AWS re:Invent in Review — Part 2
Let's go over the all the major announcements from the Week-2 of the AWS re:Invent 2020.
Amazon re:Invent’s Week 2 was filled with more than a few exciting announcements. From machine learning to cloud infrastructure, there’s a lot to unpack. As promised, here’s a quick overview of the most important announcements made during keynotes and partner events during the second week.
Week 2 AWS re:Invent Coverage
In our last re:Invent roundup, we covered a few updates to Amazon SageMaker. However, even as I was wrapping it up, Amazon plowed on with a flurry of SageMaker-related announcements. Here’s what was announced –
SageMaker Feature Store: With Feature Store, Amazon is simplifying the process of securely storing, discovering, and sharing curated data used in training and prediction workflows. It is a centralized repository for ML features and is now completely integrated into Amazon SageMaker Studio.
SageMaker Clarify: Clarify is a new Amazon SageMaker feature that helps customers detect bias in machine learning models. It increases transparency by ensuring model behavior is explained to stakeholders and customers. Using Clarify, Data Scientists should be able to detect bias before training, measure bias using various metrics, explain how feature values contribute to predicted outcomes, and detect bias drift as well as feature importance drift over time.
SageMaker JumpStart: Addressing the concern that newer ML developers find it challenging to get started with Machine Learning. JumpStart allows users to discover models, solutions, and so much more intuitively within Amazon SageMaker Studio. It ensures that you can accelerate machine learning workflows with one-click access to a collection of popular models and end-to-end solutions.
SageMaker Edge Manager: Edge computing is on the rise, and Amazon brings an essential update for the audience that trains ML models in the cloud before deploying them at the edge to extract deeper insights from local data. SageMaker Edge Manager now allows developers to use the same tools on both the cloud and edge devices. It reduces the time and effort needed to get models to production while monitoring and improving model quality at the same time.
SageMaker Model Monitor: Concept Drift can cause deviations over time, and it’s critical to track them to allow for any remedial actions (if needed). With improvements to the SageMaker Model Monitor, users can now use it to detect drift in model quality, bias, and feature importance. An automated alert can now be configured to know when a model exceeds the users’ bias metric thresholds.
SageMaker Debugger: ML Model Training is a complex process that heavily depends on the dataset’s quality, the algorithm, its parameters, and the infrastructure. However, for a complicated training job that takes hours or even days, an error can prove disastrous. The SageMaker Debugger can now profile machine learning models to find and fix training issues caused by hardware resource usage.
Health data is often incomplete, inconsistent, and even unstructured. Amazon HealthLake is a HIPAA-eligible service that eases the heavy lifting of organizing, indexing, and structuring patient information — allowing for a complete view of individual patient health or the health of entire populations in a secure, auditable, and compliant manner. We can now employ advanced analytics and ML services like Amazon QuickSight and SageMaker to analyze and process data to make data-driven predictions. Amazon promises to help health organizations trying to streamline everything from early detection of diseases to population health trends to improve patient care and bring down costs.
Incremental Learning with Amazon Kendra
Recognizing the need to address the growth of unstructured data at an enterprise level, Amazon Kendra is getting new incremental learning capabilities to continuously optimize search results based on end-user search patterns and feedback. Incremental learning tunes future search results in a data-driven manner, without users training or deploying machine learning models.
Amazon Lookout for Metrics
Test. Amazon Lookout for Metrics uses ML to detect anomalies or unexpected changes in your metrics. Finding anomalies such as a dip in product sales or a rapid increase in sales leads is challenging. Traditional rule-based methods raise too many false alarms if the range is too narrow and cannot reliably account for factors like time of day, weekday or weekend, seasons, and so on. Amazon Lookout for Metrics connects to 19 different data sources. In addition to Amazon’s services like S3, CloudWatch, RDS, and Redshift, Lookout for Metrics allows users to get data from Salesforce, Marketo, and Amplitude.
AWS Audit Manager
Amazon announced a new service that helps continuously audit your AWS usage to assess risk and compliance with present regulations and industry standards. In addition to featuring frameworks like CIS AWS Foundations Benchmark, the General Data Protection Regulation (GDPR), and the Payment Card Industry Data Security Standard (PCI DSS), business users can fully customize a framework to tailor it to their unique regulatory and compliance requirements.
AWS Security Hub Integrated with AWS Audit Manager: Further simplifying the process of assessing risk, Amazon has integrated the AWS Security Hub into AWS Audit Manager. It gives you a comprehensive view of your security posture across all your AWS accounts.
AWS Panorama Appliance
Amazon announced the ability to develop a computer vision model using Amazon SageMaker, which can then be deployed to a Panorama Appliance that can run the model on video feeds from multiple networks and IP cameras. Third-party device manufacturers can now use the Panorama SDK to build Panorama-enabled devices.
You might also like
How AI and Personalized Marketing are Transforming Retail Sales
How AI/ML, CDP, personalization, and BI are revolutionizing retail, fashion, and beauty. Dive into brand examples from Sephora, ThredUp, and H&M.Read article
19 Cloud Computing Statistics You Need to Know in 2023
By 2025, over 100 zettabytes of data will be stored in the cloud—50% of all global data storage.Read article
5 Ways to Transform Grocery Retail with an AI-Driven Data Strategy
Explore 5 AI-driven data strategies for grocery retail. Learn how to solve challenges like workforce management, pricing, and disconnected CX.Read article
Copilot and the Future of AI-Assisted Coding: Insights from a Software Engineer
GitHub's Copilot is an IDE-integrated tool that streamlines coding by offering real-time text completion suggestions, predicting what developers might type next.Read article
Data Mesh and Event-Driven Architecture: Unleashing Healthcare Data Potential
Harness the power of data mesh and event-driven architecture. Streamline healthcare data management, drive innovation, and unlock lasting growth.Read article
Unlocking Hidden Revenue: Optimizing Data Systems for a Leading Car Retailer
Explore how a leading care retailer teamed up with us to optimize workflows, boost security, and uncover hidden revenue.Read article
Rapid Migration of a Legacy Java Monolith for a Major Retail Brand
Find out how we helped a grocery giant tackle a two-year standstill and the loss of key team members to conquer a critical deadline.Read article
How an HC & Tech Services Provider Optimized Scalability, Agility, and Security
Discover how a human capital and tech services provider unlocked the power of cloud infrastructure to optimize scalability, agility, and security.Read article
Egen Cuts Sifter's Infrastructure Costs by Over 30%
Learn how Sifter, an online grocery platform, reduced infrastructure costs by over 30%, allowing the startup to focus on growth and profitability.Read article