Tech Guides

article-image-computer-vision-is-an-expanding-market-heres-why

12 Jun 2018

6 min read

Computer vision is growing quickly. Here's why.

12 Jun 2018

Computer Vision is one of those technologies that has grown in leaps and bounds over the past few years. If you look back 10 years, it wasn’t the case, as CV was more a topic of academic interest. Now, however, computer vision is clearly a driver and benefactor of the renowned Artificial Intelligence. Through this article, we’ll understand the factors that have sparked the rise of Computer Vision. A billion $ market You heard it right! Computer Vision is a billion dollar market, thanks to the likes of Intel, Amazon, Netflix, etc investing heavily in the technology’s development. And from the way events are unfolding, the market is expected to hit a record $ 17 billion, by 2023. That’s at a cumulative growth rate of over 7% per year, from 2018 to 2023. Now this is a joint figure for both the hardware and software components related to Computer Vision. Under the spotlight Let’s talk a bit about a few companies that are already taking advantage of Computer Vision, and are benefiting from it. Intel There are several large organisations that are investing heavily in Computer Vision. Last year, we saw Intel invest $15 Billion towards acquiring Mobileye, an Israeli auto startup. Intel published its findings stating that the autonomous vehicle market itself would rise to $ 7 Trillion by 2050. The autonomous vehicle industry will be one of the largest implementers of computer vision technology. These vehicles will use Computer Vision to “see” their surroundings and communicate with other vehicles. Netflix Netflix on the other hand, is using Computer Vision for more creative purposes. With the rise of Netflix’s original content, the company is investing in Computer Vision to harvest static image frames directly from the source videos to provide a flexible source of raw artwork, which is used for digital merchandising. For example, within a single episode of Stranger Things, there are nearly 86k static video frames, that would had to have been analysed by human teams to identify the most appropriate stills to be featured. This meant first going through each of those 86k images, then understanding what worked for viewers of the previous episode and then applying the learning in the selection of future images. Need I estimate how long that would have taken to do? Now, Computer Vision performs this task seamlessly, with a much higher accuracy than that of humans. Pinterest Pinterest, the popular social networking application, sees millions of images, GIFs and other visuals shared every day. In 2017, they released an application feature callen Lens, that allows users to use their phone’s camera to search for similar looking decor, food and clothing, in the real world. Users can simply point their cameras at an image and Pinterest will show them similar styles and ideas. Recent reports reveal that Pinterest’s revenue has grown by a staggering 58%! National Surveillance in CCTV The world’s biggest AI startup, SenseTime, provides China with the world’s largest and most sophisticated CCTV network. With over 170 Mn CCTV cameras, the government authorities and police departments are able to seamlessly identify people. They perform this by wearing smart glasses, that have facial recognition capabilities. Bring this technology to Dubai and you’ve got a supercop in a supercar! The nation-wide surveillance project that’s named Skynet, began as early as 2005, although recent advances in AI have given it a boost. Reading through discussions like these is real fun. People used to quip that such “fancy” machines are only for the screen. If only they knew that such a machine would be a reality just a few years from then. Clearly, computer vision is one of the most highly valued commercial applications of machine learning and when integrated with AI, it’s an offer only a few can resist! Star Acquisitions that matter Several acquisitions have taken place in the field of Computer Vision in the past two years alone. The most notable of them being Intel’s acquisition of Movidius, to the tune of $400 Mn. Here are some of the others that have happened since 2016: Twitter acquires Magic Pony Technology for $150Mn Snap Inc acquires Obvious Engineering for $47 Mn Salesforce acquires Metamind for $32.8 Mn Google acquires Eyefluence for $21.6 Mn This shows the potential of the computer vision market and how big players are in the race to dive deep into the technology. Three little things driving computer vision I would say there are 3 clear growth factors that are contributing to the rise of Computer Vision: Deep Learning Advancements in Hardware Growth of the Datasets Deep Learning The advancements in the field of Deep Learning are bound to boost Computer Vision. Deep Learning algorithms are capable of processing tonnes of images, much more accurately than humans. Take Feature Extraction for example. The primary pain point with feature extraction is that you have to choose which features to look for in a given image. This becomes cumbersome and almost impossible when the number of classes you are trying to define, starts to grow. There are so many features, that you have to deal with a plethora of parameters, that have to be fine-tuned. Deep Learning simplifies this process for you. Advancements in Hardware With new hardware like GPUs capable of processing petabytes of data, algorithms are capable of running faster and more efficiently. This has led to the advancement in real-time processing and vision capabilities. Pioneering hardware manufacturers like NVIDIA and Intel are in a race to create more powerful and capable hardware to support deep learning capabilities for Computer Vision. Growth of the Datasets Training Deep Learning algorithms isn’t a daunting task anymore. There are plenty of open source data sets that you can choose from to train your algorithms. The more the data, the better is the training and accuracy. Here are some of the most notable data sets for computer vision. ImageNet with 15 million images, is a massive dataset Open Images has 9 million images Microsoft Common Objects in Context (COCO) has around 330K images CALTECH-101 has approximately 9,000 images Where tha money at? The job market for Computer Vision is on a rise too, with Computer Vision featuring at #3 on the list of top jobs in 2018, according to Indeed. Organisations are looking for Computer Vision Engineers who are well versed with writing efficient algorithms for handling large amounts of data. Source: Indeed.com So is it the right time to invest or perhaps learn Computer Vision? You bet it is! It’s clear that Computer Vision is a rapidly growing market and will have a sustained growth for the next few years. If you’re just planning to start out or even if you’re competent in using tools for Computer Vision, here are some resources to help you skill up with popular CV tools and techniques. Introducing Intel’s OpenVINO computer vision toolkit for edge computing Top 10 Tools for Computer Vision Computer Vision with Keras, Part 1

0
0
8201

article-image-exploring-language-improvements-c-72-and-73-0

Mark J.

28 Nov 2017

9 min read

Exploring Language Improvements in C# 7.2 and 7.3

Mark J.

28 Nov 2017

9 min read

0
0
8179

article-image-capsnet-capsule-networks-convolutional-neural-networks-cnns

Savia Lobo

13 Dec 2017

5 min read

CapsNet: Are Capsule networks the antidote for CNNs kryptonite?

