No. 04 · 7 parts · read in order
Machine Learning 2
learning what to do, not just what to predict
- 01Machine learning for trading, part two, in ten ideas11 min
- 02Reinforcement learning: learning from reward, not from labels15 min
- 03MDPs: a state, an action, a reward and the value of what follows13 min
- 04Q-learning: one update rule, repeated15 min
- 05Exploration: when to try something worse on purpose16 min
- 06Dyna: learn a model of the world, then practise in it15 min
- 07States and rewards: what you tell the learner decides what it learns22 min