMTP is used in mail servers, as with the POP and the Internet Message Access Protocol (IMAP). SMTP is used for sending mail, while POP and IMAP are used to retrieve mail from an email server. SMTP supports various commands, such as EXPN and VRFY. The EXPN command can be used to verify whether a particular mailbox exists on a local system, while the VRFY command can be used to validate a username on a mail server. An attacker can establish a connection between the attacker's machine and the mail server on port 25. Once a successful connection has been established, the server will send a banner back to the attacker's machine displaying the server name and the status of the port (open). Once this occurs, the attacker can then use the VRFY command followed by a user name to check for a valid user on the mail system using the VRFY bob syntax. SYN flooding One of the protocols that exist at the Transport Layer is TCP. TCP is used to establish a connection-oriented session between two devices that want to communication or exchange data. Let's recall how TCP works. There are two devices that want to exchange some messages, Bob and Alice. Bob sends a TCP Synchronization (SYN) packet to Alice, and Alice responds to Bob with a TCP Synchronization/Acknowledgment (SYN/ACK) packet. Finally, Bob replies with a TCP Acknowledgement (ACK) packet. The following diagram shows the TCP 3-Way Handshake mechanism: For every TCP SYN packet received on a device, a TCP ACK packet must be sent back in response. One type of attack that takes advantage of this design flaw in TCP is known as a SYN Flood attack. In a SYN Flood attack, the attacker sends a continuous stream of TCP SYN packets to a target system. This would cause the target machine to process each individual packet and response accordingly; eventually, with the high influx of TCP SYN packets, the target system will become too overwhelmed and stop responding to any requests: TCP reassembly and sequencing During a TCP transmission of datagrams between two devices, each packet is tagged with a sequence number by the sender. This sequence number is used to reassemble the packets back into data. During the transmission of packets, each packet may take a different path to the destination. This may cause the packets to be received in an out-of-order fashion, or in the order they were sent over the wire by the sender. An attacker can attempt to guess the sequencing numbers of packets and inject malicious packets into the network destined for the target. When the target receives the packets, the receiver would assume they came from the real sender as they would contain the appropriate sequence numbers and a spoofed IP address. Summary In this article, we have explored the different types of vulnerabilities that exist at the Application and Transport Layer of the TCP/IP protocol suite. To understand other networking concepts like network architecture, security, network monitoring, and troubleshooting; and ace the CompTIA certification exam, check out our book CompTIA Network+ Certification Guide AWS announces more flexibility its Certification Exams, drops its exam prerequisites Top 10 IT certifications for cloud and networking professionals in 2018 What matters on an engineering resume? Hacker Rank report says skills, not certifications

TensorFlow is an open source machine learning framework for carrying out high-performance numerical computations. It provides excellent architecture support which allows easy deployment of computations across a variety of platforms ranging from desktops to clusters of servers, mobiles, and edge devices. Have you ever thought, why TensorFlow has become so popular in such a short span of time? What made TensorFlow so special, that we seeing a huge surge of developers and researchers opting for the TensorFlow framework? Interestingly, when it comes to artificial intelligence frameworks showdown, you will find TensorFlow emerging as a clear winner most of the time. The major credit goes to the soaring popularity and contributions across various forums such as GitHub, Stack Overflow, and Quora. The fact is, TensorFlow is being used in over 6000 open source repositories showing their roots in many real-world research and applications. How TensorFlow came to be The library was developed by a group of researchers and engineers from the Google Brain team within Google AI organization. They wanted a library that provides strong support for machine learning and deep learning and advanced numerical computations across different scientific domains. Since the time Google open sourced its machine learning framework in 2015, TensorFlow has grown in popularity with more than 1500 projects mentions on GitHub. The constant updates made to the TensorFlow ecosystem is the real cherry on the cake. This has ensured all the new challenges developers and researchers face are addressed, thus easing the complex computations and providing newer features, promises, and performance improvements with the support of high-level APIs. By open sourcing the library, the Google research team have received all the benefits from a huge set of contributors outside their existing core team. Their idea was to make TensorFlow popular by open sourcing it, thus making sure all new research ideas are implemented in TensorFlow first allowing Google to productize those ideas. Read Also: 6 reasons why Google open sourced TensorFlow What makes TensorFlow different from the rest? With more and more research and real-life use cases going mainstream, we can see a big trend among programmers, and developers flocking towards the tool called TensorFlow. The popularity for TensorFlow is quite evident, with big names adopting TensorFlow for carrying out artificial intelligence tasks. Many popular companies such as NVIDIA, Twitter, Snapchat, Uber and more are using TensorFlow for all their major operations and research areas. On one hand, someone can make a case that TensorFlow’s popularity is based on its origins/legacy. TensorFlow being developed under the house of “Google” enjoys the reputation of the household name. There’s no doubt, TensorFlow has been better marketed than some of its competitors. Source: The Data Incubator However that’s not the full story. There are many other compelling reasons why small scale to large scale companies prefer using TensorFlow over other machine learning tools TensorFlow key functionalities TensorFlow provides an accessible and readable syntax which is essential for making these programming resources easier to use. The complex syntax is the last thing developers need to know given machine learning’s advanced nature. TensorFlow provides excellent functionalities and services when compared to other popular deep learning frameworks. These high-level operations are essential for carrying out complex parallel computations and for building advanced neural network models. TensorFlow is a low-level library which provides more flexibility. Thus you can define your own functionalities or services for your models. This is a very important parameter for researchers because it allows them to change the model based on changing user requirements. TensorFlow provides more network control. Thus allowing developers and researchers to understand how operations are implemented across the network. They can always keep track of new changes done over time. Distributed training The trend of distributed deep learning began in 2017, when Facebook released a paper showing a set of methods to reduce the training time of a convolutional neural network model. The test was done on RESNET-50 model on ImageNet dataset which took one hour to train instead of two weeks. 256 GPUs spread over 32 servers were used. This revolutionary test has open the gates for many research work which have massively reduced the experimentation time by running many tasks in parallel on multiple GPUs. Google’s distributed TensorFlow has allowed all the researchers and developers to scale out complex distributed training using in-built methods and operations that optimizes distributed deep learning among servers. . Google’s distributed TensorFlow engine which is part of the regular TensorFlow repo, works exceptionally well with the existing TensorFlow’s operations and functionalities. It has allowed exploring two of the most important distributed methods: Distribute the training time of a neural network model over many servers to reduce the training time. Searching for good hyperparameters by running parallel experiments over multiple servers. Google has given distributed TensorFlow engine the required power to steal the share of the market acquired by other distributed projects such as Microsoft’s CNTK, AMPLab's SparkNet, and CaffeOnSpark. Even though the competition is tough, Google has still managed to become more popular when compared to the other alternatives in the market. From research to production Google has, in some ways, democratized deep learning., The key reason is TensorFlow’s high-level APIs making deep learning accessible to everyone. TensorFlow provides pre-built functions and advanced operations to ease the task of building different neural network models. It provides the required infrastructure and hardware which makes them one of the leading libraries used extensively by researchers and students in the deep learning domain. In addition to research tools, TensorFlow extends the services by bringing the model in production using TensorFlow Serving. It is specifically designed for production environments, which provides a flexible, high-performance serving system for machine learning models. It provides all the functionalities and operations which makes it easy to deploy new algorithms and experiments as per changing requirements and preferences. It provides an excellent feature of out-of-the-box integration with TensorFlow models which can be easily extended to serve other types of models and data. TensorFlow’s API is a complete package which is easier to use and read, plus provides helpful operators, debugging and monitoring tools, and deployment features. This has led to growing use of TensorFlow library as a complete package within the ecosystem by the emerging body of students, researchers, developers, production engineers from various fields who are gravitating towards artificial intelligence. There is a TensorFlow for web, mobile, edge, embedded and more TensorFlow provides a range of services and modules within their existing ecosystem making them as one of the ground-breaking end-to-end tools to provide state-of-the-art deep learning. TensorFlow.js for machine learning on the web JavaScript library for training and deploying machine learning models in the browser. This library provides flexible and intuitive APIs to build and train new and pre-existing models from scratch right in the browser or under Node.js. TensorFlow Lite for mobile and embedded ML It is a TensorFlow lightweight solution used for mobile and embedded devices. It is fast since it enables on-device machine learning inference with low latency. It supports hardware acceleration with the Android Neural Networks API. The future releases of TensorFlow Lite will bring more built-in operators, performance improvements, and will support more models to simplify the developer’s experience of bringing machine learning services within mobile devices. TensorFlow Hub for reusable machine learning A library which is used extensively to reuse machine learning models. Thus you can transfer learning by reusing parts of machine learning models. TensorBoard for visual debugging While training a complex neural network model, the computations you use in TensorFlow can be very confusing. TensorBoard makes it very easy to understand and debug your TensorFlow programs in the form of visualizations. It allows