Computer Engineering is coming soon
We're still writing and reviewing the notes for every subject, and we'll open it when every subject is ready. Meanwhile you can practise aptitude, ask Aria about any subject, or see the branches that are ready.
Computer Engineering
Reinforcement Learning
Bandits and MDPs, dynamic programming, Monte Carlo and TD learning, policy gradients, deep RL.
Notes for Reinforcement Learning are being written and reviewed. Until they're up, Aria teaches every topic — ask for any part of the syllabus, with worked numericals.
Coming next in Reinforcement Learning
- Reinforcement learning: elements and scope· notes coming
- Multi-armed bandits· notes coming
- Action selection: optimistic values, UCB and gradient bandits· notes coming
- Finite Markov decision processes· notes coming
- Policies, value functions and optimality· notes coming
- Policy evaluation and policy iteration· notes coming
- Value iteration and generalised policy iteration· notes coming
- Monte Carlo prediction and control· notes coming
- Off-policy learning and importance sampling· notes coming
- Temporal-difference prediction· notes coming
- Sarsa and Q-learning· notes coming
- Expected Sarsa and double learning· notes coming
- n-step bootstrapping· notes coming
- Planning with models, prioritised sweeping and MCTS· notes coming
- Value-function approximation· notes coming
- Semi-gradient methods and linear function approximation· notes coming
- On-policy control with approximation· notes coming
- Off-policy approximation and the deadly triad· notes coming
- Eligibility traces and TD(λ)· notes coming
- Policy gradient theorem· notes coming
- REINFORCE and baselines· notes coming
- Actor–critic methods· notes coming
- Deep Q-networks· notes coming
- DQN improvements: double DQN, prioritised replay and dueling· notes coming
- Continuous actions: DDPG and PPO· notes coming
- Trends: multi-agent RL, offline RL and RLHF· notes coming
- PAC bandit algorithms and median elimination· notes coming
- Hierarchical reinforcement learning· notes coming
- Partially observable MDPs· notes coming