Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/devintdha/pml-schnet
Project Machine Learning WS23/24: SchNet: A continuous-filter convolutional neural network for modeling quantum interactions
https://github.com/devintdha/pml-schnet
Last synced: 14 days ago
JSON representation
Project Machine Learning WS23/24: SchNet: A continuous-filter convolutional neural network for modeling quantum interactions
- Host: GitHub
- URL: https://github.com/devintdha/pml-schnet
- Owner: DevinTDHa
- Created: 2023-11-11T16:10:16.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-03-19T16:29:24.000Z (10 months ago)
- Last Synced: 2025-01-10T23:47:44.559Z (14 days ago)
- Language: Python
- Size: 4.08 MB
- Stars: 3
- Watchers: 4
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# PML-SchNet
Project Machine Learning WS23/24: [SchNet: A continuous-filter convolutional neural
network for modeling quantum interactions](https://arxiv.org/pdf/1706.08566.pdf)This is a repository covering the code and reports for the module "Project Machine Learning" at TU Berlin. We attempted to reproduce the paper and expand on it by introducing the following:
1. Two different confidence measures for the model predictions, using ensemble models and statistical inference
2. Exploration of the hyperparameter space, revealing better suited hyperparameters
3. Explainability measure for model predictions
4. Regularization, namely [RMSNorm](https://arxiv.org/abs/1910.07467)
5. A prototype SchNet Transformer architecture> https://github.com/DevinTDHa/PML-SchNet/assets/33089471/f73b561e-d28c-4848-afc0-f27a2f2f6d39
>
> Figure: Molecular Dynamics Animation for an aspirin molecule of the MD17 dataset. The goal of SchNet is to predict the energy of the molecule (color from black to white) and the forces acting on the atoms (arrows for each atom).## Reports
For this project, we wrote 3 reports for a significant milestone each:
1. [Milestone 1: Data Sets and Prototype](reports/Project_Machine_Learning_Report-1.pdf)
2. [Milestone 2: Model Selection and Evaluation](reports/Project_Machine_Learning_Report-2.pdf)
3. [Milestone 3: The Final Prediction Method and Explainability](reports/Project_Machine_Learning_Report-3.pdf)