• Contents:

Amazon Mechanical Turk (MTurk)

Ref: https://www.udemy.com/course/aws-ai-practitioner-certified/learn/lecture/44887679

• 🔧 Crowdsourcing marketplace to perform simple human tasks
  • Distributed virtual workforce
• 💡 Origin: Chess Mechanical Turk → https://en.wikipedia.org/wiki/Mechanical_Turk
  • Image
• Example:
  • You have a dataset of 10 million unlabeled images
  • You distribute the task on Mechanical Turk and humans will tag/label those images
  • You set the reward per image (e.g. $0.10 per image)
• Use cases: image classification, data collection, business processing
• Integrates with Amazon A2I, SageMaker Ground Truth…

Amazon Augmented AI (A2I)

Ref: https://www.udemy.com/course/aws-ai-practitioner-certified/learn/lecture/44887687

• 🔧 Human oversight of ML predictions in production
  • Humans can be your own employees, over 500 000 contractors from AWS, or Amazon Mechanical Turk
  • Some vendors are pre-screened for confidentiality requirements
• ML model can be built on AWS (SageMaker, Rekognition…) or elsewhere

Amazon Forecast

Ref: https://learn.cantrill.io/courses/1820301/lectures/42194099

• ‼️ NOTE!! Amazon Forecast has been DISCONTINUED for new customers! Instead, AWS suggests using SageMaker Canvas:
  • Reference: https://aws.amazon.com/blogs/machine-learning/transition-your-amazon-forecast-usage-to-amazon-sagemaker-canvas/
  • Feel free to skip this service, it shouldn't appear in exams any more
• 🔧 Time-series data forecasting service
  • ‼️ NOT necessarily for weather forecasting! xD
  • Examples: retail demand, supply chain, staffing, energy, server capacity, web traffic…
• Imports from customers:
  1. Historical data
     • e.g. track how popular a sold item is throughout a time period
  2. Related data (extra contextual information)
     • e.g. promotions running in that time period