Logbook for July 21

Week 26 - July 21

Thursday 7/1

Paper reviewed on arxiv about revisiting Deep Learning Models for Tabular Data. arXiv:2106.11959v1 (FT-Transformer, ResNet, MLP) architectures can compete with GBDT models. Examples are not detailed in paper but code seems available at https://github.com/yandex-research/rtdl (rtdl (Revisiting Tabular Deep Learning) based on PyTorch)

RL Course by David Silver Exploration and Exploitation (lecture 9)

Week 27 - July 21

Wednesday 7/7

Start of Deep Neural Networks with PyTorch on Coursera by IBM.

Friday 7/9

Deep Neural Networks with PyTorch - week 2 - linear regression

Week 28 - July 21

Monday 7/12

Deep Neural Networks with PyTorch - week 3 - multiple input output linear regression, logistic regression for classification

Deep Neural Networks with PyTorch - week 4 - softmax regression

Thursday 7/15

Deep Neural Networks with PyTorch - week 4 - shallow neural networks

Deep Neural Networks with PyTorch - week 5 - deep neural networks

Friday 7/16

Deep Learning with PyTorch book

img

PyTorch Ecosystem day 2021:

• lightning to remove boilerplate code - to be tested!
• torchstudio to visualize dataset and models
• pytorch-ignite from my colleague Sylvain - high-level library to help training evaluating NN
• pytorchts to play with timeseries
• multitask RL environments and baselines for RL
• rlstructures python library for RL research
• mbrl-lib to write model-based RL algorithms
• pystiche framework for style transfer

Deep Neural Networks with PyTorch - week 6 - convolutional neural networks - even if I don’t need it

Week 30 - July 21

Monday 7/26

Deep Neural Networks with PyTorch - week 7 - fashion mnist

Reading of some papers:

• Deep Reinforcement Learning Approaches for Process Control by Steven Spielberg Pon Kumar - thesis, UBC 2017
• A Deep Learning Architecture for Predictive Control by Steven Spielberg Pon Kumar - ScienceDirect 2018
• Deep Reinforcement Learning for Process Control: A Primer for Beginners by Steven Spielberg et al. 2004
• Reinforcement Learning for Statistical Process Control in Manufacturing - by Zsolt János Viharos 2020