you to easily inspect and understand your TensorFlow runs and graphs. Sonnet Sonnet is a DeepMind library which is built on top of TensorFlow extensively used to build complex neural network models. All of this factors have made the TensorFlow library immensely appealing for building a wide spectrum of machine learning and deep learning projects. This tool has become a preferred choice for everyone from space research giant NASA and other confidential government agencies, to an impressive roster of private sector giants. Road Ahead for TensorFlow TensorFlow no doubt is better marketed compared to the other deep learning frameworks. The community appears to be moving very fast. In any given hour, there are approximately 10 people around the world contributing or improving the TensorFlow project on GitHub. TensorFlow dominates the field with the largest active community. It will be interesting to see what new advances TensorFlow and other utilities make possible for the future of our digital world. Continuing the recent trend of rapid updates, the TensorFlow team is making sure they address all the current and active challenges faced by the contributors and the developers while building machine learning and deep learning models. TensorFlow 2.0 will be a major update, we can expect the release candidate by next year early March. The preview version of this major milestone is expected to hit later this year. The major focus will be on ease of use, additional support for more platforms and languages, and eager execution will be the central feature of TensorFlow 2.0. This breakthrough version will add more functionalities and operations to handle current research areas such as reinforcement learning, GANs, building advanced neural network models more efficiently. Google will continue to invest and upgrade their existing TensorFlow ecosystem. According to Google’s CEO, Sundar Pichai “artificial intelligence is more important than electricity or fire.” TensorFlow is the solution they have come up with to bring artificial intelligence into reality and provide a stepping stone to revolutionize humankind. Read more The 5 biggest announcements from TensorFlow Developer Summit 2018 The Deep Learning Framework Showdown: TensorFlow vs CNTK Tensor Processing Unit (TPU) 3.0: Google’s answer to cloud-ready Artificial Intelligence

A history of cybercrime As computer systems have now become integral to the daily functioning of businesses, organizations, governments, and individuals we have learned to put a tremendous amount of trust in these systems. As a result, we have placed incredibly important and valuable information on them. History has shown, that things of value will always be a target for a criminal. Cybercrime is no different. As people flood their personal computers, phones, and so on with valuable data, they put a target on that information for the criminal to aim for, in order to gain some form of profit from the activity. In the past, in order for a criminal to gain access to an individual's valuables, they would have to conduct a robbery in some shape or form. In the case of data theft, the criminal would need to break into a building, sifting through files looking for the information of greatest value and profit. In our modern world, the criminal can attack their victims from a distance, and due to the nature of the internet, these acts would most likely never meet retribution. Cybercrime in the 70s and 80s In the 70s, we saw criminals taking advantage of the tone system used on phone networks. The attack was called phreaking, where the attacker reverse-engineered the tones used by the telephone companies to make long distance calls. In 1988, the first computer worm made its debut on the internet and caused a great deal of destruction to organizations. This first worm was called the Morris worm, after its creator Robert Morris. While this worm was not originally intended to be malicious it still caused a great deal of damage. The U.S. Government Accountability Office in 1980 estimated that the damage could have been as high as $10,000,000.00. 1989 brought us the first known ransomware attack, which targeted the healthcare industry. Ransomware is a type of malicious software that locks a user's data, until a small ransom is paid, which will result in the issuance of a cryptographic unlock key. In this attack, an evolutionary biologist named Joseph Popp distributed 20,000 floppy disks across 90 countries, and claimed the disk contained software that could be used to analyze an individual's risk factors for contracting the AIDS virus. The disk however contained a malware program that when executed, displayed a message requiring the user to pay for a software license. Ransomware attacks have evolved greatly over the years with the healthcare field still being a very large target. The birth of the web and a new dawn for cybercrime The 90s brought the web browser and email to the masses, which meant new tools for cybercriminals to exploit. This allowed the cybercriminal to greatly expand their reach. Up till this time, the cybercriminal needed to initiate a physical transaction, such as providing a floppy disk. Now cybercriminals could transmit virus code over the internet in these new, highly vulnerable web browsers. Cybercriminals took what they had learned previously and modified it to operate over the internet, with devastating results. Cybercriminals were also able to reach out and con people from a distance with phishing attacks. No longer was it necessary to engage with individuals directly. You could attempt to trick millions of users simultaneously. Even if only a small percentage of people took the bait you stood to make a lot of money as a cybercriminal. The 2000s brought us social media and saw the rise of identity theft. A bullseye was painted for cybercriminals with the creation of databases containing millions of users' personal identifiable information (PII), making identity theft the new financial piggy bank for criminal organizations around the world. This information coupled with a lack of cybersecurity awareness from the general public allowed cybercriminals to commit all types of financial fraud such as opening bank accounts and credit cards in the name of others. Cybercrime in a fast-paced technology landscape Today we see that cybercriminal activity has only gotten worse. As computer systems have gotten faster and more complex we see that the cybercriminal has become more sophisticated and harder to catch. Today we have botnets, which are a network of private computers that are infected with malicious software and allow the criminal element to control millions of infected computer systems across the globe. These botnets allow the criminal element to overload organizational networks and hide the origin of the criminals: We see constant ransomware attacks across all sectors of the economy People are constantly on the lookout for identity theft and financial fraud Continuous news reports regarding the latest point of sale attack against major retailers and hospitality organizations This is an extract from Information Security Handbook by Darren Death. Follow Darren on Twitter: @DarrenDeath. 

From data to web development, Python has come to stand as one of the most important and most popular open source programming languages being used today. But whilst some see it as almost a new kid on the block, Python is actually older than both Java, R, and JavaScript. So what are the origins of our favorite open source language? In the beginning... Python's origins lie way back in distant December 1989, making it the same age as Taylor Swift. Created by Guido van Rossum (the Python community's Benevolent Dictator for Life) as a hobby project to work on during week around Christmas, Python is famously named not after the constrictor snake but rather the British comedy troupe Monty Python's Flying Circus. (We're quite thankful for this at Packt - we have no idea what we'd put on the cover if we had to pick for 'Monty' programming books!) Python was born out of the ABC language, a terminated project of the Dutch CWI research institute that van Rossum worked for, and the Amoeba distributed operating system. When Amoeba needed a scripting language, van Rossum created Python. One of the principle strengths of this new language was how easy it was to extend, and its support for multiple platforms - a vital innovation in the days of the first personal computers. Capable of communicating with libraries and differing file formats, Python quickly took off. Computer Programming for Everybody Python grew throughout the early nineties, acquiring lambda, reduce(), filter() and map() functional programming tools (supposedly courtesy of a Lisp hacker who missed them and thus submitted working patches), key word arguments, and built in support for complex numbers. During this period, Python also served a central role in van Rossum's Computer Programming for Everybody initiative. The CP4E's goal was to make programming more accessible to the 'layman' and encourage a basic level of coding literacy as an equal essential knowledge alongside English literacy and math skills. Because of Python's focus on clean syntax and accessibility, it played a key part in this. Although CP4E is now inactive, learning Python remains easy and Python is one of the most common languages that new would-be programmers are pointed at to learn. Going Open with 2.0 As Python grew in the nineties, one of the key issues in uptake was its continued dependence on van Rossum. 'What if Guido was hit by a bus?' Python users lamented, 'or if he dropped dead of exhaustion or if he is rubbed out by a member of a rival language following?' In 2000, Python 2.0 was released by the BeOpen Python Labs team. The ethos of 2.0 was very much more open and community oriented in its development process, with much greater transparency. Python moved its repository to SourceForge, granting write access to its CVS tree more people and an easy way to report bugs and submit patches. As the release notes stated, 'the most important change in Python 2.0 may not be to the code at all, but to how Python is developed'. Python 2.7 is still used today - and will be supported until 2020. But the word from development is clear - there will be no 2.8. Instead, support remains focused upon 2.7's usurping younger brother - Python 3. The Rise of Python 3 In 2008, Python 3 was released on an almost-unthinkable premise - a complete overhaul of the language, with no backwards compatibility. The decision was controversial, and born in part of the desire to clean house on Python. There was a great emphasis on removing duplicative constructs and modules, to ensure that in Python 3 there was one - and only one - obvious way of doing things. Despite the introduction of tools such as '2to3' that could identify quickly what would need to be changed in Python 2 code to make it work in Python 3, many users stuck with their classic codebases. Even today, there is no assumption that Python programmers will be working with Python 3. Despite flame wars raging across the Python community, Python 3's future ascendancy was something of an inevitability. Python 2 remains a supported language (for now). But as much as it may still be the default choice of Python, Python 3 is the language's future. The Future Python's userbase is vast and growing - it's not going away any time soon. Utilized by the likes of Nokia, Google, and even NASA for it's easy syntax, it looks to have a bright future ahead of it supported by a huge community of OS developers. Its support of multiple programming paradigms, including object-oriented Python programming, functional Python programming, and parallel programming models makes it a highly adaptive choice - and its uptake keeps growing.

