https://github.com/nicolossus/fys9429-project1
https://github.com/nicolossus/fys9429-project1
Last synced: 3 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/nicolossus/fys9429-project1
- Owner: nicolossus
- License: mit
- Created: 2024-01-23T13:51:17.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2024-06-26T00:26:26.000Z (11 months ago)
- Last Synced: 2025-01-14T01:34:43.151Z (4 months ago)
- Language: Jupyter Notebook
- Size: 642 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# FYS9429-Project1
This repository contains programs, material and report for the project in FYS9429: Advanced machine learning and data analysis for the physical sciences
## VAE-based representation learning of neural data
In this project, the objective is to investigate whether variational autoencoders (VAEs) can be used to learn features from neural recordings simulated from the Hodgkin-Huxley model. We use both vanilla and $\beta$-VAEs utilizing simple dense and convolutional neural networks as encoder/decoder architecure. We embarked on this quest in the traditional way -- by testing our simple models on the (binarized) MNIST dataset.
## Repo structure
Here is a brief description of the directories in this repo:
* `src/neurovae`: contains the `neurovae` package, which consists of the VAE models, data generators and loaders and utility functions.
* `experiments`: contains notebooks and scripts with the procedures for (re)producing the results.
* `exploration`: contains notebooks and scripts used for development and exploration
* `tests`: contains a very minimal set of tests for verifying `neurovae` functionality
* `latex`: contains the LaTeX source for building the report, as well as figures generated in the analyses.
* `report`: contains the report rendered to PDF from the LaTeX source.
## Run the code (reproducibility)
### Set up development environment
We have an `environment.yml` file that contains all packages to build and work with the `neurovae` package.
Install the `conda` environment:
$ conda env create --file environment.yml
### Install the project package
`cd` into the root of the repository and install the `neurovae` package in editable mode from source:
$ python3 -m pip install .