Official repository for the CVPR 2026 Findings paper:
Learning to Reason: Targeted Knowledge Discovery and Fuzzy Logic Update for Robust Image Recognition
Thank you for your interest in our work.
The source code and pretrained models are currently undergoing internal organizational review and approval procedures. The repository has been created to provide a permanent reference corresponding to the link included in the published paper.
The implementation, training scripts, evaluation code, pretrained checkpoints, and documentation will be released as soon as the approval process is completed.
We appreciate your patience and understanding.
- Training and evaluation code
- Pretrained model checkpoints
- Dataset preparation scripts
- Configuration files
- Reproducibility instructions
- Documentation and examples
Expected code release: July 2026
If you find our work useful, please cite:
@inproceedings{srinivas2026learning,
title={Learning to Reason: Targeted Knowledge Discovery and Fuzzy Logic Update for Robust Image Recognition},
author={Srinivas, Gurucharan and Niemeijer, Joshua and K{\"o}ster, Frank},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={7122--7131},
year={2026}
}For questions regarding the paper, please open an issue in this repository or contact the authors.
Note: The repository link was included in the published CVPR 2026 paper. The code release is temporarily delayed due to internal approval requirements and will be made publicly available once these procedures have been completed.