Living in the digital age brings its own challenges. News of security breaches in well-known companies is becoming a normal thing. In the battle between those who want to secure the Internet and those who want to exploit its security vulnerabilities, here's a list of five significant security challenges that I think information security is/will be facing in 2017. Army of young developers Everyone's beloved celebrity is encouraging the population to learn how to code, and it's working. Learning to code is becoming easier every day. There are loads of apps and programs to help people learn to code. But not many of them care to teach how to write secure code. Security is usually left as an afterthought, an "advanced" topic to learn sometime in future. Even without the recent fame, software development is a lucrative career. It has attracted a lot of 9-to-5ers who just care about getting through the day and collecting their paycheck. This army of young developers who care little about the craft is most to blame when it comes to vulnerabilities in applications. It would astonish you to learn how many people simply don't care about the security of their applications. The pressure to ship and ever-slipping deadlines don't make it any better. Rise of the robots I mean IoT devices. Sorry, I couldn't resist the temptation. IoT devices are everywhere. "Internet of Things" they call it. As if Internet wasn't insecure enough already, it's on "things" now. Most of these things rarely have any concept of security. Your refrigerator can read your tweets, and so can your 13-year-old neighbor. We've already seen a lot of famous disclosures of cars getting hacked. It's one of the examples of how dangerous it can get. Routers and other such infrastructure devices are becoming smarter and smarter. The more power they get, the more lucrative they become for a hacker to attack them. Your computer may have a firewall and anti-virus and other fancy security software, but your router might not. Most people don't even change the default password for such devices. It's much easier for an attacker to simply control your means of connecting to the Internet than connecting to your device directly. On the other front, these devices can be (and have been) used as bots to launch attacks (like DDoS) elsewhere. Internet advertisements as malware The Internet economy is hugely dependent on advertisements. Advertisements is a big big business, but it is becoming uglier and uglier every day. As if tracking users all over the webs and breaching their privacy was not enough, advertisements are now used for spreading malware. Ads are very attractive to attackers as they can be used to distribute content on fully legitimate sites without actually compromising them. They've already been in the news for this very reason lately. So the Internet can potentially be used to do great damage. Mobile devices Mobile apps go everywhere you go. That cute little tap game you installed yesterday might result in the demise of your business. But that's just the tip of the iceberg. Android will hopefully add essential features to limit permissions granted to installed apps. New exploits are emerging everyday for vulnerabilities in mobile operating systems and even in the processor chips. Your company might have a secure network with every box checked, but what about the laptop and mobile device that Cindy brought in? Organizations need to be ever more careful about the electronic devices their employees bring into the premises, or use to connect to the company network. The house of security cards crumbles fast if attackers get access to the network through a legitimate medium. The weakest links If you follow the show Mr. Robot (you should, it's brilliant), you might remember a scene from the first Season when they plan to attack the "impenetrable" Steel Mountain. Quoting Elliot: Nothing is actually impenetrable. A place like this says it is, and it’s close, but people still built this place, and if you can hack the right person, all of a sudden you have a piece of powerful malware. People always make the best exploits. People are the weakest links in many technically secure setups. They're easiest to hack. Social engineering is the most common (and probably easiest) way to get access to an otherwise secure system. With the rise in advanced social engineering techniques, it is becoming crucial everyday to teach the employees how to detect and prevent such attacks. Even if your developers are writing secure code, it's doesn’t matter if the customer care representative just gives the password away or grants access to an attacker. Here's a video of how someone can break into your phone account with a simple call to your phone company. Once your phone account is gone, all your two-factor authentications (that depend on SMS-based OTPs) are worth nothing. About the author Charanjit Singh is a freelance JavaScript (React/Express) developer. Being an avid fan of functional programming, he’s on his way to take on Haskell/Purescript as his main professional languages.

Let's say we have two groups of video games. In the first group, we have games like The Witcher 3, Civilization VI, and Overwatch. And in the second group, we have games like Super Meat Boy, Braid, and Stardew Valley. Can you tell the difference between these two groups? Is one group of games better than the other? No, they are all good games that have achieved both critical and financial success. Are the games in the first group sequels, while games in the second group are new? No, Overwatch is a new, original IP. Are the games in the first group more expensive than the second group? Now we're getting closer. The truth is, the first group of games comes from searching Google for "popular AAA games," while the second group comes from searching for "popular indie games." In short, the games in the first group are AAA games, and in the second group are indie games. Indie vs. AAA game development Now that we've seen the difference between the two groups, why do people separate these games into two different groups? What makes these two groups of games different from each other? Some would say that they are priced differently, but there are actually AAA games with low pricing as well as indie games with expensive pricing. How about the scale of the games? Again, there are indie games with big, massive worlds, and there are also AAA games set in short, small worlds. From my perspective, the key difference between the two groups of games is the size of the company developing the games. Indie games are usually made by companies with less than 30 people, and some are even made by less than five people. On the other hand, AAA games are made by much bigger companies, usually with hundreds of employees. Game development teams: size matters Earlier, I mentioned that company size is the key difference between indie games and AAA games. So it's not surprising that it's also the main difference between indie and AAA game development. In fact, the difference in team or company size leads to every difference between the two game development processes. Let's start with something personal, your role or position in the development team. Big teams usually have every position they need already filled. If they need someone to work on the game engine, they already have a engine programmer there. If they need someone to design a level, they already have a level designer working on it. In a big team, your role is already determined from the start, and you will rarely work on any task outside of your job description. If AAA game development values specialists, then indie game development values generalists who can fill multiple roles. It's not weird at all in a small development team if a programmer is asked to deal with both networking as well as enemy AI. Small teams usually aren't able to individually cover all the needed positions, so they turn to people who are able to work on a variety of tasks. Funding across the games industry Let's move to another difference, this time from the funding aspect. A large team requires a large amount of funding, simply because it has more people that need to be paid. And, if you look at the bigger picture, it also means that video games made by a large team have a large development cost. The opposite rings true as well; indie game development has much smaller development costs because they have smaller teams. Because every project has a chance of failure, the large development cost of AAA games becomes a big problem. If you're only spending a little money, maybe you're fine with a small chance of failure, but if you're spending a large sum of money, you definitely want to reduce that risk as much as possible. This ends up with AAA game development being much more risk-averse; they're trying to avoid risk as much as possible. In AAA game development, when there's a decision that needs to be made, the team will try to make sure that they don't make the wrong choice. They will do extensive market research and they will see what is trending in the market. They'd want to grab as many audience members as possible, so if there's any design that will exclude a significant amount of customers, it will be cut out. On the other hand, indie game development doesn't spend that much money. With a smaller development cost, indie games don't need to have a massive amount of sale to recoup their costs. Because of that, they're willing to take risks with experimental and unorthodox design, giving the team the creative freedom without needing to do market research. That said, indie game development harbors a different kind of risk. Unlike their bigger counterpart, indie game developers tend to live from one game to the next. That is, they use the revenue from their current game to fund the development of their next game. So if any of their games don't perform well, they could immediately close down. And that's another difference between the two game development process, AAA game development tends to be more financially stable compared to indie development. There are more differences between indie and AAA game development, but the ones listed above are definitely some of the most prominent. All in all, one development process isn't better than the other, and it falls back on you to decide which one is better suited for you. Raka Mahesa is a game developer at Chocoarts, who is interested in digital technology in general. Outside of work hours, he likes to work on his own projects, with Corridoom VR being his latest released game. Raka also regularly tweets as @legacy99.