Savia Lobo

13 Dec 2017

5 min read

Convolutional Neural networks (CNNs), are a group from the neural network family that has manifested in areas such as Image recognition, classification, etc. They are one of the popular neural network models present in nearly all of the image recognition tasks that provide state-of-the-art-results. However, these CNNs have drawbacks, which are to be discussed later in the article. In order to address the issue with CNNs, Geoffrey Hinton, popularly known as the Godfather of Deep Learning, recently proposed a research paper along with two other researchers, Sara Sabour and Nicholas Frosst. In this paper, they introduced CapsNet or Capsule Network--a neural network, based on multi-layer capsule system. Let’s explore the issue with CNNs and how CapsNet came as an advancement to it. What is the issue with CNNs? Convolutional Neural Network or CNNs are known to seamlessly handle image classification tasks. They are experts in learning at a granular level; where the lower layers detect edges and shape of an object, and the higher layers detect the image as a whole. However, CNNs perform poorly when an image possesses a slightly different orientation (rotation or a tilt), as it compares every image with the ones it learns during training. For instance, if an image of a face is to be detected, it checks for facial features such as nose, two eyes, mouth, eyebrows, etc; irrespective of the placement. This means CNNs may identify an incorrect face in cases where the placement of an eye and the nose is not as conventionally expected, for example in case of the profile view. So, the orientation and the spatial relationships between the objects within an image is not considered by a CNN. To make CNNs understand orientation and spatial relationships, they were trained profusely with images taken from all possible angles. Unfortunately, it resulted in excess amount of time required to train the model. Also, the performance of the CNNs did not improve largely. Pooling methods were also introduced at each layer within the CNN model for two reasons; first to reduce the time invested in training, and second to bring out positional invariance within CNNs. It resulted in triggering false positives in an image, i.e., it detected the object within an image but did not check its orientation. Also it incorrectly declared it as a right image. Thus, positional invariance made the CNNs susceptible to minute changes in viewpoint. Instead of invariance, what CNNs require is equivariance-- a feature that makes CNNs adapt to change in rotation or proportion within an image. This equivariance feature is now possible via Capsule Network! The Solution: Capsule Network CapsNet or Capsule network is an encapsulation of nested neural network layers. Traditional neural network contains multiple layers whereas a capsule network contains multiple layers within a single capsule. CNNs go deeper in terms of height, whereas the capsule network deepens in terms of nesting or internal structure. Such a model is highly robust to geometric distortions and transformations, which are a result of non-ideal camera angles. Thus, it is able to exceptionally handle orientations, rotations and so on. CapsNet Architecture Source: https://arxiv.org/pdf/1710.09829.pdf Key Features: Layer based Squashing In a typical Convolutional Neural Network, the squashing function is added to each layer of the CNN model. A squashing function compresses the input to one of the ends of a small interval, introducing nonlinearity to the neural network and enables the network to be effective. Whereas, in a Capsule network, the squashing function is applied to the vector output of each capsule. Given below is a squashing function proposed by Hinton in his research paper. Squashing function Source: https://arxiv.org/pdf/1710.09829.pd Instead of applying non-linearity to each neuron, the squashing function applies squashing to a group of neurons i.e the capsule. To be more precise, it applies nonlinearity to the vector output of each capsule. The squashing function also tries to squash the vector output to zero if it is a small vector. If the vector is too long, the function tries to limit the output vector to 1. Dynamic Routing Dynamic routing algorithm in CapsNet replaces the scalar-output feature detectors of the CNN with the vector-output capsules. Also, the max pooling feature in CNNs, which led to positional invariance, is replaced with ‘routing by agreement’. The algorithm ensures that when they forward propagate the data, it goes to the next most relevant capsule in the layer above. Although dynamic routing adds an extra computational cost to the capsule network, it has been proved to be advantageous to the network by making it more scalable and adaptable. Training the Capsule Network The capsule network is trained using the MNIST. MNIST is a dataset which includes more than 60,000 handwritten digit images. It is used to test machine learning algorithms. The capsule model is trained for 50 epochs with a batch size of 128 parts, where each epoch is responsible for a complete run through the training dataset. A TensorFlow implementation of the CapsNet based on Hinton’s research paper is available in GitHub repository. Similarly, CapsNet can also be implemented using other deep learning frameworks such as Keras, PyTorch, MXNet, etc. CapsNet is a recent breakthrough in the field of Deep learning and have a promise to benefit organizations with accurate image recognition tasks. Also, implementations with CapsNet is slowly catching up and is expected to reach at par like CNNs. They have been trained on a very simplistic dataset i.e the MNIST. They will still require to prove themselves on various other datasets. However, as time advances and we see CapsNet being trained within different domains, it will be exciting to discern how it moulds itself as a faster and more efficient training technique for deep learning models.

0
0
8158

article-image-one-shot-learning-solution-low-data-problem

Savia Lobo

04 Dec 2017

5 min read

One Shot Learning: Solution to your low data problem

Savia Lobo

04 Dec 2017

5 min read

The fact that machines are successful in replicating human intelligence is mind-boggling. However, this is only possible if machines are fed with correct mix of algorithms, huge collection of data, and most importantly the training given to it, which in turn leads to faster prediction or recognition of objects within the images. On the other hand, when you train humans to recognize a car for example, you simply have to show them a live car or an image. The next time they see any vehicle, it would be easy for them to distinguish a car among-st other vehicles. In a similarly way, can machines learn with single training example like humans do? Computers or machines lack a key part that distinguishes them from humans, and that is, ‘Memory’. Machines cannot remember; hence it requires millions of data to be fed in order to understand the object detection, be it from any angle. In order to reduce this supplement of training data and enabling machines to learn with less data at hand, One shot learning is brought to its assistance. What is one shot learning and how is it different from other learning? Deep Neural network models outperform various tasks such as image recognition, speech recognition and so on. However, such tasks are possible only due to extensive, incremental training on large data sets. In cases when there is a smaller dataset or fewer training examples, a traditional model is trained on the data that is available. During this process, it relearns new parameters and incorporates new information, and completely forgets the one previously learned. This leads to poor training or catastrophic inference. One shot learning proves to be a solution here, as it is capable of learning with one, or a minimal number of training samples, without forgetting. The reason for this is, they posses meta-learning; a capability often seen in neural network that has memory. How One shot learning works? One shot learning strengthens the ability of the deep learning models without the need of a huge dataset to train on. Implementation of One shot learning can be seen in a Memory Augmented Neural Network (MANN) model. A MANN has two parts, a controller and an external memory model. The controller is either a feed forward neural network or an LSTM (Long Short Term Memory) network, which interacts with the external memory module using number of read/write heads. These heads fetch or place representations to and fro the memory. LSTMs are proficient in long term storage through slow updates of weights and short term storage via the external memory module. They are trained to meta-learn; i.e. it can rapidly learn unseen functions with fewer data samples.Thus, MANNs are said to be capable of metalearning. The MANN model is later trained on datasets that include different classes with very few samples. For instance, the Omniglot dataset, a collection of handwritten samples of different languages, with very few samples of each language. After continuously training the model with thousands of iterations by using few samples, the model was able to recognize never-seen-before image samples, taken from a disjoint sample of the Omniglot dataset. This proves that MANN models are able to outperform various object categorization tasks with minimal data samples. Similarly, One shot learning can also be achieved using Neural Turing Machine and Active One shot learning. Therefore, learning with a single attempt/one shot actually involves meta-learning. This means, the model gradually learns useful representations from the raw data using certain algorithms, for instance, the gradient descent algorithm. Using these learnings as a base knowledge, the model can rapidly cohere never seen before information with a single or one-shot appearance via an external memory module. Use cases of One shot learning Image Recognition: Image representations are learnt using supervised metric based approach. For instance, siamese neural network, an identical sister network, discriminates between the class-identity of an image pair. Features of this network are reused for one-shot learning without the need for retraining. Object Recognition within images: One shot learning allows neural network models to recognize known objects and its category within an image. For this, the model learns to recognize the object with a few set of training samples. Later it compares the probability of the object to be present within the image provided. Such a model trained on one shot can recognize objects in an image despite the clutter, viewpoint, and lighting changes. Predicting accurate drugs: The availability of datasets for a drug discovery are either limited or expensive. The molecule found during a biological study often does not end up being a drug due to ethical reasons such as toxicity, low-solubility and so on. Hence, a less amount of data is available about the candidate molecule. Using one shot learning, an iterative LSTM combined with Graph convolutional neural network is used to optimize the candidate molecule. This is done by finding similar molecules with increased pharmaceutical activity and lesser risks to patients. A detailed explanation of how using low data, accurate drugs can be predicted is discussed in a research paper published by the American Chemical Society(ACS). One shot learning is in its infancy and therefore use cases can be seen in familiar applications such as image and object recognition. As the technique will advance with time and the rate of adoption, other applications of one shot learning will come into picture. Conclusion One shot learning is being applied in instances of machine learning or deep learning models that have less data available for their training. A plus point in future is, that organizations will not have to collect huge amount of data for their ML models to be trained, only a few training samples would do the job! Large number of organizations are looking forward to adopt one shot learning within their deep learning models. It would be exciting to see how one shot learning will glide through being the base of every neural network implementation.

0
0
8152

article-image-how-nodejs-changing-web-development

Antonio Cucciniello

05 Jul 2017

5 min read

How is Node.js Changing Web Development?

