Logbook for March 21

logbook
Published

March 1, 2021

Week 9 - Mar 21

Monday 3/1

MIT 6S191 Deep Generative Modeling (lecture 4) - vaes and gans.

MIT 6S191 De-biasing Facial Recognition Systems (lab 2): CNN, VAE, DB-VAE

Tuesday 3/2

College de France Approximations non linéaires et réseaux de neurones (lecture 4)

RL Course by David Silver lecture 1 - intro (22’/88’)

Future of Manufacturing@MIT - interesting landscape about Manufacturing and AI

Wednesday 3/3

Interpretable Machine Learning by Christoph Molnar. LIME reading to understand context of local surrogate models. SHAP chapter using Janus data.

Deep Reinforcement Learning by Thomas Simonini (Chapter 3 v1) on DQN with temporal limitation using LSTM, and experience replay. (replay buffer)

Thursday 3/4

Interpretable Machine Learning by Christoph Molnar. PDP chapter using Janus data.

Friday 3/5

RL - Sutton book (p220-223) - full vs sample backups, trajectory sampling, heuristic search

RL - Sutton book (p223+) - start of Approximate Solution Methods, why to use NN.

Week 10 - Mar 21

Monday 3/8

MIT 6S191 Deep Reinforcement Learning. Q-learning vs Policy Gradient.

Tuesday 3/9

College de France Ondelettes et échantillonnage (lecture 5)

RL Course by David Silver Introduction to Reinforcement Learning (lecture 1)

Installation of clustergit to detect local (=uncommited or unpushed) changes in repo

Wednesday 3/10

Deep Reinforcement Learning by Thomas Simonini (Chapter 4 v1) on four strategies to improve DQN (fixed Q-targets, double DQN, dueling DQN (DDQN), Prioritized Experience Replay (PER))

t-SNE using Janus data.

RL Course by David Silver Markov Decision Processes (lecture 2)

Friday 3/12

RL - Sutton book (p287-352) - Applications and case studies, end of the book

RL Course by David Silver Planning by Dynamic Programming (lecture 3)

Week 11 - Mar 21

Monday 3/15

MIT 6S191 Limitations and New Frontiers.

MIT 6S191 Pixels-to-Control Learning (lab 3): Cartpole and Pong

RL Course by David Silver Model-Free Prediction (lecture 4)

Tuesday 3/16

College de France Multi-résolutions (lecture 6)

Wednesday 3/17

Deep Reinforcement Learning by Thomas Simonini (Chapter 5 v1) - Policy Gradient

Thursday 3/18

Deep Reinforcement Learning by Thomas Simonini (Chapter 5 v1) - Policy Gradient notebooks

RL Course by David Silver Model-Free Control (lecture 5)

Friday 3/19

Deep Reinforcement Learning by Thomas Simonini (Chapter 6 v1) - Advantage Actor Critic (A2C) and Asynchronous Advantage Actor Critic (A3C)

College de France - l’apprentissage profond par Yann Lecunn 2016 - Pourquoi l’apprentissage profond ?

Week 12 - Mar 21

Monday 3/22

MIT 6S191 Evidential Deep Learning and Uncertainty (lecture 7).

Deep Reinforcement Learning by Thomas Simonini (v1 Part 5) - Advantage Actor Critic (A2C) - implementation and video

Tuesday 3/23

College de France Bases orthonormales d’ondelettes (lecture 7)

Wednesday 3/24

Deep Reinforcement Learning by Thomas Simonini (Chapter 7 v1) - Proximal Policy Optimization PPO

Stable baselines 3 - init and 1st tutorial

Thursday 3/25

setup headless raspberry pi to bridge wifi (tethering from phone) to ethernet (to wifi-router)

Stable baselines 3 - finalize init and go through documentation

Create a patch for a github project (by forking and pulling request)

Friday 3/26

Stable baselines 3 - Documentation > Examples

Rename my branches named master to main

Week 13 - Mar 21

Monday 3/29

MIT 6S191 Bias and Fairness (lecture 8).

Wednesday 3/31

College de France Parcimonie et compression d’images (lecture 8)