Data

Tools

Indexes

Applications

Experiments

Open Source Science

zut

[AAAI 2025 Oral] Fair Training with Zero Inputs

https://github.com/asd123pwj/zut

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (6.9%) to scientific vocabulary
Last synced: 10 months ago · JSON representation ·

Repository

[AAAI 2025 Oral] Fair Training with Zero Inputs

Basic Info
  • Host: GitHub
  • Owner: asd123pwj
  • License: apache-2.0
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 5.36 MB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created over 1 year ago · Last pushed about 1 year ago
Metadata Files
Readme License Citation

README.md

Fair Training with Zero Inputs [PDF] [Oral Presentation/Poster] [中文介绍]

This repository is the official implementation for semantic segmentation in our AAAI 2025 oral presentation paper Fair Training with Zero Inputs.

Results

Results on ADE20K

Note: This repository is a clean reimplementation based on MMSegmentation after paper acceptance, enabling direct comparison with the original codebase to inspect ZUT's modifications. Sample training results on VAN-b0 are shown in the table below.

| Method | Reported mAcc (%) | Reported mIoU (%) | Sample mAcc (%) | Sample mIoU (%) | Logs/Weight | | ------ | ----------------- | ----------------- | --------------- | --------------- | ---------------------------------------------------------------------------------------------------- | | VAN-b0 | 52.56 | 37.87 | 53.33 | 38.10 | Google Drive |

Install

  1. Following the install steps of MMSegmentation 1.2.1.
  2. Download ADE20K dataset, and modify data_root in ./configs/_base_/datasets/ade20k.py to match dataset path.
  3. Download pretrained weight for VAN-{b0, b1, b2, b3}, and put them like ./pretrained/van_b0.pth.
    • MMSeg will auto download pretrained weights for ResNet-50, Poolformer-s12, and ConvNeXt-tiny.

Training

```python

ZUT

python tools/train.py configs/0ZUT/vanb0.py python tools/train.py configs/0ZUT/vanb1.py python tools/train.py configs/0ZUT/vanb2.py python tools/train.py configs/0ZUT/vanb3.py python tools/train.py configs/0ZUT/r50.py python tools/train.py configs/0ZUT/poolformers12.py python tools/train.py configs/0ZUT/convnext_tiny.py

Baseline

python tools/train.py configs/0ZUTbaseline/vanb0.py python tools/train.py configs/0ZUTbaseline/vanb1.py python tools/train.py configs/0ZUTbaseline/vanb2.py python tools/train.py configs/0ZUTbaseline/vanb3.py python tools/train.py configs/0ZUTbaseline/r50.py python tools/train.py configs/0ZUTbaseline/poolformers12.py python tools/train.py configs/0ZUTbaseline/convnexttiny.py ```

Citation

bibtex @inproceedings{Fairness_ZUT_WJPan, author = {Pan, Wenjie and Zhu, Jianqing and Zeng, Huanqiang}, title = {Fair Training with Zero Inputs}, booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence}, volume = {39}, pages = {6317-6325}, address= {Pennsylvania, USA}, year = {2025}, type = {Conference Proceedings} }

License

This project is released under the Apache 2.0 license.

Owner

  • Name: MWHLS
  • Login: asd123pwj
  • Kind: user

Citation (CITATION.cff) 

cff-version: 1.0.0
message: "If you use this software, please cite it as below."
authors:
  - name: "Wenjie Pan"
title: "Source Code of Fair Training with Zero Inputs"
date-released: 2025-03-11
url: "https://github.com/asd123pwj/ZUT"
license: Apache-2.0

GitHub Events

Total
  • Issues event: 1
  • Watch event: 2
  • Issue comment event: 2
  • Push event: 9
  • Create event: 2
Last Year
  • Issues event: 1
  • Watch event: 2
  • Issue comment event: 2
  • Push event: 9
  • Create event: 2