Antonio Cucciniello

05 Jul 2017

5 min read

If you have remotely been paying attention to what is going on in the web development space, you know that Node.js has become extremely popular and is many developers’ choice of backend technology. It all started in 2009 by Ryan Dahl. It is a JavaScript runtime that is built on Google Chrome's V8 JavaScript Engine.Over the past couple of years, more and more engineers have moved towards Node.js in many of the their web applications. With plenty of people using it now, how has Node.js changed web development? Scalability Scalability is the one thing that makes Node.js so popular. Node.js runs everything in a single thread. This single thread is event driven (due to JavaScript being the language that it is written with). It is also non-blocking. Now, when you spin up a server in your Node web app, every time a new user connects to the server, that launches an event. That event gets handled concurrently with the other events that are occurring or users that are connecting to the server. In web applications built with other technologies, this would slow down the server after a large amount of users. In contrast, with a Node application, and the non-blocking event driven nature, this allows for highly scalable applications. This now allows companies that are attempting to scale, to build their apps with Node, which will prevent any slowdowns they may have had. This also means they do not have to purchase as much server space as someone using a web app that was not developed with Node. Ease of Use As previously mentioned, Node.js is written with JavaScript. Now, JavaScript was always used to add functionality to the frontend of applications. But with the addition of Node.js, you can now write the entire application in JavaScript. This now makes it so much easier to be a frontend developer who can edit some backend code, or be a backend engineer who can play around with some frontend code. This in turn makes it so much easier to become a Full Stack Engineer. You do not really need to know anything new except the basic concepts of how things work in the backend. As a result, we have recently seen the rise in a full stack JavaScript developer. This also reduces the complexity of working with multiple languages; it minimizes any confusion that might arise when you have to switch from JavaScript on the front end to whatever language would have been used on the backend. Open Source Community When Node was released, NPM, node package manager, was also given to the public. The Node package manager does exactly what it says on the tin: it allows developers to quickly add and use third party libraries and frameworks in their code. If you have used Node, then you can vouch for me here when I say there is almost always a package that you can use in your application that can make it easier to develop your application or automate a larger task. There are packages to help create http servers, help with image processing, and help with unit testing. If you need it, it’s probably been made. The even more awesome part about this community is that it’s growing by the day, and people are extremely active by contributing the many open source packages out there to help developers with various needs. This increases the productivity of all developers that are using Node in their application because they can shift the focus from something that is not that important in their application, to the main purpose of it. Aid in Frontend Development With the release of Node it did not only benefit the backend side of development, it also benefitted the frontend. With new frameworks that can be used on the frontend such as React.js or virtual-dom, these are all installed using NPM. With packages like browserify you can also use Node’s require to use packages on the frontend that normally would be used on the backend! You can be even more productive and develop things faster on the front end as well! Conclusion Node.js is definitely changing web development for the better. It is making engineers more productive with the use of one language across the entire stack. So, my question to you is, if you have not tried out Node in your application, what are you waiting for? Do you not like being more productive? If you enjoyed this post, tweet about your opinion of how Node.js changed web development. If you dislike Node.js, I would love to hear your opinion as well! About the author Antonio Cucciniello is a Software Engineer with a background in C, C++ and JavaScript (Node.Js) from New Jersey. His most recent project called Edit Docs is an Amazon Echo skill that allows users to edit Google Drive files using your voice. He loves building cool things with software, reading books on self-help and improvement, finance, and entrepreneurship. Follow him on twitter @antocucciniello, and follow him on GitHub here: https://github.com/acucciniello.

0
0
8149

article-image-machine-learning-apis-for-google-cloud-platform

Amey Varangaonkar

28 Jun 2018

7 min read

Machine learning APIs for Google Cloud Platform

Amey Varangaonkar

28 Jun 2018

7 min read

Google Cloud Platform (GCP) is considered to be one of the Big 3 cloud platforms among Microsoft Azure and AW. GCP is widely used cloud solutions supporting AI capabilities to design and develop smart models to turn your data into insights at a cheap, affordable cost. The following excerpt is taken from the book 'Cloud Analytics with Google Cloud Platform' authored by Sanket Thodge. GCP offers many machine learning APIs, among which we take a look at the 3 most popular APIs: Cloud Speech API A powerful API from GCP! This enables the user to convert speech to text by using a neural network model. This API is used to recognize over 100 languages throughout the world. It can also support filter of unwanted noise/ content from a text, under various types of environments. It supports context-awareness recognition, works on any device, any platform, anywhere, including IoT. It has features like Automatic Speech Recognition (ASR), Global Vocabulary, Streaming Recognition, Word Hints, Real-Time Audio support, Noise Robustness, Inappropriate Content Filtering and supports for integration with other APIs of GCP. The architecture of the Cloud Speech API is as follows: In other words, this model enables speech to text conversion by ML. The components used by the Speech API are: REST API or Google Remote Procedure Call (gRPC) API Google Cloud Client Library JSON API Python Cloud DataLab Cloud Data Storage Cloud Endpoints The applications of the model include: Voice user interfaces Domotic appliance control Preparation of structured documents Aircraft / direct voice outputs Speech to text processing Telecommunication It is free of charge for 15 seconds per usage, up to 60 minutes per month. More than that will be charged at $0.006 per usage. Now, as we have learned about the concepts and the applications of the model, let's learn some use cases where we can implement the model: Solving crimes with voice recognition: AGNITIO, A voice biometrics specialist partnered with Morpho (Safran) to bring Voice ID technology into its multimodal suite of criminal identification products. Buying products and services with the sound of your voice: Another most popular and mainstream application of biometrics, in general, is mobile payments. Voice recognition has also made its way into this highly competitive arena. A hands-free AI assistant that knows who you are: Any mobile phone nowadays has voice recognition software in the form of AI machine learning algorithms. Cloud Translation API Natural language processing (NLP) is a part of artificial intelligence that focuses on Machine Translation (MT). MT has become the main focus of NLP group for many years. MT deals with translating text from the source language to text in the target language. Cloud Translation API provides a graphical user interface to translate an inputted string of a language to targeted language, it’s highly responsive, scalable and dynamic in nature. This API enables translation among 100+ languages. It also supports language detection automatically with accuracy. It provides a feature to read a web page contents and translate to another language, and need not be text extracted from a document. The Translation API supports various features such as programmatic access, text translation, language detection, continuous updates and adjustable quota, and affordable pricing. The following image shows the architecture of the translation model: In other words, the cloud translation API is an adaptive Machine Translation Algorithm. The components used by this model are: REST API Cloud DataLab Cloud data storage Python, Ruby Clients Library Cloud Endpoints The most important application of the model is the conversion of a regional language to a foreign language. The cost of text translation and language detection is $20 per 1 million characters. Use cases Now, as we have learned about the concepts and applications of the API, let's learn two use cases where it has been successfully implemented: Rule-based Machine Translation Local Tissue Response to Injury and Trauma We will discuss each of these use cases in the following sections. Rule-based Machine Translation The steps to implement rule-based Machine Translation successfully are as follows: Input text Parsing Tokenization Compare the rules to extract the meaning of prepositional phrase Find word of inputted language to word of the targeted language Frame the sentence of the targeted language Local tissue response to injury and trauma We can learn about the Machine Translation process from the responses of a local tissue to injuries and trauma. The human body follows a process similar to Machine Translation when dealing with injuries. We can roughly describe the process as follows: Hemorrhaging from lesioned vessels and blood clotting Blood-borne physiological components, leaking from the usually closed sanguineous compartment, are recognized as foreign material by the surrounding tissue since they are not tissue-specific Inflammatory response mediated by macrophages (and more rarely by foreign-body giant cells) Resorption of blood clot Ingrowth of blood vessels and fibroblasts, and the formation of granulation tissue Deposition of an unspecific but biocompatible type of repair (scar) tissue by fibroblasts Cloud Vision API Cloud Vision API is powerful image analytic tool. It enables the users to understand the content of an image. It helps in finding various attributes or categories of an image, such as labels, web, text, document, properties, safe search, and code of that image in JSON. In labels field, there are many sub-categories like text, line, font, area, graphics, screenshots, and points. How much area of graphics involved, text percentage, what percentage of empty area and area covered by text, is there any image partially or fully mapped in web are included web contents. The document consists of blocks of the image with detailed description, properties show that the colors used in image is visualized. If any unwanted or inappropriate content is removed from the image through safe search. The main features of this API are label detection, explicit content detection, logo and landmark detection, face detection, web detection, and to extract the text the API used Optical Character Reader (OCR) and is supported for many languages. It does not support face recognition system. The architecture for the Cloud Vision API is as follows: We can summarize the functionalities of the API as extracting quantitative information from images, taking the input as an image and the output as numerics and text. The components used in the API are: Client Library REST API RPC API OCR Language Support Cloud Storage Cloud Endpoints Applications of the API include: Industrial Robotics Cartography Geology Forensics and Military Medical and Healthcare Cost: Free of charge for the first 1,000 units per month; after that, pay as you go. Use cases This technique can be successfully implemented in: Image detection using an Android or iOS mobile device Retinal Image Analysis (Ophthalmology) We will discuss each of these use cases in the following topics. Image detection using Android or iOS mobile device Cloud Vision API can be successfully implemented to detect images using your smartphone. The steps to do this are simple: Input the image Run the Cloud Vision API Executes methods for detection of Face, Label, Text, Web and Document properties Generate the response in the form of phrase or string Populate the image details as a text view Retinal Image Analysis – ophthalmology Similarly, the API can also be used to analyze retinal images. The steps to implement this are as follows: Input the images of an eye Estimate the retinal biomarkers Do the process to remove the effected portion without losing necessary information Identify the location of specific structures Identify the boundaries of the object Find similar regions in two or more images Quantify the image with retinal portion damage You can learn a lot more about the machine learning capabilities of GCP on their official documentation page. If you found the above excerpt useful, make sure you check out our book 'Cloud Analytics with Google Cloud Platform' for more information on why GCP is a top cloud solution for machine learning and AI. Read more Google announces Cloud TPUs on the Cloud Machine Learning Engine (ML Engine) How machine learning as a service is transforming cloud Google announce the largest overhaul of their Cloud Speech-to-Text