Now that you’ve opened this article, I’ll assume you’re a web developer who is all excited with the prospect of building a machine learning project. You may be here for one of these reasons. Either you have been in a circle of people who find web development is dying? (Is it really dying or just unwell?). Or maybe you are stagnating in your current trajectory. And so, you want to learn something different, something trending, something like Artificial Intelligence. Or you/your employer/your client is aware of the capabilities of machine learning and want to include it in some part of your web app to make it more powerful. Or like the majority of the folks, you just want to see first hand if all the fuss about artificial intelligence is really worth all the effort to switch gears, by building a side toy ML project. Either way, there are different approaches to fulfill these needs. Learning Machine Learning for the Web with Javascript Learning machine learning coming from a web development background comes with its own constraints. You might worry about having to learn entirely different concepts from scratch - from different algorithms to programming languages like Python to mathematical concepts like linear algebra, calculus, and statistics. However, chances are you can skip learning a new language. You probably know some Javascript in some form or the other thanks to your web development experience. As such, you can learn Machine Learning in JavaScript (You don’t have to learn another programming language from scratch) and take it right to your browsers with WebGL. There are some advantages to using JavaScript for ML. Its popularity is one; while ML in JavaScript is not as popular as Python’s ML ecosystem, at the moment, the language itself is. As demand for ML applications rises, and as hardware becomes faster and cheaper, it's only natural for machine learning to become more prevalent in the JavaScript world. The JavaScript ecosystem offers a rich set of libraries suited for most Machine Learning tasks. Math: math.js Data Analysis: d3.js Server: node.js (express, koa, hapi) Performance: Tensorflow.js (e.g. GPU accelerated via WebGL API in the browser), Keras.js etc. Read also: 5 JavaScript machine learning libraries you need to know BRIIM is a good collection of materials to get you started as web developer or JavaScript enthusiast in machine learning. In case you’re interested in learning Python instead of Javascript, here are the set of libraries you should pick. Math: numpy Data Analysis: Pandas Data Mining: PySpark Server: Flask, Django Performance: TensorFlow (because it is written with a Python API over a C/C++ engine) or Keras (sits on top of TensorFlow). Using Machine Learning as a service If you don’t want to spend your time learning frameworks, tools, and languages suited for machine learning, you can adopt Machine Learning as a service or MLaaS. These services provide machine learning tools as part of cloud computing services. So basically, you can benefit from machine learning without the allied cost, time and risk of establishing an in-house internal machine learning team. All you need is sufficient knowledge of incorporating APIs. All Machine Learning tasks including data pre-processing, model training, model evaluation, and predictions can be completed through MLaaS. Read also: How machine learning as a service is transforming cloud A large number of companies provide Machine Learning as a service. Most prominent ones include: Amazon Machine Learning Amazon ML makes it easy for web developers to build smart applications using simple APIs. This includes applications for fraud detection, demand forecasting, targeted marketing, and click prediction. They provide a Developer Guide, which provides a conceptual overview of Amazon ML and includes detailed instructions for using the service. They also have a API reference, which describes all the API operations and provides sample requests and responses for supported web service protocols. Azure ML web app templates The web app templates available in the Azure Marketplace can build a custom web app that knows your web service's input data and expected results. All you need to do is give the web app access to your web service and data, and the template does the rest. There are two available templates: Azure ML Request-Response Service Web App Template Azure ML Batch Execution Service Web App Template Each template creates a sample ASP.NET application by using the API URI and key for your web service. The template then deploys the application as a website to Azure. No coding is required to use these templates. You just supply the API key and URI, and the template builds the application for you. Google Cloud based APIs Google also provides machine learning services, with pre-trained models and a service to generate your own tailored models. Google’s Cloud AutoML is a suite of machine learning products that enables developers with limited machine learning expertise to train high-quality models specific to their business needs. Cloud AutoML is used by Disney on their website shopDisney to enhance guest experience through more relevant search results, expedited discovery, and product recommendations. Building Conversational Interfaces As a web developer, another thing you might be looking into, is developing conversational interfaces or chatbots to enhance your web apps. Amazon, Google, and Microsoft provide Machine learning powered tools to help developers with building their own chatbots. Amazon Lex You can embed chatbots in your web apps with the Amazon Lex featuring ASR (Automatic Speech Recognition) and NLP (Natural Language Processing) capabilities. The API can recognize written and spoken text and the Lex interface allows you to hook the recognized inputs to various back-end solutions. Lex currently supports deploying chatbots for Facebook Messenger, Slack, and Twilio. Google Dialogflow Google’s Dialogflow can build voice and text-based conversational interfaces, such as voice apps and chatbots, powered by AI. Dialogflow incorporates Google's machine learning expertise and products such as Google Cloud Speech-to-Text. The API can be tweaked and customized for needed intents using Java, Node.js, and Python. It is also available as an enterprise edition. Microsoft Azure Cognitive Services Microsoft Cognitive Services simplify a variety of AI-based tasks, giving you a quick way to add intelligence technologies to your bots with just a few lines of code. It provides tools and APIs for aiding the development of conversational interfaces. These include: Translator Speech API Bing Speech API to convert text into speech and speech into text Speaker Recognition API for voice verification tasks Custom Speech Service to apply Azure NLP capacities using own data and models Language Understanding Intelligent Service (LUIS) is an API that analyzes intentions in text to be recognized as commands Text Analysis API for sentiment analysis and defining topics Bing Spell Check Translator Text API Web Language Model API that estimates probabilities of words combinations and supports word autocompletion Linguistic Analysis API used for sentence separation, tagging the parts of speech, and dividing texts into labeled phrases Read also: Top 4 chatbot development frameworks for developers These tools should be enough to get your feet off the ground quickly and move into the specific area of machine learning. Ultimately your choice of tool relies on the kind of application you want to build, your level of expertise, and how much time and effort you’re willing to put to learn. Obviously, depending on your area of choice, you would have to do more research and develop yourself in those areas. How should web developers learn machine learning? 5 examples of Artificial Intelligence in Web apps The most valuable skills for web developers to learn in 2018

