Amazon reinvents Data Science at the re:invent 2017

AWS AppSync – A service for building real-time, collaborative apps with offline capabilities

AWS announced the release of AppSync, a service to build responsive web and mobile applications driven by data-intensive services in the cloud. The distinguishing feature of Appsync is the fact that shared app data is updated in real time. It also offers an offline programming model, using which the service can automatically manage the data operations for offline users as well. The service can add the app data locally for offline users and then update it to the cloud, once the device reconnects.

AWS AppSync makes use of GraphQL, an open standard data query language for retrieving data from the cloud. GraphQL works at the application layer to define operations to be performed on the retrieved data and also works on the structuring of the data retrieved.  With GraphQL-backed AppSync, developers can retrieve and manipulate data from multiple sources with ease. Thus creating robust, collaborative and data-driven apps.

H1 Instances – New Storage family

AWS also introduced a new Storage Optimized Instance family during re:invent, called the H1 instances specifically for Big Data Applications, These include data-intensive workloads such as MapReduce, distributed file systems, network file systems, log or data processing, and big data clusters. The H1 instances provide more memory per terabyte of local magnetic storage and more vCPUs as compared to the dense storage (D2) instances.

The instances in H1 are based on Intel Xeon E5-2686 v4 processors running at a base clock frequency of 2.3 GHz and come in four instance sizes (all VPC-only and HVM-only):

Source: aws.amazon.com

The two largest sizes support Intel Turbo and CPU power management, with all-core Turbo at 2.7 GHz and single-core Turbo at 3.0 GHz.

Amazon GuardDuty – Threat Analysis and Protection

Amazon’s GuardDuty is a Security Monitoring and Threat Detection service powered by Machine Learning. Announced on the 2nd day of re:invent, it helps cloud users protect their AWS accounts and workloads from potential threats. GuardDuty monitors information from multiple sources such as VPC Flow Logs, AWS CloudTrail Event Logs, and DNS logs to detect malicious or unauthorized behavior in the AWS accounts. GuardDuty then uses machine learning to find trends, patterns, and anomalies in the customer’s activities and data.

Customers can enable Amazon GuardDuty with a few clicks in the AWS Management Console. Amazon GuardDuty requires no installation or management of any agents, sensors, or network appliances. Most importantly it does not affect performance and reliability. The findings from the security activities are presented at three levels, low, medium, or high. GuardDuty can also be integrated into systems such as Splunk, Sumo Logic, and PagerDuty and in workflow systems like JIRA, ServiceNow, and Slack.

Amazon Sumerian: AI platform to build VR, AR, and 3D experiences

Re:invent also saw Amazon unveil a new platform, Amazon Sumerian, for developers to build and host VR, AR, and 3D apps. These apps can run on head-based displays like the Oculus, HTC Vive, smartphones and tablets, digital signage, and web browsers. The tools and resources included in Sumerian include:

• A web-based editor, for constructing 3D scenes, importing assets, scripting interactions and special effects, with cross-platform publishing.
• An Object library of pre-built objects and templates.
• Sumerian supports importing FBX, OBJ, and other 3D assets to use in a scene.
• A JavaScript scripting library for advanced scripting capabilities.
• Animated hosts which are lifelike 3D characters customizable for gender, voice, and language.
• Integration with Amazon Polly and Amazon Lex to add speech and natural language to Sumerian hosts.

Sumerian is free to use. Customers only pay for the storage they create.

AWS Machine Learning Partner Solutions

Next in line for AWS re:invent, was the launch of new AWS Machine Learning Partner Solutions program dedicated specifically to Machine Learning. AWS Machine Learning Partners provide solutions to help organizations address their data challenges, or enable ML workloads. They also offer SaaS-based capabilities for enhancing end applications with machine intelligence.

The current data partners include Alteryx, CrowdFlower, Paxata, etc. for Data Services, Bonsai, C3 IoT, Databricks, Dataiku, and more for Platform Solutions and Anodot, Luminoso, and many others for SaaS/API Solutions. More data partners are listed on their official site.

AWS Machine Learning Research Awards

In order to acknowledge people contributing towards the machine learning domain, Amazon announced novel, AWS Machine Learning Research Awards. This program funds university departments, faculty, Ph.D. students, and postdoctoral who are conducting research in Machine Learning.

The program would award recipients with unrestricted cash gifts and AWS credits redeemable towards Amazon’s cloud services. Selected participants would also be awarded training resources and the opportunity to attend an annual research seminar at Amazon’s Seattle HeadQuarters.

Andrew Moore, Dean of the School of Computer Science at Carnegie Mellon University says “The capabilities students have available to them at their fingertips today via the AWS Cloud are amazing, and it’s great that a sophisticated framework tool such as Apache MXNet is available. The opportunity to have the next generation of machine learning practitioners and researchers make the most of these tools is exciting and is uniquely enabled through this funding program. We can’t wait to get started.”

There are still 2 more days to go before the conclusion of the mega re:invent. Keep an eye on our website for upcoming releases. In case you want to view the live coverage you can find it here.