0
0
8098

article-image-python-software-foundation-jetbrains-python-developer-survey

Aaron Lazar

24 May 2018

7 min read

What the Python Software Foundation & Jetbrains 2017 Python Developer Survey had to reveal

Aaron Lazar

24 May 2018

7 min read

Python Software Foundation together with Jetbrains conduct their developer survey every year and at the end of 2017, over 9,500 developers from all over the world participated in this insightful Python developer survey. The learnings are pretty interesting and so, I thought I’d quickly summarise the most relevant points here. So here we go. TL;DR: Adoption of Python 3 is growing rapidly, but 25% of developers are yet to migrate to Python 3 despite the closely looming deadline (1st of Jan, 2020). There are as many Python developers doing primarily data science as there are those focused on web development. This is quite different from the 2016 findings, where there was a difference of 17% between Web and Data. A majority of Python developers use JavaScript, HTML/CSS and SQL along with Python. Django, Pandas, NumPy and Matplotlib are the most popular python frameworks. Jupyter Notebook and Docker are the most popular technologies used along with Python. Among cloud platforms, AWS is the most popular. Both editions of PyCharm (Community and Professional) are the most popular tools for Python development, followed by SublimeText. Code autocompletion, code refactorings and writing unit tests are the most widely used features in Python. More than half the respondents had a full-time job, working on Python. A majority of respondents held the role of a Developer/Programmer and belonged to the age group 21-39. Most of the developers are located in the US, India and China. The above stats are quite interesting no doubt. This got me thinking about the why behind those numbers. Here’s my perspective on some of those findings. I would love to hear yours in the comments section below. How is Python being used? Starting with the usage of Python, the survey revealed that close to 80% of the respondents used Python as their primary language for development. When asked which languages they generally used it with, the top responses were JavaScript, HTML/CSS and SQL. On the other hand, a lot of Java developers and those using Bash/shell, seem to use Python as their secondary language. This shows that Python is quite interoperable with a number of other languages, making it versatile to use in web, enterprise and server side scripting. Now when it comes to what tasks Python is used for in day to day development, it wasn’t a surprise when respondents mentioned Data Analysis. More than 50% use Python as their primary language for data analysis, however, only 32% claimed that they used it for Machine Learning. On the other hand, 54% mentioned that they used it for web development. 36% responded that they used Python for DevOps and system administration purposes. This isn’t surprising as most developers tend to stick to a particular tool/stack as far as possible. Developers also responded that they used Python the most for Web Development, apart from anything else, with Machine Learning + Data Analysis close on its heels. Most DevOps and Sys admins use Python as their secondary language - that might be because shell/bash are their primary languages. In the 2016 survey, the percentage of web developers was much more than ML/Data Analysts, but the difference has reduced greatly. What roles do these developers hold? When asked what roles these developers hold, the responses were quite interesting! While nearly a quarter were in a combination of Data Analysis and Machine Learning roles, another quarter were in a combination of Data Analysis and Web Development! 15% claimed to be in Web Development and Machine Learning. This relationship, although quite unlikely, is extremely interesting and worth exploring further. One reason could be that developers are building machine learning solutions that are offered to customers as a web application, rather than as a desktop application. Another reason could also be that a lot of web apps these days are becoming more data driven and require some kind of machine learning components running under the hood. What version of Python are developers rolling with and what tools do they use it with? A very surprising fact that surfaced from the survey was that 25% of developers still haven’t migrated their codebases to Python 3 and are still working with Python 2. This is quite astonishing, since the support for Python 2 will be discontinued in less than two years (from Jan 1, 2020 to be precise). Although, the adoption for Python 3 has been growing steadily over the years, most of the developers who were still using Python 2 turned out to be web developers. This is so because data scientists might have moved into using Python quite recently, as compared to web developers who might have been using Python for a long time and hence, haven’t migrated their legacy code. What are their prefered tool set with Python? When asked about the tools that developers used, the web developers responded that a majority of them used Django(76%), while 53% used Requests and 49% used Flask. When it came to GUI frameworks, 9% of developers used PyQT / PyGTK / wxPython while 6% used TkInter. 29% of these developers mentioned that they used scientific libraries like NumPy / pandas / Matplotlib / scipy. This is quite supportive of the overlap between both the GUI development and Data Science roles. On the other hand, Data Scientists responded that 65% used NumPy / pandas / Matplotlib / scipy. 38% used Keras / Theano / TensorFlow / scikit-learn, while 31% and 27% used Django and Flask respectively. Django was a clear winner in the tools section, with an overall of 41% developers using it. When asked about what tools they used along with Python, the web developers responded that 47% used Docker, 46% used an ORM like SQLAlchemy, PonyORM, etc. and 40% used Redis. 27% of them used Jupyter Notebook. The Data Scientists on the other hand, used Jupyter Notebook a lot (52%). 40% of them used Anaconda and 23% Docker. Of the various cloud platforms, developers chose AWS the most (65%). When it came to Python features that were used the most, Code autocompletion (84%), code refactorings (82%) and writing unit tests (81%), made the top of the list. 75% developers used SQL databases while only 46% used NoSQL. Of the various IDEs and Editors, PyCharm in both its versions, community and professional, was the most popular, closely tailed by Sublime, Vim, IDLE, Atom, and VS Code. While Web Developers preferred PyCharm, data scientists prefer Jupyter Notebook. Developer Profile: Employment, Job Roles and Experience Unsurprisingly, 52% of Python developers claimed that they were in a full-time job. This ties in well with the 2018 StackOverflow Developer survey which labeled Python as the “most wanted” programming language. So developers out there, if you’re well versed with Python, you’re more likely to be hired. Talking about job roles, 73% held the role of a Developer/Programmer, while 19% held the role of a Data Analyst and 18% an Architect. Interestingly, 7% of them held the role of a CIO / CEO / CTO. In terms of years of experience, the results were well balanced with almost as many developers having more than 5 years of experience as those with less than 5 years of experience. 67% of the respondents were in the age group of 21-39, meaning that a majority of young folk seem to be using Python. If you’re one of them, and are looking to progress in your career, check out our extensive catalog of Python titles. As for geographic location of the developers, 18% were from the US while 13% were from India and 7% from China. Should you move to Python 3? 7 Python experts’ opinions Python experts talk Python on Twitter: Q&A Recap Introducing Dask: The library that makes scalable analytics in Python easier

