What's changed: Deepened MLS-C01 Chapter 6 (component roles/spot training/Experiments/Debugger/Pipelines, AI service Custom features & decision/Textract vs Rekognition/Forecast vs DeepAR, Ground Truth workforces/auto-labeling/build-buy spectrum + tables, scenarios, FAQ, traps; ja figures)
6.2High-Level AI Services
Use without building—understand Rekognition, Comprehend, Transcribe/Polly, Translate, Textract, and Personalize (the time-series Forecast service is closed to new use; its successor is SageMaker Canvas, etc.). Implement quickly via no-training APIs.
For common tasks, high-level AI services—just call an API, no training—are the fastest. Before building a custom model, check if these suffice.
6.2.1AI services by use case
- Rekognition: image/video analysis (objects, faces, moderation).
- Comprehend / Translate / Textract: NLP (sentiment/entities) / translation / extracting text & data from documents.
- Transcribe / Polly: speech→text / text→speech.
- Personalize: recommendations with no training (the time-series Forecast service is closed to new use; use SageMaker Canvas, etc.).
Common on MLS-C01: images/video = Rekognition, sentiment/entities = Comprehend, translation = Translate, docs → data = Textract, speech → text = Transcribe, recommendations = Personalize, managed time-series forecasting = Forecast. For common tasks, first see if a no-training AI service fits (build custom on SageMaker only if not).
Continue reading — free sign-up
You're reading the free preview. Sign up free to read this section in full, plus every chapter (including 4+) and all questions.