This artificial intelligence wiki is a beginner’s guide to machine learning and deep learning. The goal is to give readers an intuition for how algorithms work, with calculations and code you can follow.

Getting Started

  1. Neural Networks and Deep Learning. Follow a crop-yield prediction through two hidden layers and see how a training step changes the answer.
  2. Datasets and Machine Learning. Choose examples that fit your problem, then learn why training, validation and test sets have different jobs.
  3. LSTMs and Recurrent Neural Networks. Trace a character prediction through two steps and see how memory carries information through a sequence.

For the broader vocabulary, read our comparison of artificial intelligence, machine learning and deep learning.

Reinforcement Learning and Simulation

Reinforcement learning trains an agent by trial and error: it acts, sees the result, and adjusts toward higher reward. Simulations are where that trial and error is cheap. A model of a warehouse, factory floor or supply chain lets an agent try thousands of policies before any of them touches real operations.

  1. Deep Reinforcement Learning. How agents, states, actions and rewards fit together, and how neural networks let RL handle large problems.
  2. Simulation, AI, Optimization and Complexity. Why simulations are useful, how optimization works inside them, and the tools used to build them for business operations.
  3. Reinforcement Learning Definitions. The core vocabulary: agent, state, action, reward, policy, discount factor and Q-value.
  4. Genetic and Evolutionary Algorithms. Another way to search for good solutions when the problem is too complex to solve directly.

The line between mathematics and philosophy is blurry when we talk about artificial intelligence, because with AI, we ask the mineral called silicon to perceive and to think, actions once thought exclusive to meat, and now possible with computation.