0
2
8097

article-image-generative-adversarial-networks-gans-next-milestone-deep-learning

Savia Lobo

09 Nov 2017

7 min read

Generative Adversarial Networks (GANs): The next milestone In Deep Learning

Savia Lobo

09 Nov 2017

7 min read

With the rise in popularity of deep learning as a concept and a paradigm, neural networks are captivating the interest of machine learning enthusiasts and developers alike, by being able to replicate the human brain for efficient predictions, image recognition, text recognition, and much more. However, can these neural networks do something more, or are they just limited to predictions? Can they self-generate new data by learning from a training dataset? Generative Adversarial networks (GANs) are here, to answer all these questions. So, what are GANs all about? Generative Adversarial Networks follow unsupervised machine learning, unlike traditional neural networks. When a neural network is taught to identify a bird, it is fed with a huge number of images including birds, as training data. Each picture is labeled before it is put to use in training the models. This labeling of data is both costly and time-consuming. So, how can you train your neural networks by giving it less data to train on? GANs are of a great help here. They cast out an easy way to train the DL algorithms by slashing out the amount of data required to train the neural network models, that too, with no labeling of data required. The architecture of a GAN includes a generative network model(G), which produces fake images or texts, and an adversarial network model--also known as the discriminator model (D)--that distinguishes between the real and the fake productions by comparing the content sent by the generator with the training data it has. Both of these are trained separately by feeding each of them with training data and a competitive goal. Source: Learning Generative Adversarial Networks GANs in action GANs were introduced by Ian Goodfellow, an AI researcher at Google Brain. He compares the generator and the discriminator models with a counterfeiter and a police officer. “You can think of this being like a competition between counterfeiters and the police,” Goodfellow said. “Counterfeiters want to make fake money and have it look real, and the police want to look at any particular bill and determine if it’s fake.” Both the discriminator and the generator are trained simultaneously to create a powerful GAN architecture. Let’s peek into how a GAN model is trained- Specify the problem statement and state the type of manipulation that the GAN model is expected to carry out. Collect data based on the problem statement. For instance, for image manipulation, a lot of images are required to be collected to feed in. The discriminator is fed with an image; one from the training set and one produced by the generator The discriminator can be termed as ‘successfully trained’ if it returns 1 for the real image and 0 for the fake image. The goal of the generator is to successfully fool the discriminator and getting the output as 1 for each of its generated image. In the beginning of the training, the discriminator loss--the ability to differentiate real and fake image or data--is minimal. As the training advances, the generator loss decreases and the discriminator loss increases, This means, the generator is now able to generate real images. Real world applications of GANs The basic application of GANs can be seen in generating photo-realistic images. But there is more to what GANs can do. Some of the instances where GANs are majorly put to use include: Image Synthesis Image Synthesis is one of the primary use cases of GANs. Here, multilayer perceptron models are used in both the generator and the discriminator to generate photo-realistic images based on the training dataset of the images. Text-to-image synthesis Generative Adversarial networks can also be utilized for text-to-image synthesis. An example of this is in generating a photo-realistic image based on a caption. To do this, a dataset of images with their associated captions are given as training data. The dataset is first encoded using a hybrid neural network called the character-level convolutional Recurrent Neural network, which creates a joint representation of both in multimodal space for both the generator and the discriminator. Both Generator and Discriminator are then trained based on this encoded data. Image Inpainting Images that have missing parts or have too much of noise are given as an input to the generator which produces a near to real image. For instance, using TensorFlow framework, DCGANs (Deep Convolutional GANs), can generate a complete image from a broken image. DCGANs are a class of CNNs that stabilizes GANs for efficient usage. Video generation Static images can be transformed into short scenes with plausible motions using GANs. These GANs use scene dynamics in order to add motion to static images. The videos generated by these models are not real but illusions. Drug discovery Unlike text and image manipulation, Insilico medicine uses GANs to generate an artificially intelligent drug discovery mechanism. To do this, the generator is trained to predict a drug for a disease which was previously incurable.The task of the discriminator is to determine whether the drug actually cures the disease. Challenges in training a GAN Whenever a competition is laid out, there has to be a distinct winner. In GANs, there are two models competing against each other. Hence, there can be difficulties in training them. Here are some challenges faced while training GANs: Fair training: While training both the models, precaution has to be taken that the discriminator does not overpower the generator. If it does, the generator would fail to train effectively. On the other hand, if the discriminator is lenient, it would allow any illegitimate content to be generated. Failure to understand the number of objects and the dimensions of objects, present in a particular image. This usually occurs during the initial learning phase. For instance, GANs, at times output an image which ends up having more than two eyes, which is not normal in the real world. Sometimes, it may present a 3D image like a 2D one. This is because they cannot differentiate between the two. Failure to understand the holistic structure: GANs lack in identifying universally correct images. It may generate an image which can be totally opposed to how they look in real. For instance, a cat having an elongated body shape, or a cow standing on its hind legs, etc. Mode collapse is another challenge, which occurs when a low variation dataset is processed by a GANs. Real world includes complex and multimodal distributions, where data may have different concentrated sub-groups. The problem here is, the generator would be able to yield images based on anyone sub-group resulting in an inaccurate output. Thus, causing a mode collapse. To tackle these and other challenges that arise while training GANs, researchers have come up with DCGANs (Deep Convolutional GANs), WassersteinGANs, CycleGANs to ensure fair training, enhance accuracy, and reduce the training time. AdaGANs are implemented to eliminate mode collapse problem. Conclusion Although the adoption of GANs is not as widespread as one might imagine, there’s no doubt that they could change the way unsupervised machine learning is used today. It is not too far-fetched to think that their implementation in the future could find practical applications in not just image or text processing, but also in domains such as cryptography and cybersecurity. Innovations in developing newer GAN models with improved accuracy and lesser training time is the key here - but it is something surely worth keeping an eye on.

