• Ref: https://www.udemy.com/course/aws-ai-practitioner-certified/learn/lecture/44886569
• Contents:

Artificial Intelligence (AI)

• 🔧 Broad field for the development of intelligent systems capable of performing tasks that typically require human intelligence:
  • Perception, Reasoning, Learning, Problem solving, Decision-making…
  • Umbrella-term for various techniques
• Diagram

Machine Learning (ML)

• 🔧 Subset of AI whereby lots of data (training data) is fed into an algorithm, which outputs a statistical model and rules
  • ❗No explicit programming of rules → rules “learned” from training data
  • ML systems can make predictions based on patterns of their training data
• Examples: Regression and Classification algorithms
• ‼️ AI ≠ ML!!
  • Example: MYCIN Expert System from 1970s is AI with hundreds of predefined rules to determine culprit bacteria
  • A set of if-else statements could be considered barebones AI

When is Machine Learning NOT appropriate?

Ref: https://www.udemy.com/course/aws-ai-practitioner-certified/learn/lecture/45796429

• Deterministic problems (where the solution can be computed) are better solved with computer code adapted to the problem than with ML
  • ML algorithms can offer an approximation of the result, but they're always worse and less optimal than a direct coding solution →
• Example: Probability of drawing a blue card in a deck of cards

Deep Learning (DL)

• 🔧 Subset of ML which involves applying several “learning layers” to the training data, which form a neural network
  • DL can process and detect more complex patterns than traditional ML
  • Examples:
    • Computer Vision → images
    • Natural Language Processing (NLP) → text
  • Diagram of a neural network
• Requires even more input data than traditional ML