Skip to content

CompPhysics/AdvancedMachineLearning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 

Repository files navigation

Advanced Machine Learning

This repository contains information about the course on Advanced Data Analysis and Machine Learning, spanning from weekly plans to lecture material and various reading assignments. The emphasis is on deep learning algorithms, starting with the mathematics of neural networks (NNs), moving on to convolutional NNs (CNNs) and recurrent NNs (RNNs), autoencoders, transformers, graph neural networks and other dimensionality reduction methods to finally discuss generative methods. These will include Boltzmann machines, variational autoencoders, generalized adversarial networks, diffusion methods and other.

alt text

Practicalities

  • Lectures Thursdays 1215pm-2pm, room FØ434, Department of Physics
  • Lab and exercise sessions Thursdays 215pm-4pm, , room FØ434, Department of Physics
  • We plan to work on two projects which will define the content of the course, the format can be agreed upon by the participants
  • No exam, only two projects. Each projects counts 1/2 of the final grade. Alternatively one long project.
  • All info at the GitHub address URL:"https://github.com/CompPhysics/AdvancedMachineLearning"

Deep learning methods covered (tentative plan)

Deep learning, classics

  • Feed forward neural networks and its mathematics (NNs)
  • Convolutional neural networks (CNNs)
  • Recurrent neural networks (RNNs)
  • Graph neural networks
  • Transformers
  • Autoencoders and principal component analysis

Deep learning, generative methods

  • Basics of generative models
  • Boltzmann machines and energy based methods
  • Diffusion models
  • Variational autoencoders (VAEe)
  • Generative Adversarial Networks (GANs)
  • Autoregressive methods (tentative)

Physical Sciences (often just called Physics informed) informed machine learning

  • Basic set up of PINNs with discussion of projects

FYS5429 zoom link to be announced when semester starts

All teaching material is available from this GitHub link.

The course can also be used as a self-study course and besides the lectures, many of you may wish to independently work on your own projects related to for example your thesis or research. In general, in addition to the lectures, we have often followed five main paths:

  • Projects (two in total) and exercises that follow the lectures

  • The coding path. This leads often to a single project only where one focuses on coding for example CNNs or RNNs or parts of LLMs from scratch.

  • The Physics Informed neural network path (PINNs). Here we define some basic PDEs which are solved by using PINNs. We start normally with studies of selected differential equations using NNs, and/or RNNs, and/or GNNs or Autoencoders before moving over to PINNs.

  • The own data path. Some of you may have data you wish to analyze with different deep learning methods

  • The Bayesian ML path is not covered by the present lecture material. It is normally based on independent work.

January 20-24: Presentation of couse, review of neural networks and deep Learning and discussion of possible projects

January 27-31

February 3-7

February 10-14

February 17-21

February 24-28

March 3-7

March 10-14

March 17-21: Autoencoders

March 24-28: Generative models

March 31-April 4: Deep generative models, Boltzmann machines

April 7-11: Deep generative models

  • Implementation of Boltzmann machines using TensorFlow and Pytorch
  • Energy-based models and Langevin sampling
  • Generative Adversarial Networks (GANs)
  • Reading recommendation: Goodfellow et al chapters 18.1-18.2, 20.1-20-7; To create Boltzmann machine using Keras, see Babcock and Bali chapter 4
  • See also Foster, chapter 7 on energy-based models

April 14-18: Public holiday, no lectures

April 21-25: Deep generative models

April 28 - May 2: May 1 is a public holiday, no lectures:

May 5-9: Deep generative models

May 12-16: Deep generative models

May 19-23: Only and discussion of projects

Recommended textbooks:

o Goodfellow, Bengio and Courville, Deep Learning at https://www.deeplearningbook.org/

o Sebastian Raschka, Yuxi Lie, and Vahid Mirjalili, Machine Learning with PyTorch and Scikit-Learn at https://www.packtpub.com/product/machine-learning-with-pytorch-and-scikit-learn/9781801819312, see also https://sebastianraschka.com/blog/2022/ml-pytorch-book.html

o David Foster, Generative Deep Learning, https://www.oreilly.com/library/view/generative-deep-learning/9781098134174/

o Babcock and Gavras, Generative AI with Python and TensorFlow, https://github.com/PacktPublishing/Hands-On-Generative-AI-with-Python-and-TensorFlow-2

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published