0
0
8088

Packt Editorial Staff

09 Apr 2018

6 min read

A brief history of Blockchain

Packt Editorial Staff

09 Apr 2018

6 min read

History - where do we start? Blockchain was introduced with the invention of Bitcoin in 2008. Its practical implementation then occurred in 2009. Of course, both Blockchain and Bitcoin are very different, but you can't tell the full story behind the history of Blockchain without starting with Bitcoin. Electronic cash before Blockchain The concept of electronic cash or digital currency is not new. Since the 1980s, e-cash protocols have existed based on a model proposed by David Chaum. This is an extract from the new edition of Mastering Blockchain. Just as you need to understand the concept of distributed systems is to properly understand Blockchain, you also need to understand electronic cash. This concept pre-dates Blockchain and Bitcoin, but without it, we would certainly not be where we are today. Two fundamental e-cash system issues need to be addressed: accountability and anonymity. Accountability is required to ensure that cash is spendable only once (double-spend problem) and that it can only be spent by its rightful owner. Double spend problem arises when same money can be spent twice. As it is quite easy to make copies of digital data, this becomes a big issue in digital currencies as you can make many copies of same digital cash. Anonymity is required to protect users' privacy. As with physical cash, it is almost impossible to trace back spending to the individual who actually paid the money. David Chaum solved both of these problems during his work in the 1980s by using two cryptographic operations, namely blind signatures and secret sharing. Blind signatures allow for signing a document without actually seeing it, and secret sharing is a concept that enables the detection of double spending, that is using the same e-cash token twice (double spending). In 2009, the first practical implementation of an electronic cash (e-cash) system named Bitcoin appeared. The term cryptocurrency emerged later. For the very first time, it solved the problem of distributed consensus in a trustless network. It used public key cryptography with a Proof of Work (PoW) mechanism to provide a secure, controlled, and decentralized method of minting digital currency. The key innovation was the idea of an ordered list of blocks composed of transactions and cryptographically secured by the PoW mechanism. Other technologies that used something like a precursor to Bitcoin, include Merkle trees, hash functions, and hash chains. Looking at all the technologies mentioned earlier and their relevant history, it is easy to see how concepts from electronic cash schemes and distributed systems were combined to create Bitcoin and what now is known as Blockchain. This concept can also be visualized with the help of the following diagram: Blockchain and Sakoshi Nakamoto In 2008, a groundbreaking paper entitled Bitcoin: A Peer-to-Peer Electronic Cash System was written on the topic of peer-to-peer electronic cash under the pseudonym Satoshi Nakamoto. It introduced the term chain of blocks. No one knows the actual identity of Satoshi Nakamoto. After introducing Bitcoin in 2009, he remained active in the Bitcoin developer community until 2011. He then handed over Bitcoin development to its core developers and simply disappeared. Since then, there has been no communication from him whatsoever, and his existence and identity are shrouded in mystery. The term chain of blocks evolved over the years into the word Blockchain. Since that point, the history of Blockchain is really the history of its application in different industries. The most notable area is unsurprisingly within finance. Blockchain has been shown to improve the speed and security of financial transactions. While it hasn't yet become embedded in the mainstream of the financial sector, it surely only remains a matter of time before it begins to take hold. How it has evolved in recent years In Blockchain: Blueprint for a New Economy, Melanie Swann identifies three different tiers of Blockchain. These three tiers all showcase how Blockchain is currently evolving. It's worth noting that these various tiers or versions aren't simple chronological points in the history of Blockchain. The lines between each are blurred, and it ultimately depends on how Blockchain technology is being applied that different features and capabilities will be appear. Blockchain 1.0: This tier was introduced with the invention of Bitcoin, and it is primarily used for cryptocurrencies. Also, as Bitcoin was the first implementation of cryptocurrencies, it makes sense to categorize this first generation of Blockchain technology to include only cryptographic currencies. All alternative cryptocurrencies, as well as Bitcoin, fall into this category. It includes core applications such as payments and applications. This generation started in 2009 when Bitcoin was released and ended in early 2010. Blockchain 2.0: This second Blockchain generation is used by financial services and smart contracts. This tier includes various financial assets, such as derivatives, options, swaps, and bonds. Applications that go beyond currency, finance, and markets are incorporated at this tier. Ethereum, Hyperledger, and other newer Blockchain platforms are considered part of Blockchain 2.0. This generation started when ideas related to using blockchain for other purposes started to emerge in 2010. Blockchain 3.0: This third Blockchain generation is used to implement applications beyond the financial services industry and is used in government, health, media, the arts, and justice. Again, as in Blockchain 2.0, Ethereum, Hyperledger, and newer blockchains with the ability to code smart contracts are considered part of this blockchain technology tier. This generation of Blockchain emerged around 2012 when multiple applications of Blockchain technology in different industries were researched. Blockchain X.0: This generation represents a vision of Blockchain singularity where one day there will be a public Blockchain service available that anyone can use just like the Google search engine. It will provide services for all realms of society. It will be a public and open distributed ledger with general-purpose rational agents (Machina economicus) running on a Blockchain, making decisions, and interacting with other intelligent autonomous agents on behalf of people, and regulated by code instead of law or paper contracts. This does not mean that law and contracts will disappear, instead, law and contracts will be implementable in code. Like any history, this history of Blockchain isn't exhaustive. But it does hopefully give you an idea of how it has developed to where we are today. Check out this tutorial to write your first Blockchain program.

0
0
8082

article-image-what-is-quantum-entanglement

Amarabha Banerjee

05 Aug 2018

3 min read

What is Quantum Entanglement?

Amarabha Banerjee

05 Aug 2018

3 min read

Einstein described it as “Spooky action at a distance”. Quantum entanglement is a phenomenon observed in photons where particles share information of their state - even if separated by a huge distance. This state sharing phenomenon happens almost instantaneously. Quantum particles can be in any possible state until their state is measured by an observer. These states are called Eigen-Values. In case of quantum entanglement, two particles separated by several miles of distance, when observed, change into the same state. Quantum entanglement is hugely important for modern day computation tasks. The reason is that the state information between photons travel sometimes at speeds like 10k times the speed of light. This if implemented in physical systems, like quantum computers, can be a huge boost. Source: picoquant One important concept for us to understand this idea is ‘Qubit’. What is a Qubit? It’s the unit of information in Quantum computing. Like ‘Bit’ in case of normal computers. A bit can be represented by two states - ‘0’ or ‘1’. Qbits are also like ‘bits’, but they are governed by the weirder rules of Quantum Computing. Qubits don’t just contain pure states like ‘0’ and ‘1’, but they can also exist as superposition of these two states like {|0>,|1>},{ |1>,|0>}, {|0>,|0>}, {|1>,|1>}. This particular style of writing particle states is called the Dirac Notation. Because of these unique superposition of states, the quantum particles get entangled and share their state related information. A recent research experiment by a Chinese group has claimed to have packed 18 Qubits of information in just 6 entangled photons. This is revolutionary. What this basically means is that if one bit can pack in three times the information that it can carry presently, then our computers would become three times faster and smoother to work with. The reasons which make this a great start for future implementation of faster and practical quantum computers are: It’s very difficult to entangle so many electrons There are instances of more than 18 qubits getting packed into a larger number of photons, however the degree of entanglement has been much simpler Entanglement of each new particle takes increasingly more computer simulation time Introducing each new qubit creates a separate simulation taking up more processing time. The possible reason why this experiment has worked might be credited to the multiple degrees of freedom that photons can have. This particular experiment has been performed using Photons in a networking system. The fact that such a system allows multiple degrees of freedom for the Photon meant that this result is specific to this particular quantum system. It would be difficult to replicate the results in other systems like a Superconducting Network. Still this result means a great deal for the progress of quantum computing systems and how they can evolve to be a practical solution and not just remain in theory forever. Quantum Computing is poised to take a quantum leap with industries and governments on. PyCon US 2018 Highlights: Quantum computing, blockchains and serverless rule! Q# 101: Getting to know the basics of Microsoft’s new quantum computing language

