T-SKM-Net: Trainable Neural Network Framework for Linear Constraint Satisfaction via Sampling Kaczmarz-Motzkin Method

This is the official repository for T-SKM-Net, a trainable neural network framework for linear constraint satisfaction via the Sampling Kaczmarz-Motzkin method.

• Paper (arXiv): https://arxiv.org/abs/2512.10461
• PDF in this repository: paper.pdf

Status: Accepted at AAAI-26.
Code availability: We will release the code once it is ready.

---

Citation

If you find this work useful, please cite:

@misc{zhuTSKMNetTrainableNeural2025,
  title         = {T-SKM-Net: Trainable Neural Network Framework for Linear Constraint Satisfaction via Sampling Kaczmarz-Motzkin Method},
  author        = {Zhu, Haoyu and Zhang, Yao and Ren, Jiashen and Hou, Qingchun},
  year          = {2025},
  eprint        = {2512.10461},
  archiveprefix = {arXiv},
  doi           = {10.48550/arXiv.2512.10461}
}