CompSciProgram

This repository contains exercises and projects on computational science and AI for the CompSci program

First project:

Machine learning with linear and non-linear regression, logistic regression and support vector machines as well as Bayesian linear regression. This involves linear algebra (matrix inversion, determinants, eigenvalues, SVD and more from FYS4150), convex optimization problem (gradient descent, steepest descent, stochastic gradient descent, iterative solvers) and several central (deterministic) ML methods. Calculation-oriented statistics with Bayes' theorem and MCMC sampling can also be included. Bayesian linear regression can be omitted.

Workload: 6 ECTS.

Datasets you study can be adapted to your research field, whether it is astro, physics, chemistry, bioscience, geoscience or mathematics. Planned finished december 2021

Second project:

Deep learning: standard neural networks, convolution and neural networks (CNN), recursive neural networks, Boltzmann machines, various autoencoders and possibly general adversial networks. Reduction of dimensionality in scientific problems. Possible topic to work with: solution of ordinary and partial differential equations. Here we can take this from a deep learning perspective and a traditional final difference form taught in FYS4150. But we can also focus on classification problems. Datasets can again be adapted to the field.

Workload: 7 ECTS. Planned finished end February 2022

Third project:

Three possible alternative paths that combine elements from both courses. -Unsupervised learning: PCA, other dimensionality reduction methods and clustering, k-means or similar methods. -Bayesian machine learning: brings in MCMC, statistics and deep learning. -Quantum machine learning: Boltzmann machines, classical and quantum machines. MCMC simulations, gradient methods. -Or simulate data and themes related to own research or other user defined topics.

Workload: 7 ECTS. Planned finished end April 2022

In total 20 ECTS.

Lectures

October 18

October 25

November 1

November 15

November 22

December 9

December 13

ssjha-iist / compsciprogram Goto Github PK

compsciprogram's Introduction

CompSciProgram

First project:

Workload: 6 ECTS.

Second project:

Workload: 7 ECTS. Planned finished end February 2022

Third project:

Workload: 7 ECTS. Planned finished end April 2022

Lectures

October 18

October 25

November 1

November 15

November 22

December 9

December 13

January 17

January 24

January 31

February 7

February 14 and 21

February 28

March 7

April 4, Bayesian Statistics lecture series by Anders Kvellestad

April 19, Bayesian Statistics lecture series by Anders Kvellestad

April 25, Bayesian Statistics lecture series by Anders Kvellestad

May 2, Bayesian Statistics lecture series by Anders Kvellestad

May 9, Bayesian Statistics lecture series by Anders Kvellestad

Recommend Projects

Recommend Topics

Recommend Org

Jobs