0
0
8067

article-image-amazon-sagemaker-machine-learning-cloud-easy

Amey Varangaonkar

12 Apr 2018

5 min read

Amazon Sagemaker makes machine learning on the cloud easy

Amey Varangaonkar

12 Apr 2018

5 min read

Amazon Sagemaker was launched by Amazon back in November 2017. It was built with the promise of simplifying machine learning on the cloud. The software was a response not only to the increasing importance of machine learning, but also the fact that there is a demand to perform machine learning in the cloud. Amazon Sagemaker is clearly a smart move by Amazon that will consolidate the dominance of AWS in the cloud market. What is Amazon Sagemaker? Amazon Sagemaker is Amazon’s premium cloud-based service which serves as a platform for machine learning developers and data scientists to build, train and deploy machine learning models on the cloud. One of the features that makes Sagemaker stand out from the rest is that it is business-ready. This means machine learning models can be optimized for high performance and deployed at scale to work on data with varying sizes and complexity. The basic intention of Sagemaker, as Vogels mentioned in his keynote, is to remove any barriers that slow down the machine learning process for developers. In a standard machine learning process, a developer spends most of the time doing the following standard tasks: Collecting, cleaning and preparing the training data set. Selecting the most appropriate algorithm for the machine learning problem Training the model for accurate prediction Optimizing the model’s performance Integrating the model with the application Deploying the application to production Most of these tasks require a lot of expertise, and more importantly, time and efforts. Not to mention the computational resources such as storage space and processing memory. The larger the dataset, the bigger this problem becomes. Amazon Sagemaker removes these complexities by providing a solid platform with built-in modules that can be used together or individually to complete each of the above tasks with relative ease. How Amazon Sagemaker Works Amazon Sagemaker offers a lot of options for machine learning developers to train and optimize their machine learning models to work at scale. For starters, Sagemaker comes integrated with hosted Jupyter notebooks to allow developers to visually explore and analyze their dataset. You can also move your data directly from popular Amazon databases such as RDS, DynamoDB and Redshift into S3 and conduct your analysis there. The simple block diagram below demonstrates the core working of Amazon Sagemaker: Amazon Sagemaker includes 12 high performance, production-ready algorithms which can be used to build and deploy models at scale. Some of the popular ones include k-means clustering, Principal Component Analysis (PCA), neural topic modeling, and more. It comes pre-configured with popular machine learning and deep learning frameworks such as Tensorflow, PyTorch, Apache MXNet and more, but you can also use your own framework without any hassle. Once your model is trained, Sagemaker makes use of the AWS’ auto-scaled clusters to deploy the model, making sure the model doesn’t lack in performance and is highly available at all times. Not just that, Sagemaker also includes built-in testing capabilities for you to test and check your model for any issues, before it can be deployed for production. Benefits of using Amazon Sagemaker Business are likely to adopt Amazon Sagemaker, mainly because of the fact that it makes the whole machine learning process so effortless. With Sagemaker, it becomes very easy to build and deploy smarter applications that give accurate predictions, and thereby help increase the business profitability. Significantly reduces time: With built-in modules, Sagemaker significantly reduces the time required to do a variety of machine learning tasks, and the models can be deployed to production in very little time. This is important for businesses, as near-real time insights obtained from smart applications help them optimize their processes quickly, and effectively get an edge over their competition. Effortless and more productive machine learning: By virtue of the one-click training and deployment feature offered by Sagemaker, machine learning engineers and developers can now focus on asking the right questions of the data, and focus on the results rather than the process. They can also devote more time to optimizing the model rather than focusing on collecting and cleaning the data, which takes up most of their time. Flexibility in using the algorithms and frameworks: With Sagemaker, developers have the freedom to choose the best-possible algorithm and tool for performing machine learning effectively. Easy integration, access and optimization: The models trained using Sagemaker can be integrated into an existing business application seamlessly, and are optimized for speed and high performance. Backed by the computational power of AWS, business can rest assured their applications will continue to perform optimally without any risk of failure. Sagemaker - Amazon’s answer to Cloud Auto ML In a 3-way cloud war between Google, Microsoft and Amazon, it is clear Google and Amazon are trying to go head to head in order to establish their supremacy in the market, especially in the AI space. Sagemaker is Amazon’s answer to Google’s Cloud Auto ML, which was made publicly available in January, and delivers a similar promise - making machine learning easier than ever for developers. With Amazon serving a large customer-base, a platform like Sagemaker helps them to create a system that runs at scale and handles vast amounts of data quite effortlessly. Amazon is yet to release any technical paper on how Sagemaker’s streaming algorithms work, but that will certainly be something to look out for in the near future. Considering Amazon identifies AI as key to their future product development, to think of Sagemaker as a better, more complete cloud service which also has deep learning capabilities is definitely not far-fetched.

0
0
8055

Amey Varangaonkar

28 Aug 2017

7 min read

Is Python edging R out in the data science wars?

Amey Varangaonkar

28 Aug 2017

7 min read

When it comes to the ‘lingua franca’ of data science, there seems to be a face-off between R and Python. R has long been established as the language of researchers and statisticians but Python has come up quickly as a bona-fide challenger, helping embed analytics as a necessity for businesses and other organizations in 2017. If a tech war does exist between the two languages, it’s a battle fought not so much on technical features but instead on the wider changes within modern business and technology. R is a language purpose-built for statistics, for performing accurate and intensive analysis. So, the fact that R is being challenged by Python — a language that is flexible, fast, and relatively easy to learn — suggests we are seeing a change in who’s actually doing data science, where they’re doing it, and what they’re trying to achieve. Python versus R — A Closer Look Let’s make a quick comparison of the two languages on aspects important to those working with data and see what we can learn about the two worlds where R and Python operate. Learning curve Python is the easier language to learn. While R certainly isn’t impenetrable, Python’s syntax marks it as a great language to learn even if you’re completely new to programming. The fact that such an easy language would come to rival R within data science indicates the pace at which the field is expanding. More and more people are taking on data-related roles, possibly without a great deal of programming knowledge — Python makes the barrier to entry much lower than R. That said, once you get to grips with the basics of R, it becomes relatively easier to learn the more advanced stuff. This is why statisticians and experienced programmers find R easier to use. Packages and libraries Many R packages are in-built. Python, meanwhile, depends upon a range of external packages. This obviously makes R much more efficient as a statistical tool — it means that if you’re using Python you need to know exactly what you’re trying to do and what external support you’re going to need. Data Visualization R is well-known for its excellent graphical capabilities. This makes it easy to present and communicate data in varied forms. For statisticians and researchers, the importance of that is obvious. It means you can perform your analysis and present your work in a way that is relatively seamless. The ggplot2 package in R, for example, allows you to create complex and elegant plots with ease and as a result, its popularity in the R community has increased over the years. Python also offers a wide range of libraries which can be used for effective data storytelling. The breadth of external packages available with Python means the scope of what’s possible is always expanding. Matplotlib has been a mainstay of Python data visualization. It’s also worth remarking on upcoming libraries like Seaborn. Seaborn is a neat little library that sits on top of Matplotlib, wrapping its functionality and giving you a neater API for specific applications. So, to sum up, you have sufficient options to perform your data visualization tasks effectively — using either R or Python! Analytics and Machine Learning Thanks to libraries like scikit-learn, Python helps you build machine learning systems with relative ease. This takes us back to the point about barrier to entry. If machine learning is upending how we use and understand data, it makes sense that more people want a piece of the action without having to put in too much effort. But Python also has another advantage; it’s great for creating web services where data can be uploaded by different people. In a world where accessibility and data empowerment have never been more important (i.e., where everyone takes an interest in data, not just the data team), this could prove crucial. With packages such as caret, MICE, and e1071, R too gives you the power to perform effective machine learning in order to get crucial insights out of your data. However, R falls short in comparison to Python, thanks to the latter’s superior libraries and more diverse use-cases. Deep Learning Both R and Python have libraries for deep learning. It’s much easier and more efficient with Python though — most likely because the Python world changes much more quickly, new libraries and tools springing up as quickly as the data science world hooks on to a new buzzword. Theano, and most recently Keras and TensorFlow have all made a huge impact on making it relatively easy to build incredibly complex and sophisticated deep learning systems. If you’re clued-up and experienced with R it shouldn’t be too hard to do the same, using libraries such as MXNetR, deepr, and H2O — that said, if you want to switch models, you may need to switch tools, which could be a bit of a headache. Big Data With Python, you can write efficient MapReduce applications with ease, or scale your R program on Hadoop to work with petabytes of data. Both R and Python are equally good when it comes to working with Big Data, as they can be seamlessly integrated with Big Data tools such as Apache Spark and Apache Hadoop, among many others. It’s likely that it’s in this field that we’re going to see R moving more and more into industry as businesses look for a concise way to handle large datasets. This is true in industries such as bioinformatics which have a close connection with the academic world and necessarily depend upon a combination of size and accuracy when it comes to working with data. So, where does this comparison leave us? Ultimately, what we see are two different languages offering great solutions to very different problems in data science. In Python, we have a flexible and adaptable language with a vibrant community of developers working on a huge range of problems and tasks, each one trying to find more effective and more intelligent ways of doing things. In R, we have a purely statistical language with a large repository of over 8000 packages for data analysis and visualization. While Python is production-ready and is better suited for organizations looking to harness technical innovation to its advantage, R’s analytical and data visualization capabilities can make your life as a statistician or data analyst easier. Recent surveys indicate that Python commands a higher salary than R — that is because it’s a language that can be used across domains; a problem-solving language. That’s not to say that R isn’t a valuable language; rather, Python is the language that just seems to fit the times at the moment. In the end, it all boils down to your background, and the kind of data problems you want to solve. If you come from a statistics or research background and your problems only revolve around statistical analysis and visualization, then R would best fit your bill. However, if you’re a Computer Science graduate looking to build a general-purpose, enterprise-wide data model which can integrate seamlessly with the other business workflows, you will find Python easier to use. R and Python are two different animals. Instead of comparing the two, maybe it’s time we understood where and how each can be best used and then harnessed their power to the fullest to solve our data problems. One thing is for sure, though — neither is going away anytime soon. Both R and Python occupy a large chunk of the data science market-share today, and it will take a major disruption to take either one of them out of the equation completely.