Men in rags became millionaires and rich people bite the dust within minutes, thanks to crypto currencies. According to a research, over 1500 crypto currencies are being traded globally and with over 6 million wallets, proving that digital currency is here not just to stay but to rule. The rise and fall of crypto market isn’t hidden from anyone but the catch is—cryptocurrency still sells like a hot cake. According to Bill Gates, “The future of money is digital currency”. With thousands of digital currencies rolling globally, crypto traders are immensely occupied and this is where cryptocurrency trading bots come into play. They ease out the currency trade and research process that results in spending less effort and earning more money not to mention the hours saved. According to Eric Schmidt, ex CEO of Google, “Bitcoin is a remarkable cryptographic achievement and the ability to create something that is not duplicable in the digital world has enormous value.” The crucial part is - whether the crypto trading bot is dependable and efficient enough to deliver optimum results within crunch time. To make sure you don't miss an opportunity to chip in cash in your digital wallet, here are the top 15 crypto trading bots ranked according to the performance: 1- Gunbot Gunbot is a crypto trading bot that boasts of detailed settings and is fit for beginners as well as professionals. Along with making custom strategies, it comes with a“Reversal Trading” feature. It enables continuous trading and works with almost all the exchanges (Binance, Bittrex, GDAX, Poloniex, etc). Gunbot is backed by thousands of users that eventually created an engaging and helpful community. While Gunbot offers different packages with price tags of 0.02 to 0.15 BTC, you can always upgrade them. The bot comes with a lifetime license and is constantly upgraded. Haasbot Hassonline created this cryptocurrency trading bot in January 2014. Its algorithm is very popular among cryptocurrency geeks. It can trade over 500 altcoins and bitcoins on famous exchanges such as BTCC, Kraken, Bitfinex, Huobi, Poloniex, etc. You need to put a little input of the currency and the bot will do all the trading work for you. Haasbot is customizable and has various technical indicator tools. The cryptocurrency trading bot also recognizes candlestick patterns. This immensely popular trading bot is priced between 0.12BTC and 0.32 BTC for three months. 3- Gekko Gekko is a cryptocurrency trading bot that supports over 18 Bitcoin exchanges including Bitstamp, Poloniex, Bitfinex, etc. This bot is a backtesting platform and is free for use. It is a full fledged open source bot that is available on the GitHub. Using this bot is easy as it comes with basic trading strategies. The webinterface of Gekko was written from scratch and it can run backtests, visualize the test results while you monitor your local data with it. Gekko updates you on the go using plugins for Telegram, IRC, email and several different platforms. The trading bot works great with all operating systems such as Windows, Linux and macOS. You can even run it on your Raspberry PI and cloud platforms. 4- CryptoTrader CyrptoTrader is a  cloud-based platform which allows users to create automated algorithmic trading programs in minutes. It is one of the most attractive crypto trading bot and you wont need to install any unknown software with this bot. A highly appreciated feature of CryptoTrader is its Strategy Marketplace where users can trade strategies. It supports major currency exchanges such as Coinbase, Bitstamp, BTCe and is supported for live trading and backtesting. The company claims its cloud based trading bots are unique as compared with the currently available bots in the market. 5- BTC Robot One of the very initial automated crypto trading bot, BTC Robot offers multiple packages for different memberships and software. It provides users with a downloadable version of Windows. The minimum robot plan is of $149. BTC Robot sets up quite easily but it is noted that its algorithms aren't great at predicting the markets. The user mileage in BTC Robot varies heavily leaving many with mediocre profits. With the trading bot’s fluctuating evaluation, the profits may go up or down drastically depending on the accuracy of algorithm. On the bright side the bot comes with a sixty day refund policy that makes it a safe buy. 6- Zenbot Another open source trading bot for bitcoin trading, Zenbot can be downloaded and its code can be modified too. This trading bot hasn't got an update in the past months but still, it is among one of the few bots that can perform high frequency trading while backing up multiple assets at a time. Zenbot is a lightweight artificially intelligent crypto trading bot and supports popular exchanges such as Kraken, GDAX, Poloniex, Gemini, Bittrex, Quadriga, etc. Surprisingly, according to the GitHub’s page, Zenbot’s version 3.5.15 bagged an ROI of 195% in just a mere period of three months. 7- 3Commas 3Commas is a famous cryptocurrency trading bot that works well with various exchanges including Bitfinex, Binance, KuCoin, Bittrex, Bitstamp, GDAX, Huiboi, Poloniex and YOBIT. As it is a web based service, you can always monitor your trading dashboard on desktop, mobile and laptop computers. The bot works 24/7 and it allows you to take-profit targets and set stop-loss, along with a social trading aspect that enables you to copy the strategies used by successful traders. ETF-Like feature allows users to analyze, create and back-test a crypto portfolio and pick from the top performing portfolios created by other people. 8- Tradewave Tradewave is a platform that enables users to develop their own cryptocurrency trading bots along with automated trading on crypto exchanges. The bot trades in the cloud and uses Python to write the code directly in the browser. With Tradewave, you don't have to worry about the downtime. The bot doesn't force you to keep your computer on 24x7 nor it glitches if not connected to the internet. Trading strategies are often shared by community members that can be used by others too. However, it currently supports very few cryptocurrency exchanges such as Bitstamp and BTC-E but more exchanges will be added in coming months. 9- Leonardo Leonardo is a cryptocurrency trading bot that supports a number of exchanges such as Bittrex, Bitfinex, Poloniex, Bitstamp, OKCoin, Huobi, etc. The team behind Leonardo is extremely active and new upgrades including plugins are in the funnel. Previously, it cost 0.5 BTC but currently, it is available for $89 with a license of single exchange. Leonardo boasts of two trading strategy bots including Ping Pong Strategy and Margin Maker Strategy. The first strategy enables users to set the buy and sell price leaving all of the other plans to the bot while the Margin Maker strategy can buy and sell on price adjusted according to the direction in the market. This trading bot stands out in terms of GUI. 10- USI Tech USI Tech is a trading bot that is majorly used for forex trading but it also offers BTC packages. While majority of trading bots require an initial setup and installation, USI uses a different approach and it isn't controlled by the users. Users are needed to buy-in from their expert mining and bitcoin trade connections and then, the USI Tech bot guarantees a daily profit from the transactions and trade. To earn one percent of the capital daily, customers are advised to choose feature rich plans.. 11- Cryptohopper Cryptohopper  is a 24/7 cloud based trading bot that means it doesn't matter  if you are on the computer or not. Its system enables users to trade on technical indicators with subscription to a signaler who sends buy signals. According to the Cryptohopper’s website, it is the first crypto trading bot that is integrated with professional external signals. The bot helps in leveraging bull markets and has a latest dashboard area where users can monitor and configure everything. The dashboard also includes a configuration wizard for the major exchanges including Bittrex, GDAX, Kraken,etc. 12- My Bitcoin Bot MBB is a team effort from Brad Sheridon and his proficient teammates who are experts of cryptocurrency investment. My Bitcoin Bot is an automated trading software that can be accessed by anyone who is ready to pay for it. While the monthly plan is of $39 a month, the yearly subscription for this auto-trader bot is available for of $297. My bitcoin bot comes with heaps of advantages such as unlimited technical support, free software updates, access to trusted brokers list, etc. 13- Crypto Arbitrager A standalone application that operates on a dedicated server, Crypto Arbitrager can leverage robots even when the PC is off. The developers behind this cryptocurrency trading bot claim that this software uses code integration of financial time series. Users can make money from the difference in rates of Litecoins and Bitcoins. By implementing the advanced strategy of hedge funds, the trading bot effectively manages savings of users regardless of the state of the cryptocurrency market. 14- Crypto Robot 365 Crypto Robot 365 automatically trades your digital currency. It buys and sells popular cryptocurrencies such as Ripple, Bitcoin, ethereum, Litecoin, Monero, etc. Rather than a signup fee, this platform charges its commision on a per trade basis. The platform is FCA-Regulated and offers a realistic achievable win ratio. According to the trading needs, users can tweak the system. Moreover, it has an established trading history and  it even offers risk management options. Down The Line While cryptocurrency trading is not a piece of cake, trading with currency bots may be confusing for many. The aforementioned trading bots are used by many and each is backed by years of extensive hard work. With reliability, trustworthiness, smartwork and proactiveness being top reasons for choosing any cryptocurrency trading bot, picking up a trading bot is a hefty task. I recommend you experiment with small amount of money first and if your fate gets to a shining start, pick the trading bot that perfectly suits your way of making money via cryptocurrency. About the Author Rameez Ramzan is a Senior Digital Marketing Executive of Cubix - mobile app development company.  He specializes in link building, content marketing, and site audits to help sites perform better. He is a tech geek and loves to dwell on tech news. Crypto-ML, a machine learning powered cryptocurrency platform Beyond the Bitcoin: How cryptocurrency can make a difference in hurricane disaster relief Apple changes app store guidelines on cryptocurrency mining

