Machine Learning

Theory

We produce a lot of data (see: data capitalism)

• Data mining: automatically extract useful knowledge from large datasets
• Machine learning: automatically detect patterns in data and use these to make predictions or decisions
  • Typically, AI $\subset$ ML $\subset$ Deep Learning
• Typically, data mining is more human-in-the-loop and more application specific whereas machine learning is more hands-off and general
• Both similar to statistics but more emphasis on larger datasets, predictions instead of descriptions, and more general models

Healthy skepticism is good though:

“The combination of some data and an aching desire for an answer does not ensure that reasonable answer can be extracted from a given body of data”

• John Tukey

Main topics:

• Exploratory Data Analysis
• Supervised learning
• No Free Lunch Theorem
• Unsupervised Learning
• Optimization
  • Gradient Descent
  • hyper-parameter optimization
• Regularization
• Regression
  • Linear Regression
  • Multi-class classification
• Binary classification
• MLE
• Latent-factor models
• Recommender System
• Neural Networks
• CNNs
• Autoencoders
• Deep learning Semantics
• Transformers
• Generative Models
• Philosophy
  • GOFAI
  • NFAI

Related background:

• Linear Algebra
• Probability
• Calculus
  • Automatic Differentiation