"Machine Learning will empower and improve every business, every government organization, every philanthropy - basically there's no institution in the world that cannot be improved with Machine Learning"
- Jeff Bezos, Founder, Amazon
If you are brand new to the topic of machine learning, rather than diving into the specific AWS machine learning services you may want to explore the topic of what is deep learning first.
According to the AWS website primary machine learning, there are 32 machine learning services (as of 2022), however, this number is a miscount. By digging a little further you can find a total of 34 AWS machine learning services. Within this broad umbrella of machine learning, there are actually five major categories:
AWS Supported Machine Learning Frameworks
Among the AWS machine learning services offered, the machine learning frameworks are the most rudimentary. AWS provides the hardware and optimizes performance for the following write-your-own-algorithm frameworks:
- Amazon SageMaker - Amazon's very own machine learning framework.
- PyTorch on AWS - a machine learning framework managed by Facebook's AI Research (FAIR) Lab.
- Apache MXNet on AWS - a machine learning framework from the Apache Software Foundation.
- TensorFlow on AWS - a machine learning framework managed by the Google Brain team.
In addition to these AWS also does support the following via their available deep-learning Amazon Machine Images (AMIs):
- Chainer
- Theano
- Keras
- Gluon
AWS Deep Learning Algorithms
The most robust offering —and by far the most interesting— are the AWS deep learning algorithm which spans a large cross-section of data and brings a tremendous amount of value with no need for any kind of training. These deep learning algorithms include:
- Amazon Comprehend - discover insights and relationships in text
- Amazon Comprehend Medical - a medical-specific spinoff of Comprehend
- Amazon DevOps Guru - ML-powered cloud operations service to improve application availability
- Amazon Forecast - increase forecast accuracy using machine learning
- Amazon Rekognition - machine learning computer vision to analyze image and video
- Amazon Personalize - create real-time personalized user experiences faster at scale
- Amazon CodeGuru - automate code reviews and optimize application performance with ML-powered recommendations
- Amazon Fraud Detector - a real-time fraud detection service
- Amazon Kendra - an intelligent search service powered by machine learning
- Amazon Textract - extract printed text, handwriting, and data from any document
- Amazon Translate - translate written text from one language to another
- Amazon Transcribe - convert spoken language into written text
- Amazon Lookout for Equipment - detect abnormal behavior by analyzing sensor data
- Amazon Lookout for Metrics - detect anomalies in metrics
- Amazon Lookout for Vision - spot product defects using computer vision to automate quality inspection
AWS Machine Learning Add-on Services
Not exactly stand-alone products (but branded that way), the machine learning add-ons category includes offerings that generally make some of the drudgery involved in machine learning less painful or somehow improve performance. These services include:
- Amazon Augmented AI - easily implement a human review of ML predictions
- Amazon Elastic Inference - lower machine learning inference costs by up to 75%
- Amazon SageMaker Ground Truth - create datasets for training machine learning models
- Amazon SageMaker Neo - run ML models anywhere with up to 25x better performance
- AWS Deep Learning AMIs - Amazon Machine Images (AMI) for different ML frameworks
- AWS Deep Learning Containers - read-to-go containers for different ML frameworks
- Amazon HealthLake - Securely store, transform, query, and analyze health data in minutes
AWS Machine Learning Powered Hardware
- Amazon Inferentia - high-performance machine learning inference chip, custom designed by AWS
- AWS DeepLens - a deep learning-enabled video camera
- Amazon Monitron - an end-to-end system for equipment monitoring (including a physical sensor)
- AWS Panorama - hardware-enabled computer vision at the edge