Cloud computing has risen massively in terms of popularity in recent times. This is due to the way it reduces on-premise infrastructure cost and improves efficiency. Primarily, the cloud model has been divided into three major service categories: Infrastructure as a Service (IaaS) Platform as a Service (PaaS) Software as a Service (SaaS) We will discuss each of these instances in the following sections: The article is an excerpt taken from the book 'Cloud Analytics with Google Cloud Platform', written by Sanket Thodge. Infrastructure as a Service (IaaS) Infrastructure as a Service often provides the infrastructure such as servers, virtual machines, networks, operating system, storage, and much more on a pay-as-you-use basis. IaaS providers offer VM from small to extra-large machines. The IaaS gives you complete freedom while choosing the instance type as per your requirements: Common cloud vendors providing the IaaS services are: Google Cloud Platform Amazon Web Services IBM HP Public Cloud Platform as a Service (PaaS) The PaaS model is similar to IaaS, but it also provides the additional tools such as database management system, business intelligence services, and so on. The following figure illustrates the architecture of the PaaS model: Cloud platforms providing PaaS services are as follows: Windows Azure Google App Engine Cloud Foundry Amazon Web Services Software as a Service (SaaS) Software as a Service (SaaS) makes the users connect to the products through the internet (or sometimes also help them build in-house as a private cloud solution) on a subscription basis model. Below image shows the basic architecture of SaaS model. Some cloud vendors providing SaaS are: Google Application Salesforce Zoho Microsoft Office 365 Differences between SaaS, PaaS, and IaaS The major differences between these models can be summarized to a table as follows: Software as a Service (SaaS) Platform as a Service (PaaS) Infrastructure as a Service (IaaS) Software as a service is a model in which a third-party provider hosts multiple applications and lets customers use them over the internet. SaaS is a very useful pay-as-you-use model. Examples: Salesforce, NetSuite This is a model in which a third-party provider application development platform and services built on its own infrastructure. Again these tools are made available to customers over the internet. Examples: Google App Engine, AWS Lambda In IaaS, a third-party application provides servers, storage, compute resources, and so on. And then makes it available for customers for their utilization. Customers can use IaaS to build their own PaaS and SaaS service for their customers. Examples: Google Cloud Compute, Amazon S3 How PaaS, IaaS, and SaaS are separated at a service level In this section, we are going to learn about how we can separate IaaS, PaaS, and SaaS at the service level: As the previous diagram suggests, we have the first column as OPS, which stands for operations. That means the bare minimum requirement for any typical server. When we are going with a server to buy, we should consider the preceding features before buying. It includes Application, Data, Runtime, Framework, Operating System, Server, Disk, and Network Stack. When we move to the cloud and decide to go with IaaS—in this case, we are not bothered about the server, disk, and network stack. Thus, the headache of handling hardware part is no more with us. That's why it is called Infrastructure as a Service. Now if we think of PaaS, we should not be worried about runtime, framework, and operating system along with the components in IaaS. Things that we need to focus on are only application and data. And the last deployment model is SaaS—Software as a Service. In this model, we are not concerned about literally anything. The only thing that we need to work on is the code and just a look at the bill. It's that simple! If you found the above excerpt useful, make sure to check out the book 'Cloud Analytics with Google Cloud Platform' for more of such interesting insights into Google Cloud Platform. Read more Top 5 cloud security threats to look out for in 2018 Is cloud mining profitable? Why AWS is the prefered cloud platform for developers working with big data?

The orchestration war between Kubernetes and Docker Swarm appears to be over. Back in October, Docker announced that its Enterprise Edition could be integrated with Kubernetes. This move was widely seen as the Docker team conceding to Kubernetes dominance as an orchestration tool. But Docker Swarm nevertheless remains popular; it doesn't look like it's about to fall off the face of the earth. So what is the difference between Kubernetes and Docker Swarm? And why should you choose one over the other?  To start with it's worth saying that both container orchestration tools have a lot in common. Both let you run a cluster of containers, allowing you to increase the scale of your container deployments significantly without cloning yourself to mess about with the Docker CLI (although as you'll see, you could argue that one is more suited to scalability than the other). Ultimately, you'll need to view the various features and key differences between Docker Swarm and Kubernetes in terms of what you want to achieve. Do you want to get up and running quickly? Are you looking to deploy containers on a huge scale? Here's a brief but useful comparison of Kubernetes and Docker Swarm. It should help you decide which container orchestration tool you should be using. Docker Swarm is easier to use than Kubernetes One of the main reasons you’d choose Docker Swarm over Kubernetes is that it has a much more straightforward learning curve. As popular as it is, Kubernetes is regarded by many developers as complex. Many people complain that it is difficult to configure. Docker Swarm, meanwhile, is actually pretty simple. It’s much more accessible for less experienced programmers. And if you need a container orchestration solution now, simplicity is likely going to be an important factor in your decision making. ...But Docker Swarm isn't as customizable Although ease of use is definitely one thing Docker Swarm has over Kubernetes, it also means there's less you can actually do with it. Yes, it gets you up and running, but if you want to do something a little different, you can't. You can configure Kubernetes in a much more tailored way than Docker Swarm. That means that while the learning curve is steeper, the possibilities and opportunities open to you will be far greater. Kubernetes gives you auto-scaling - Docker Swarm doesn't When it comes to scalability it’s a close race. Both tools are able to run around 30,000 containers on 1,000 nodes, which is impressive. However, when it comes to auto-scaling, Kubernetes wins because Docker doesn’t offer that functionality out of the box. Monitoring container deployments is easier with Kubernetes This is where Kubernetes has the edge. It has in-built monitoring and logging solutions. With Docker Swarm you’ll have to use third-party applications. That isn’t necessarily a huge problem, but it does make life ever so slightly more difficult. Whether or not it makes life more difficult to outweigh the steeper Kubernetes learning curve however is another matter… Is Kubernetes or Docker Swarm better? Clearly, Kubernetes is a more advanced tool than Docker Swarm. That's one of the reasons why the Docker team backed down and opened up their enterprise tool for integration with Kubernetes. Kubernetes is simply the software that's defining container orchestration. And that's fine - Docker has cemented its position within the stack of technologies that support software automation and deployment. It's time to let someone else take on the challenge of orchestration But although Kubernetes is the more 'advanced' tool, that doesn't mean you should overlook Docker Swarm. If you want to begin deploying container clusters, without the need for specific configurations, then don't allow yourself to be seduced by something shinier, something ostensibly more popular. As with everything else in software development, understand and define what job needs to be done - then choose the right tool for the job.

