ML

有三种。

  1. Supervised Learning

    Utilizes labeled datasets to train models for prediction or classification.
    Limited by the context of the training data.

  2. Unsupervised Learning

    Processes unlabeled data to discover underlying patterns and structures.
    Includes semi-supervised learning, which uses a mix of labeled and unlabeled data to identify correlations.

  3. Reinforcement Learning

    Involves AI systems learning from interactions with their environment.
    Positive actions are reinforced through numerical rewards to improve decision-making over time.

区别

  • Supervised learning - access labeled examples with given answers.
  • UnSupervised learning - extract underlying structure in data, it’s about data representation.
  • Reinforced learning - The reward give the agent some idea of good/bad in its recent actions.

总结：RL focus on the problem of learning while interacting with an ever-changing world.

Modules

  1. Multi-armed bandits.
  2. Markov Decision Processes
  3. Value Functions & Bellman Equations
  4. Dynamic Programming