no code implementations • 23 Jan 2024 • Zhen Liang, Taoran Wu, Ran Zhao, Bai Xue, Ji Wang, Wenjing Yang, Shaojun Deng, Wanwei Liu
However, these strategies face challenges in addressing the "unknown dilemma" concerning whether the exact output region or the introduced approximation error violates the property in question.
no code implementations • 29 Jul 2023 • Ye Tao, Wanwei Liu, Fu Song, Zhen Liang, Ji Wang, Hongxu Zhu
Quantized neural networks (QNNs) have been developed, with binarized neural networks (BNNs) restricted to binary values as a special case.
1 code implementation • 27 Jun 2023 • Zhen Liang, Dejin Ren, Bai Xue, Ji Wang, Wenjing Yang, Wanwei Liu
Moreover, for NNs that do not feature these properties with respect to the input set, we explore subsets of the input set for establishing the local homeomorphism property and then abandon these subsets for reachability computations.
1 code implementation • 5 May 2023 • Zhen Liang, Taoran Wu, Changyuan Zhao, Wanwei Liu, Bai Xue, Wenjing Yang, Ji Wang
For the fine-tuning repair process, BIRDNN analyzes the behavior differences of neurons on positive and negative samples to identify the most responsible neurons for the erroneous behaviors.
no code implementations • 2 Dec 2022 • Zhen Liang, Changyuan Zhao, Wanwei Liu, Bai Xue, Wenjing Yang, Zhengbin Pang
Based on Koopman operator theory, this paper presents an alternative perspective of linear dynamics on dealing with the credit assignment problem for trained neural networks.
1 code implementation • 9 Oct 2022 • Zhen Liang, Dejin Ren, Wanwei Liu, Ji Wang, Wenjing Yang, Bai Xue
The homeomorphism property exists in some widely used NNs such as invertible NNs.
no code implementations • 13 Mar 2020 • Renyan Feng, Erman Acar, Stefan Schlobach, Yisong Wang, Wanwei Liu
To address such a scenario in a principled way, we introduce a forgetting-based approach in CTL and show that it can be used to compute SNC and WSC of a property under a given model and over a given signature.
no code implementations • 9 Feb 2020 • Yufeng Zhang, Jialu Pan, Wanwei Liu, Zhenbang Chen, Ji Wang, Zhiming Liu, Kenli Li, Hongmei Wei
For point-wise anomaly detection, our method achieves 90. 7\% AUROC on average and outperforms the baseline by 5. 2\% AUROC.