The augmented reality market is developing rapidly. Today, it has a total market value of almost $15 billion; according to Statista,  and this figure could rise to $210 billion by 2022. Augmented reality is having a huge impact on the games industry, but it’s being used by organizations in fields as diverse as publishing and retail.. For example, Layar is an app that turns static objects into live objects, while IKEA’s Catalog app lets you imagine how different types of furniture might fit into your room. But it’s not just about commerce: some apps have a distinctly educational bent, like Field Trip. Field Trip uses augmented reality to help users learn about the history that immediately surrounds them. The best augmented reality apps are always deceptively simple. But to build a really effective augmented reality application you need a diverse range of skills, that span both the domains of software and real-world physics. Let’s take a closer look at location-based augmented reality apps, including what they’re used for and how you can begin building them. How does location-based AR app work? Location-based augmented reality apps are sometimes called geo-based AR apps. Whatever you call them, one thing is important: they collate GPS mobile data and the digital compass to detect the location and position of the device. The application works like this: The AR app arranges queries to be dispatched to the sensor. Once the data has been acquired, the app can determine where it should add virtual information (such as images) should be added to the real world. Location-based augmented reality apps can be used both inside or outside. When inside and it isn’t possible to connect to GPS, the application will use beacons for location data. The best examples of existing location-based augmented reality apps While reading about location-based augmented reality apps can give you a good idea of how they work, to be really inspired, you need to try some out for yourself. Here’s a list of some of the best location-based augmented reality apps out there. Yelp Monocle Yelp Monocle helps you navigate an unknown city. Using GPS, it provides exactly the sort of information you’d expect from Yelp, but in a format that’s fully integrated with your surroundings. So, you can see restaurant reviews, shop opening hours as you move around your environment. Ingress Ingress is an augmented reality gaming app that immerses you in a (semi) virtual world. Your main mission is to find portals that the game ‘creates’ in your immediate environment and open them. Essentially, the game is a great way to explore the world around you and places a new augmented layer on a place that might otherwise be familiar. Vortex Planetarium Vortex Planetarium is an app for aspiring astronomers or anybody else with a passing interest in astronomy. The app detects the user’s location and then provides them with celestial data to better understand the night sky. Steps to create location-based AR app So, if you like the idea of a location-based augmented reality app, you’ll probably want to get started. As we’ve seen, these apps can be incredibly complex, but if you break the development process down, it should become much easier. 1. Determine what resources you need Depending on the complexity of your app, you need to determine what resources are needed - that could be anything from data to other frameworks and services will be required. For example, if you plan to create a game with 3D objects, you’ll need to use Unity to build in that level of functionality and realism. 2. Choose the right augmented reality tool There are a huge number of available augmented reality software development kits out there. However, rather than wade through every single one, here are some of the best to get started with. R SDKs, but we will list the most popular ones that can give you the widest range of possible features. AR Kit by Apple AR kit from Apple features just about everything you’d need to develop an augmented reality application, For example, it has a technology that allows combines both computer vision and camera data to track the user’s environment. AR Kit also is able to adjust the light level in the virtual model, to respond to the level of light in the real world. ARKit 2 recently brought users a number of cool new features. For example, it allows you to build interactivity into your application, and also allows you to build ‘memory’ into your app so it can ‘remember’ the location of augmented reality objects.ARCore by Google In Google’s ARCore you’ll find a mapping tool which is particularly useful for developing of location-based AR apps. ARCore can also track motion and detect vertical and horizontal surfaces. In the latest version of ARCore users can take two gadgets and work with one AR object from different viewing angles. 3. Geolocation data should be added Not all SDKs provide mapping feature. If it doesn’t, it’s essential to make sure you add in geolocation data. Without it, the app wouldn’t work! As we’ve already seen, GPS technology is typically used. It’s convenient and it can detect a user's location anywhere. It can, however, consume a lot of energy. Location services on iOS and Android will help to activate geolocation on the device. 3 augmented reality pitfalls to avoid Developing something as complex as a location-based augmented reality app is bound to lead to some challenges. So be prepared - watch for some of these pitfalls.. Ensure you have proper functionality. When users move with their camera and look for AR objects, these objects should remain static, regardless of the user’s movements. To do this, use SLAM - Simultaneous Localization and Mapping. This is a technique that allows software systems - like robots - ‘understand’ where they are situated in relation to their surroundings. Accuracy. A crucial factor for any AR app is accuracy. When developing your app, it’s essential to consider the user’s position to ensure that the app sends queries to sensors correctly. If it doesn’t the whole experience could seem plain weird for the user. Similarly, the distance between the device and the real world must be calculated correctly - again, if it isn’t your application simply will not work. Get started - build an awesome augmented reality app! Clearly, building a location-based augmented reality app isn’t easy. It requires skill and a commitment to keep going in the face of challenges. You certainly need a team of great developers around you if you’re going to deliver something that makes an impact. But, really, that’s what makes software development exciting, right? Author Bio Vitaly Kuprenko is a technical writer at Cleveroad. It's a web and mobile app development company in Ukraine. He enjoys telling about tech innovations and digital ways to boost businesses. Magic Leap unveils Mica, a human-like AI in augmented reality. Magic Leap teams with Andy Serkis’ Imaginarium Studios to enhance Augmented Reality “As Artists we should be constantly evolving our technical skills and thought processes to push the boundaries on what’s achievable,” Marco Matic Ryan, Augmented Reality Artist

The primary target for businesses while working on mobile apps is the Android platform, thanks to the massive market share the mobile operating system holds. It’s popularity can be attributed to the fact that it is open source and is regular updated with new enhancements and features. Android devices generally tend to differ based on the mobile hardware features even when powered by the same version of the Android OS. This is why it is essential that when developing apps for Android, developers create mobile apps capable of targeting a diverse range of mobile devices running on different versions of Android OS. During the various stages of planning, developing and testing, developers need to focus comprehensively on the apps functionality, accessibility, usability, performance, and security so that users can be engaged despite their choice of device. Also, they also need to look for ways to make the apps deliver a more personalized user experience across the various devices an operating system. Furthermore, developers need to understand and find solutions to the common challenges involved in android app development. Common Challenges Android App Developers Face 1. Hardware Features The Android OS is unlike any other mobile operating system. For one thing, it is an open source system. Alphabet gives manufacturers the leeway to customize the operating system to their specific needs. Also, there are no regulations on the devices being released by the different manufacturers. As a result, you can find various Android devices with different hardware features running on the same Android version. Two smartphones running on Android latest ver, for example, may have different screen resolutions, camera, screen size, and other hardware structures. During android app development, developers need to account for all of this to ensure the application delivers a personalized experience to each user. 2. Lack of Uniform User Interface Design Rules Since Google is yet to release any standard UI (user interface) design rules or process for mobile app developers, most developers don’t follow any standard UI development rules or procedure. Because developers are creating custom UI interfaces in their preferred way, a lot of apps tend to function or look different across different devices. This diversity and incompatibility of the UI usually affects the user experience that the Android app directly delivers. Smart developers prefer to go for a responsive layout that’ll keep the UI consistent across different devices. Moreover, developers need to test the UI of the app extensively by combining emulators and real mobile devices. Designing a UI that makes the app deliver the same user experience across varying Android devices is one of the more daunting challenges developers face. 3. API Incompatibility A lot of developers make use of third-party APIs to enhance the functionality and interoperability of a mobile device. Unfortunately, not all third-party APIs available for Android app development are of high quality.. Some APIs were created for a particular Android version and will not work on devices running on a different version of the operating system. Developers usually have to come up with ways to make a single API work on all Android versions, a task they often find to be very challenging. 4. Security Flaws As previously mentioned, Android is an open source software, and because of that, manufacturers find it easy to customize Android to their desired specifications. However, this openness and the massive market size makes Android a frequent target for security attacks. There have been several instances where the security of millions of Android mobile devices have been affected by security flaws and bugs like mRST, Stagefright, FakeID, ‘Certifi-gate,’ TowelRoot and Installer Hijacking. Developers need to include robust security features in their applications and utilize the latest encryption mechanisms to keep user information secure and out of the hands of hackers. 5. Search Engine Visibility The latest data from Statista shows that Google Play Store contains a higher number of mobile apps. Additionally, a large number of Android users prefer free apps than paid apps which is why developers need to promote their mobile applications to increase their download numbers and employ application monetization options. The best way to promote the app to reach their target audience is to use comprehensive digital marketing strategies. Most developers make use of digital marketing professionals to promote their apps aggressively. 6. Patent Issues Google doesn’t implement any guidelines for the evaluation of the quality of new apps that are getting submitted to the Play Store. This lack of a quality assessment guideline causes a lot of patent-related issues for developers. Some developers, to avoid patent issues, have to modify and redesign their apps in the future. As per my personal experience, I have tried to cover general challenges faced by Android app developers. I’m sure keeping wary of these challenges would help developers to build successful apps in the most hassle free way. Author Bio Harnil Oza is the CEO of Hyperlink InfoSystem, one of the leading app development companies in New York, USA and India who deliver mobile solutions mainly on Android and iOS platform. He regularly contributes his knowledge on leading blogging sites. LEGO launches BrickHeadz Builder AR, a new and free Android app to bring bricks and toys to life How Android app developers can convert iPhone apps How to Secure and Deploy an Android App

Generative adversarial networks, or GANs, are a powerful type of neural network used for unsupervised machine learning. Made up of two competing models which run in competition with one another, GANs are able to capture and copy variations within a dataset. They’re great for image manipulation and generation, but they can also be deployed for tasks like understanding risk and recovery in healthcare and pharmacology. GANs are actually pretty new - they were first introduced by Ian Goodfellow in 2014. Goodfellow developed them to tackle some of the issues with similar neural networks, including the Boltzmann machine and autoencoders. Both the Boltzmann machine and autoencoders use the Markov Decision Chain which has a pretty high computational cost. This efficiency gives engineers significant gains - which you need if you’re working at the cutting edge of artificial intelligence. How do Generative Adversarial Networks work? Let's start with a simple analogy. You have a painting - say the Mona Lisa - and we have a master forger who wants to create a duplicate painting. The forger does this by learning how the original painter - Leonardo Da Vinci - produced the painting. Meanwhile, you have an investigator trying to capture the forger and ‘second guess’ the rules the forger is learning. To map this onto the architecture of a GAN, the forger is the generator network, which learns the distribution of classes while the investigator is the discriminator network, which learning the boundaries between those classes - the formal ‘shape’ of the dataset. Applications of GANs Generative adversarial networks are used for a number of different applications. One of the best examples is a Google Brain project back in 2016 - researchers used GANs to develop a method of encryption. This project used 3 neural networks - Alice, Bob, and Eve. Alice’s job was to send an encrypted message to Bob. Bob’s job was to decode that message, while Eve’s job was to intercept it. To begin with Alice’s messages were easily intercepted by Eve. However, thanks to Eve’s adversarial work, Alice began to develop its own encryption strategy - it took 15,000 runs for Alice to successfully encrypt a message that could be deciphered by Bob that Eve couldn’t intercept. Elsewhere, GANs are also being used in fields such as drug research. The neural networks can be trained on the existing drugs and suggest new synthetic chemical structures that improve on drugs that already exist. Generative adversarial networks: the cutting edge of artificial intelligence As we’ve seen, GANs offer some really exciting opportunities in artificial intelligence. There are two key advantages you need to remember: GANs solve the problem of generating data when you don’t have enough to begin with and they require no human supervision. This is crucial when you think about the cutting edge of artificial intelligence, both in terms of the efficiency of running the models, and the real-world data we want to use - which could be poor quality or have privacy and confidentiality issues, as much healthcare data does.

