Simulating Training Dynamics to Reconstruct Training Data from Deep Neural Networks

Hanling Tian, Yuhang Liu, M. He, Zhengbao He, Zhehao Huang +2 more
2/12/2026

No official code found. Best implementation: BlueBlood6/SimuDy (10 stars, last commit 2025-03-13).

Semantic Scholar

Code Implementations(1)

BlueBlood6/SimuDyCommunity55%

[ICLR 2025] "Simulating Training Dynamics to Reconstruct Training Data from Deep Neural Networks" by Hanling Tian, Yuhang Liu, Mingzhen He, Zhengbao He, Zhehao Huang, Ruikai Yang and Xiaolin Huang

100Shell, PythonMar 1, 20251 years ago

Similar Papers

Brain Netflix: Scaling Data to Reconstruct Videos from Brain Signals

98% match2026

Learning to Search from Demonstration Sequences

97% match2026

Rethinking CNN's Generalization to Backdoor Attack from Frequency Domain

94% match1 repo72026

Variational Counterfactual Intervention Planning to Achieve Target Outcomes

93% match2026

Resolution Attack: Exploiting Image Compression to Deceive Deep Neural Networks

93% match1 repo102026

CircuitNet: A Generic Neural Network to Realize Universal Circuit Motif Modeling

93% match2026

Cite this paper

@article{tian2026simulating,
  title  = {Simulating Training Dynamics to Reconstruct Training Data from Deep Neural Networks},
  author = {Hanling Tian and Yuhang Liu and M. He and Zhengbao He and Zhehao Huang and Ruikai Yang and Xiaolin Huang},
  year   = {2026},
  url    = {https://api.semanticscholar.org/CorpusID:4046a8c652d4fb90bfff4c436aae738ddac0c499},
  journal = {ICLR 2025 2025}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/4046a8c652d4fb90bfff4c436aae738ddac0c499.svg)](https://codeofpaper.com/paper/4046a8c652d4fb90bfff4c436aae738ddac0c499)

Discussion