Machine Learning in AWS

Understanding Types of AI (Artificial Intellegence):

Strong (General) AI:

Intellegence of Machine = Intellegence of Human
For eg:

  • A machine can learn like a human on its own and solve problems.
  • Learns like a child, building on its own experiences.
  • We are far away from this achievement (it may take decades or never).

Weak (or Narrow) AI:

For eg.

  • Self driving cars
  • Machine Learning: Learn from previous data.
Challenges of Machine Learning:
  • Huge amount of sample data required.
  • Complexity in creating the ML Models.
  • Availability of skilled personnel.
Machine Learning approaches in AWS:

AWS provides below ways to work in Machine Learning even if you have no knowledge of it.

Use Pre-Trained Models:

  • AWS Comprehend, AWS Rekognition, etc.

Use simple Models: Without using Data Scientists.

  • AWS SageMaker Auto ML

Use Complex Models: Using Data Scientists where you have to build your own ML Model.

  • AWS SageMaker

Pre-Trained Models used in AWS

Below are some of the Pre-Trained models that are available as an API:

  • AWS Comprehend – Analyze text and find the meaning and intent. It can be used to comprehend meaning from emails, S3 documents, social media posts – to analyze whether its a positive or negative reviews etc.. (Compare with Kendra)
  • AWS Kendra – To do intelligent search from scattered content across multiple locations. (Compare with Comprehend). Some use cases –
    Employees can search across internal knowledge centers and wikis to find information efficiently. Healthcare professionals can search across medical literature, research papers, and clinical guidelines to find relevant information quickly.
  • AWS Textract – It extracts any text from any document- PDF, jpeg, png files.
  • AWS Rekognition – Analyze the objects in a image/video.
  • AWS Transcribe – Can convert a audio to text.
  • AWS Polly – Do the reverse of AWS Transcribe where it converts text to audio.
  • AWS Translate – Translate audio/text of one language to audio/text of another language.
  • AWS Personalize – It provides real time recommendation to the user in a app. For eg., a user in a e-commerce app gets buying recommendations based on his searches.
  • AWS Fraud Detector – For detecting online frauds.
  • AWS Forecast – to predict future of something.
  • AWS Lex – for Voice and Text Chat Bots. Powered by the same technology that powers Amazon Alexa