[box type="note" align="" class="" width=""]This is a book excerpt taken from Advanced Analytics with R and Tableau authored by Jen Stirrup & Ruben Oliva Ramos. This book will help you make quick, cogent, and data driven decisions for your business using advanced analytical techniques on Tableau and R.[/box] Today we explore popular data analytics methods such as Microsoft TDSP Process and the CRISP- DM methodology. Introduction There is an increasing amount of data in the world, and in our databases. The data deluge is not going to go away anytime soon! Businesses risk wasting the useful business value of information contained in databases, unless they are able to excise useful knowledge from the data. It can be hard to know how to get started. Fortunately, there are a number of frameworks in data science that help us to work our way through an analytics project. Processes such as Microsoft Team Data Science Process (TDSP) and CRISP-DM position analytics as a repeatable process that is part of a bigger vision. Why are they important? The Microsoft TDSP Process and the CRISP-DM frameworks are frameworks for analytics projects that lead to standardized delivery for organizations, both large and small. In this chapter, we will look at these frameworks in more detail, and see how they can inform our own analytics projects and drive collaboration between teams. How can we have the analysis shaped so that it follows a pattern so that data cleansing is included? Industry standard methodologies for analytics There are a few main methodologies: the Microsoft TDSP Process and the CRISP-DM Methodology. Ultimately, they are all setting out to achieve the same objectives as an analytics framework. There are differences, of course, and these are highlighted here. CRISP-DM and TDSP focus on the business value and the results derived from analytics projects. Both of these methodologies are described in the following sections. CRISP-DM One common methodology is the CRISP-DM methodology (the modeling agency). The Cross Industry Standard Process for Data Mining or (CRISP-DM) model as it is known, is a process model that provides a fluid framework for devising, creating, building, testing, and deploying machine learning solutions. The process is loosely divided into six main phases. The phases can be seen in the following diagram: Initially, the process starts with a business idea and a general consideration of the data. Each stage is briefly discussed in the following sections. Business understanding/data understanding The first phase looks at the machine learning solution from a business standpoint, rather than a technical standpoint. The business idea is defined, and a draft project plan is generated. Once the business idea is defined, the data understanding phase focuses on data collection and familiarity. At this point, missing data may be identified, or initial insights may be revealed. This process feeds back to the business understanding phase. CRISP-DM model — data preparation In this stage, data will be cleansed and transformed, and it will be shaped ready for the modeling phase. CRISP-DM — modeling phase In the modeling phase, various techniques are applied to the data. The models are further tweaked and refined, and this may involve going back to the data preparation phase in order to correct any unexpected issues. CRISP-DM — evaluation The models need to be tested and verified to ensure that they meet the business objectives that were defined initially in the business understanding phase. Otherwise, we may have built models that do not answer the business question. CRISP-DM — deployment The models are published so that the customer can make use of them. This is not the end of the story, however. CRISP-DM — process restarted We live in a world of ever-changing data, business requirements, customer needs, and environments, and the process will be repeated. CRISP-DM summary CRISP-DM is the most commonly used framework for implementing machine learning projects specifically, and it applies to analytics projects as well. It has a good focus on the business understanding piece. However, one major drawback is that the model no longer seems to be actively maintained. The official site, CRISP-DM.org, is no longer being maintained. Furthermore, the framework itself has not been updated on issues on working with new technologies, such as big data. Big data technologies means that there can be additional effort spend in the data understanding phase, for example, as the business grapples with the additional complexities that are involved in the shape of big data sources. Team Data Science Process The TDSP process model provides a dynamic framework to machine learning solutions that have been through a robust process of planning, producing, constructing, testing, and deploying models. Here is an example of the TDSP process: The process is loosely divided into four main phases: Business Understanding Data Acquisition and Understanding Modeling Deployment The phases are described in the following paragraphs. Business understanding The Business understanding process starts with a business idea, which is solved with a machine learning solution. The business idea is defined from the business perspective, and possible scenarios are identified and evaluated. Ultimately, a project plan is generated for delivering the solution. Data acquisition and understanding Following on from the business understanding phase is the data acquisition and understanding phase, which concentrates on familiarity and fact-finding about the data. The process itself is not completely linear; the output of the data acquisition and understanding phase can feed back to the business understanding phase, for example. At this point, some of the essential technical pieces start to appear, such as connecting to data, and the integration of multiple data sources. From the user's perspective, there may be actions arising from this effort. For example, it may be noted that there is missing data from the dataset, which requires further investigation before the project proceeds further. Modeling In the modeling phase of the TDSP process, the R model is created, built, and verified against the original business question. In light of the business question, the model needs to make sense. It should also add business value, for example, by performing better than the existing solution that was in place prior to the new R model. This stage also involves examining key metrics in evaluating our R models, which need to be tested to ensure that the models meet the original business objectives set out in the initial business understanding phase. Deployment R models are published to production, once they are proven to be a fit solution to the original business question. This phase involves the creation of a strategy for ongoing review of the R model's performance as well as a monitoring and maintenance plan. It is recommended to carry out a recurrent evaluation of the deployed models. The models will live in a fluid, dynamic world of data and, over time, this environment will impact their efficacy. The TDSP process is a cycle rather than a linear process, and it does not finish, even if the model is deployed. It is comprised of a clear structure for you to follow throughout the Data Science process, and it facilitates teamwork and collaboration along the way. TDSP Summary The data science unicorn does not exist; that is, the person who is equally skilled in all areas of data science, right across the board. In order to ensure successful projects where each team player contributes according to their skill set, the Team Data Science Summary is a team-oriented solution that emphasizes teamwork and collaboration throughout. It recognizes the importance of working as part of a team to deliver Data Science projects. It also offers useful information on the importance of having standardized source control and backups, which can include open source technology. If you liked our post, please be sure to check out Advanced Analytics with R and Tableau that shows how to apply various data analytics techniques in R and Tableau across the different stages of a data science project highlighted in this article.