SeongokRyu / ACE-Team-DLstudy

Study materials about "Deep Learning for Molecular Applications".

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

ACE-Team-DLstudy

Study materials about "Deep Learning for Molecular Applications".

Chapter 1. Brief summary of statistical modeling

  • Statistical modeling and inference
  • Likelihood and Maximum likelihood estimation
  • Bayes theorem and Bayesian inference
  • Beyond Likelihoodism

Chapter 2. Taxonomy of statistical learning

  • Supervised learning
  • Unsupervised learning
  • Reinforcement learning
  • Beyond supervised learning : semi-supervised learning, transfer learning, meta-learning

Chapter 3. Neural network

  • Multi-layer perceptron
  • Universal approximation theorem
  • Inductive biases
  • Back propagation
  • Vanishing gradient

Chapter 4. Convolutional neural network

Chapter 5. Recurrent neural network

Chapter 6. Current advanced techniques in building neural networks

  • Attention
  • Gate (Highway network)
  • Memory

Chapter 7. Generative models

  • The Goal of generative models
  • Variational autoencoder (VAE)
  • Generative adversarial network (GAN)
  • Previous problems on 1st generation generative models and solutions

Chapter 8. Graph neural network

  • Overview on graph neural network
  • Relational inference
  • Message passing neural network
  • Graph convolutional network
  • Augmented graph convolutional network
  • Graph generative models

Chapter 9. Regularization

  • Interpretation of regularization
  • L1/L2 regularization
  • adversarial training
  • ...

Chapter 10. Bayesian deep learning

  • Uncertainty in deep learning
  • Variational inference
  • MC-dropout network
  • Uncertainty quantification
  • Uncertainty-aware deep learning

Chapter 11. Towards generalization of neural network

  • Detecting out-of-distribution samples
  • Adversarial examples
  • Domain-adaptation
  • Few-shot learning

Chapter 12. Previous researches of using deep neural networks on molecular applications

  • Applications on quantum chemistry problems
  • Prediction of molecular properties
  • De novo molecular design with generative models

Chapter 13. Future directions

About

Study materials about "Deep Learning for Molecular Applications".


Languages

Language:Jupyter Notebook 92.4%Language:Python 7.6%