autoXRD

A Python package for automatic XRD pattern classification of thin-films, tweaked for small and class-imbalanced datasets.
https://github.com/PV-Lab/autoXRD

Category: Renewable Energy
Sub Category: Photovoltaics and Solar Energy

Last synced: about 21 hours ago
JSON representation

Repository metadata

Automatic XRD classification for thin-film materials using CNNs, Class Activation Maps and Data Augmentation

• Host: GitHub
• URL: https://github.com/PV-Lab/autoXRD
• Owner: PV-Lab
• License: apache-2.0
• Created: 2019-04-23T19:15:53.000Z (about 6 years ago)
• Default Branch: master
• Last Pushed: 2020-03-04T18:03:33.000Z (about 5 years ago)
• Last Synced: 2025-04-10T04:01:14.701Z (18 days ago)
• Language: Python
• Homepage:
• Size: 6.46 MB
• Stars: 49
• Watchers: 9
• Forks: 22
• Open Issues: 2
• Releases: 0

https://github.com/PV-Lab/autoXRD/blob/master/

          
autoXRD
===========
## Description


autoXRD is a python package for automatic XRD pattern classification of thin-films, tweaked for small and class-imbalanced datasets. The main application of the package is high-throughput screening of novel materials.

autoXRD performs physics-informed data augmentation to solve the small data problem, implements a state-of-the-art a-CNN architecture and allows interpretation using Average Class Activation Maps (CAMs), according to the following publications:

"**Oviedo, F., Ren, Z., Sun, S., Settens, C., Liu, Z., Hartono, N. T. P., ... & Buonassisi, T. (2019). Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks. npj Computational Materials, 5(1), 60."  Link: [https://doi.org/10.1038/s41524-019-0196-x](https://doi.org/10.1038/s41524-019-0196-x)**


"**Fast classification of small X-ray diffraction datasets using data augmentation and deep neural networks, (2019), Felipe Oviedo, Zekun Ren, et. al.  Link: [arXiv:1811.08425v](https://arxiv.org/abs/1811.08425v2)**

Accepted to NeurIPS 2018 ML for Molecules and Materials Workshop. Final version published npj Computational Materials 2019


## Installation

To install, just clone the following repository:

`$ git clone https://github.com/PV-Lab/autoXRD.git`

## Usage

Just run `space_group_a_CNN.py` , with the given datasets. Note that this performs classification for patterns into 7 space-groups. Dimensionality data is not included in the code, please contact authors if interested.
The package contains the following module and scripts:

| Module | Description |
| ------------- | ------------------------------ |
| `space_group_a_CNN.py`      | Script for XRD space-group classification with a-CNN      |
| `autoXRD`      | Module dedicated to XRD pattern preprocessing and data augmentation       |
| `autoXRD_vis`   | Visualizer module for class activation maps (CAMs)     |
| `Demo / XRD_dimensionality_demo.ipynb` | Notebook containing a demo for physics-informed data augmentation. This is a version with a modified CNN and no CAM to speed up the computation


## Authors
Felipe Oviedo and "Danny" Zekun Ren


||                    |
| ------------- | ------------------------------ |
| **AUTHORS**      | Felipe Oviedo and "Danny" Ren Zekun     | 
| **VERSION**      | 1.0 / May, 2019     | 
| **EMAIL OF REPO OWNER**      | [email protected]  | 
||                    |

## Attribution

This work is under an Apache 2.0 License and data policies of Nature Partner Journal Computational Materials. Please, acknowledge use of this work with the apropiate citation.

## Citation

    @article{oviedo2019fast, 
    title={Fast and interpretable classification of small X-ray diffraction datasets using data augmentation and deep neural networks},
    author={Oviedo, Felipe and Ren, Zekun and Sun, Shijing and Settens, Charles and Liu, Zhe and Hartono, Noor Titan Putri and Ramasamy, Savitha and DeCost, Brian L and Tian, Siyu IP and Romano, Giuseppe and others},
    journal={npj Computational Materials},
    volume={5},
    number={1},
    pages={60},
    year={2019},
    publisher={Nature Publishing Group}}

        

---

Owner metadata

• Name: Accelerated Materials Laboratory for Sustainability
• Login: PV-Lab
• Email:
• Kind: organization
• Description:
• Website: pv.mit.edu
• Location: United States of America
• Twitter:
• Company:
• Icon url: https://avatars.githubusercontent.com/u/13911947?v=4
• Repositories: 13
• Last ynced at: 2023-03-03T18:41:38.087Z
• Profile URL: https://github.com/PV-Lab

---

GitHub Events

Total

• Pull request event: 1
• Fork event: 1

Last Year

• Pull request event: 1
• Fork event: 1

---

Committers metadata

Last synced: 7 days ago

Name	Email	Commits
FELIPE OVIEDO	4****p	27
dannyzekunren	d****n@g****m	3

Committer domains:

---

Issue and Pull Request metadata

Last synced: 2 days ago

More stats: https://issues.ecosyste.ms/repositories/lookup?url=https://github.com/PV-Lab/autoXRD

---

---

Score: 4.624972813284271