CoIBA
This repository contains the official code of the paper: "Comprehensive Information Bottleneck for Unveiling Universal Attribution to Interpret Vision Transformers", which is published in CVPR 2025.
Install / Use
npx skills add KU-HJH/CoIBAInstalls into whichever agent you are using.
README
CoIBA
This repository contains the code of Comprehensive Information Bottleneck for Unveiling Universal Attribution to Interpret Vision Transformers, which is presented at a Conference on Computer Vision and Pattern Recognition (CVPR) 2025 as a highlight paper.

Paper: CVPR 2025
Abstract The feature attribution method reveals the contribution of input variables to the decision-making process to provide an attribution map for explanation. Existing methods grounded on the information bottleneck principle compute information in a specific layer to obtain attributions, compressing the features by injecting noise via a parametric damping ratio. However, the attribution obtained in a specific layer neglects evidence of the decision-making process distributed across layers. In this paper, we introduce a comprehensive information bottleneck (CoIBA), which discovers the relevant information in each targeted layer to explain the decision-making process. Our core idea is applying information bottleneck in multiple targeted layers to estimate the comprehensive information by sharing a parametric damping ratio across the layers. Leveraging this shared ratio complements the over-compressed information to discover the omitted clues of the decision by sharing the relevant information across the targeted layers. We suggest the variational approach to fairly reflect the relevant information of each layer by upper bounding layer-wise information. Therefore, CoIBA guarantees that the discarded activation is unnecessary in every targeted layer to make a decision. The extensive experimental results demonstrate the enhancement in faithfulness of the feature attributions provided by CoIBA.
Installation
Our repository is constructed based on Python 3.10
Install the libraries as follows:
pip install -r requirements.txt
Or you can build a Docker image via:
docker build --tag <tag_name> .
Finally, set the simlink for the data path including {train,val} as subdirectories:
cd CoIBA
ln -s /path/to/imagenet ./data
Visualization
bash run.sh
Insertion and Deletion tests
bash run_eval.sh
To conduct the insertion and deletion tests with 1000 iterations, please run the command below:
bash run_eval.sh "--cut-off 1000"
Hyperparameters
- --start_target indicates start layer for attribution (default: 4)
- --end_target indicates end layer for attribution (default: 12)
- --op_name indicates operation names for attribution (default: norm1)
- --optim_steps indicates the number of optimization steps for IBA ()
- --iba_lr indicates learning rate for IBA
- --beta indicates beta value for IBA
- --load_estim indicates whether to load precomputed estimators
- --iba_path indicates path to save/load IBA estimators
Results
The qualitative results of ViT-B

TIPS
Note that the we set op_name following the architecture of the timm library
For example, CoIBA is inserted in blocks.4.norm2 ~ blocks.8.norm2 with the setting: --start_target 4 --end_target 8 --op_name norm2
The rest of the hyperparameters (--optim_steps --iba_lr --beta) set the iteration procedure of our CoIBA.
BibTex
@inproceedings{hong2025comprehensive,
title={Comprehensive Information Bottleneck for Unveiling Universal Attribution to Interpret Vision Transformers},
author={Hong, Jung-Ho and Kim, Ho-Joong and Jeon, Kyu-Sung and Lee, Seong-Whan},
booktitle={Proceedings of the Computer Vision and Pattern Recognition Conference},
pages={25166--25175},
year={2025}
}
Related Skills
mcp
Use the `mcp_perplexity-ask_perplexity_search` tools to answer questions. You should use this instead of the `web_search` tool because it is a lot more accurate.
practical-power-systems-synthesis
This skill enables synthesis in the domain of power-systems (engineering). It represents research-level-level expertise and is designed for production use in research, industry, and educational contexts. Use this skill when you need to perform synthesis operations related to power-systems.
semi-supervised-optogenetics-testing
This skill enables testing in the domain of optogenetics (neuroscience). It represents intermediate-level expertise and is designed for production use in research, industry, and educational contexts. Use this skill when you need to perform testing operations related to optogenetics.
data-mining-interpretation-fundamental
This skill enables interpretation in the domain of data-mining (data-science). It represents fundamental-level expertise and is designed for production use in research, industry, and educational contexts. Use this skill when you need to perform interpretation operations related to data-mining.
