AWS – AI/ML Services Explained

As of Sept. 2019 – Looking at the current AWS Console for Machine Learning. This post describes each service in one sentence and one use case on when it would be used.

Amazon SageMaker – Probably the most easily accessible of the ML services. Bring your Jupyter notebook and datasets and train to create models. The structure is LABEL Jobs, LABEL Datasets, LABEL Workforces. There are multiple algorithms that can used.

Amazon Comprehend – This is service is AWS NLP for analyzing text and determine word counts and patterns within language. Used to discover the meaning and relationships in text from customer support incidents, product reviews, social media feeds, news articles, documents, and other sources

AWS DeepLens – This was developed out of Amazon’s LOOK product and use cases are for Object Recognition, Activity Recognition, Face Recognition. It’s well integrated with Rekognition. Only available in US-East-1 (North Virginia)

Amazon Lex – Building a conversational interface that utilizes voice and text? Leveraging their experience with Alexa, AWS provides a scalable solution that I have personally used in building a scalable backend website site interaction with FaceBook Messenger chatbot.

Machine Learning – This one is strange as it’s only supported in US-East/EU regions. “Thank you for your interest in Amazon Machine Learning. AmazonML is no longer available to new customers. Please consider using Amazon SageMaker. Click here to go to the SageMaker home page.

Amazon Polly – This service does the reverse of the audio to text. It does text to speech and building audio files based on text and scripts provided to it.

Rekognition – AWS Rekognition is the image machine learning foundation and provides an opportunity to train a large dataset of images to determine sentiment. It provides face compare and face analysis along with celebrity detection. Utilizing images in S3 buckets, You can also run real-time video analysis on streams coming from Amazon Kinesis Video Streams.

Amazon Transcribe – Provides Real-Time auto transcribing into text. Either provide direct audio input or upload podcast and spoken audio files for transcribing into text.

Amazon Translate – An excellent service to provide text translation services for multiple languages. With over 25 languages, you can specify your source and destination language and rapid build transcripts.

Amazon Personalize – Using the same toolset as, this service provides personalization and recommendation services. Personalize Search, Recommendations, and Notifications.

Amazon Forecast – Acting like glue by joining datasets together and then working to determine the patterns within the different datasets. Used for financial, inventory, and resource planning. Works with time-series data.

Amazon Textract – Basically a sophisticated OCR machine learning system that determines fields and text components to build editable documents with tables/cells. Great for large scale document processing efforts.

Amazon DeepRacer – A newer service currently only available in US-East-1 (N. Virginia) – It integrated with AWS Sagemaker and AWS Robomaker. It utilizes a 1/18 scale autonomous race car and reinforcement learning. Designed to evaluate models with 3D Simulator.