On Explaining Neural Network Robustness with Activation Path
Ziping Jiang
Abstract
<p>ICLR submission. This work is supported in part by the Huawei Technologies Co., Ltd under Grant HIRP2019041002010, the Horizon<br> Europe: European Lighthouse on Secure and Safe AI – ELSA, the UK EPSRC under Grant EP/P009727/1, and the<br> Leverhulme Trust under Grant RF-2019-492 </p>
Ask Lune
Ask anything. The answer shows its sources.
Lune reads the papers around it, then shows you the passages behind every answer.
Related papers
- Can Language Models Follow Multiple Turns of Entangled Instructions?Chi Han, Xin Liu, Haodong Wang, Shiyang Li et al.EMNLP Findings 2025
- AHA: Aligning Large Audio-Language Models for Reasoning Hallucinations via Counterfactual Hard NegativesYanxi Chen, Wenhui Zhu, Xiwen Chen, Zhipeng Wang et al.ACL Findings 2026
- D²-RAG: Dual-Decision Retrieval-Augmented Generation via Multi-Dimensional Uncertainty and Utility-Aware DecodingJinshuo Zhang, Xiaoding Zhou, Weiyu Zhang, Guoqiang Chen et al.ACL Findings 2026
- Compare to The Knowledge: Graph Neural Fake News Detection with External KnowledgeLinmei Hu, Tianchi Yang, Luhao Zhang, Wanjun Zhong et al.ACL 2021
- Sparse Growing Transformer: Training-Time Sparse Depth Allocation via Progressive Attention LoopingYao Chen, Yilong Chen, Yinqi Yang, Junyuan Shang et al.ACL Findings 2026
