| D1-M |
1. Introduction to AI and ML |
1. Colab setup |
Bruce |
|
2. Steps in ML project |
2. Python refresh |
|
|
|
3. Visualization: t-SNE |
|
| D1-A |
1. Data preprocessing & EDA |
1. Data prep: scaling, normalization, imputation |
Bruce |
|
2. Performance evaluation |
2. Feature selection |
|
|
|
3. Model evaluation: classification, regression |
|
| D2-M |
1. Model training |
1. Model tune up |
Bruce |
|
|
2. Bagging and Boosting |
|
|
|
3. Supply chain example |
|
| D2-A |
1. Supervised learning |
1. Supervised learning |
Ramy |
| D3-M |
1. Supervised learning |
1. Supervised learning |
Nathan, Ramy |
| D3-A |
1. Supervised learning |
1. Supervised learning |
Ramy |
| D4-M |
1. Deep learning |
1. Deep learning |
Nathan, Ramy |
| D4-A |
1. Deep learning |
1. Deep learning |
Ramy |
| D5-M |
1. Deep learning |
1. Deep learning |
Nathan, Ramy |
| D5-A |
Evaluation |
|
Ramy |
| D6-M |
1. Unsupervised learning |
1. PCA |
Ramy |
|
|
2. K-mean and cluster # optimization (elbow method) |
|
| D6-A |
1. Unsupervised learning |
1. Hierarchical clustering |
Ramy |
|
|
2. Soft-clustering (expectation maximization) |
|
| D7-M |
1. Markov decision process |
1. Standard methods for MDP such as PI and VI |
Matin |
|
|
• Formulation: transition probabilities, reward, … |
|
|
|
• Policy evaluation |
|
| D7-A |
1. MDP |
1. PI and VI |
Matin |
|
2. Monte Carlo method |
2. Monte Carlo method |
|
|
|
• Return computation |
|
|
|
• Generalized Policy Iteration |
|
| D8-M |
1. Tabular RL |
1. Tabular RL: Q-learning, SARSA |
Matin |
| D8-A |
1. Deep RL |
1. Deep RL: DQN and others |
Matin |
| D9-M |
1. Policy optimization |
1. Policy optimization: REINFORCE, DPG, DDPG |
Matin |
| D9-A |
1. Model-based RL, MARL |
1. Dyna-Q, MARL |
Matin |
| D10-M |
2. RL for Applications |
1. DQN for optimal maintenance, Sim2Real for robotic control |
Matin |
| D10-A |
Evaluation |
|
Matin |