0
1
8052

article-image-unity-machine-learning-agents-transforming-games-with-artificial-intelligence

Amey Varangaonkar

30 Mar 2018

4 min read

Unity Machine Learning Agents: Transforming Games with Artificial Intelligence

Amey Varangaonkar

30 Mar 2018

4 min read

Unity has undoubtedly been one of the leaders when it comes to developing cross-platform products - going from strengths to strengths in developing visually stimulating 2D as well as 3D games and simulations.With Artificial Intelligence revolutionizing the way games are being developed these days, Unity have identified the power of Machine Learning and introduced Unity Machine Learning Agents. With this, they plan on empowering the game developers and researchers in their quest to develop intelligent games, robotics and simulations. What are Unity Machine Learning Agents? Traditionally, game developers have been hard-coding the behaviour of the game agents. Although effective, this is a tedious task and it also limits the intelligence of the agents. Simply put, the agents are not smart enough. To overcome this obstacle, Unity have simplified the training process for the game developers and researchers by introducing Unity Machine Learning Agents (ML-Agents, in short). Through just a simple Python API, the game agents can be now trained to use deep reinforcement learning, an advanced form of machine learning, to learn from their actions and modify their behaviour accordingly. These agents can then be used to dynamically modify the difficulty of the game. How do they work? As mentioned earlier, the Unity ML-Agents are designed to work based on the concept of deep reinforcement learning, a branch of machine learning where the agents are trained to learn from their own actions. Here is a simple flowchart to demonstrate how reinforcement learning works: The reinforcement learning training cycle The learning environment to be configured for the ML-agents consists of 3 primary objects: Agent: Every agent has a unique set of states, observations and actions within the environment, and is assigned rewards for particular events. Brain: A brain decides what action any agent is supposed to take in a particular scenario. Think of it as a regular human brain, which basically controls the bodily functions. Academy: This object contains all the brains within the environment To train the agents, a variety of scenarios are made possible by varying the connection of different components (explained above) of the environment. Some are single agents, some simultaneous single agents, and others could be co-operative and competitive multi-agents and more. You can read more about these possibilities on the official Unity blog. Apart from the way these agents are trained, Unity are also adding some cool new features in these ML-agents. Some of these are: Monitoring the agents’ decision-making to make it more accurate Incorporating curriculum learning, by which the complexity of the tasks can eventually be increased to aid more effective learning Imitation learning is a newly-introduced feature wherein the agents simply mimic the actions we want them to perform, rather than they learning on their own. What next for Unity Machine Learning Agents? Unity recently announced the release of v0.3 beta SDK of the ML-agents, and have been making significant progress in this domain to develop smarter, more intelligent game agents which can be used with the Unity game engine. Still very much in the research phase, these agents can also be used as an example by academic researchers to study the complex behaviour of trained models in different environments and scenarios where the variables associated with the in-game physics and visual appearance can be altered. Going forward, these agents can also be used by enterprises for large scale simulations, in robotics and also in the development of autonomous vehicles. These are interesting times for game developers, and Unity in particular, in their quest for developing smarter, cutting-edge games. Inclusion of machine learning in their game development strategy is a terrific move, although it will take some time for this to be perfected and incorporated seamlessly. Nonetheless, all the research and innovation being put into this direction certainly seems well worth it!

0
0
8021

article-image-amazon-reinvents-speech-recognition-and-machine-translation-with-ai

Amey Varangaonkar

04 Jul 2018

4 min read

How Amazon is reinventing Speech Recognition and Machine Translation with AI

Amey Varangaonkar

04 Jul 2018

4 min read

In the recently held AWS Summit in San Francisco, Amazon announced the general availability of two of its premium offerings - Amazon Transcribe and Amazon Translate. What’s so special about the two products is that customers will now be able to see the power of Artificial Intelligence in action, and use it to solve their day-to-day problems. These offerings from AWS will make it easier for startups and companies looking to adopt and integrate AI into their existing process and simplify their core tasks - especially pertaining to speech and language processing. Effective speech-to-text conversion with Amazon Transcribe In the AWS summit keynote, Amazon Solutions Architect Niranjan Hira expressed his excitement talking about the features of Amazon Transcribe; the automatic speech recognition service by AWS. This API can be integrated with the other tools and services offered by Amazon such as Amazon S3, and Quicksight. Source: YouTube Amazon Transcribe boasts wonderful features like: Simple API: It is very easy to use the Transcribe API to perform speech to text conversion, with minimum need for programming. Timestamp generation: The speech when converted to text also includes the timestamps for every word, so that tracking the word becomes easy and hassle-free. Variety of use-cases: The Transcribe API can be used to generate accurate transcripts of any audio or video file, of varied quality. Subtitle generation becomes easier using this API especially for low-quality audio recordings - customer service calls are a very good example. Easy to read text: Transcribe uses the cutting edge deep learning technology to parse text from speech without any errors. With appropriate punctuations and grammar in place, the transcripts are very easy to read and understand. Machine translation simplified with Amazon Translate Amazon Translate is a machine translation service offered by Amazon. It makes use of neural networks and advanced deep learning techniques to deliver accurate, high-quality translations. Key features of Amazon Translate include: Continuous training: The architecture of this service is built in such a way that the neural networks keep learning and improving. High accuracy: The continuous learning by the translation engines from new and varied datasets results in a higher accuracy of machine translations. The machine translation capability offered by this service is almost 30% more efficient than human translation. Easy to integrate with other AWS services: With a simple API call, Translate allows you to integrate the service within third party applications to allow real-time translation capabilities. Highly scalable: Regardless of the volume, Translate does not compromise the speed and accuracy of the machine translation. Know more about Amazon Translate from Yoni Friedman’s keynote at the AWS Summit. With all the businesses slowly migrating to cloud, it is clear all the cloud vendors - mainly Amazon, Google and Microsoft - are doing everything they can to establish their dominance. Google recently launched Cloud ML for GCP which offers machine learning and predictive analytics services improving businesses. Microsoft’s Azure Cognitive Services offer effective machine translation services as well, and are slowly gaining a lot of momentum. With these releases, the onus was on Amazon to respond, and they have done so in style. With the Transcribe and Translate APIs, Amazon’s goal of making it easier for startups and small-scale businesses to adopt AWS and incorporate AI seems to be on track. These services will also help AWS distinguish their cloud offerings, given that computing and storage resources are provided by rivals as well. Read more Verizon chooses Amazon Web Services(AWS) as its preferred cloud provider Tensor Processing Unit (TPU) 3.0: Google’s answer to cloud-ready Artificial Intelligence Amazon is selling facial recognition technology to police

0
0
7909

article-image-how-ai-is-changing-game-development

Raka Mahesa

16 Jan 2018

4 min read

How AI is changing game development

Raka Mahesa

16 Jan 2018

4 min read

0
0
7895

Tech Guides

Computer vision is growing quickly. Here's why.

Exploring Language Improvements in C# 7.2 and 7.3

CapsNet: Are Capsule networks the antidote for CNNs kryptonite?

One Shot Learning: Solution to your low data problem

How is Node.js Changing Web Development?

Machine learning APIs for Google Cloud Platform

What the Python Software Foundation & Jetbrains 2017 Python Developer Survey had to reveal

Generative Adversarial Networks (GANs): The next milestone In Deep Learning

A brief history of Blockchain

What is Quantum Entanglement?

Trending Topics

Amazon Sagemaker makes machine learning on the cloud easy

Is Python edging R out in the data science wars?

Unity Machine Learning Agents: Transforming Games with Artificial Intelligence

How Amazon is reinventing Speech Recognition and Machine Translation with AI

How AI is changing game development