Interpretability of Neural Networks Based on Game-theoretic Interactions
文献类型：期刊论文
| 作者 | Huilin Zhou1; Jie Ren1; Huiqi Deng1; Xu Cheng1; Jinpeng Zhang2 ; Quanshi Zhang1
|
| 刊名 | Machine Intelligence Research
 |
| 出版日期 | 2024 |
| 卷号 | 21期号:4页码:718-739 |
| 关键词 | Model interpretability and transparency explainable AI game theory interaction deep learning |
| ISSN号 | 2731-538X |
| DOI | 10.1007/s11633-023-1419-7 |
| 英文摘要 | This paper introduces the system of game-theoretic interactions, which connects both the explanation of knowledge encoded in a deep neural networks (DNN) and the explanation of the representation power of a DNN. In this system, we define two game theoretic interaction indexes, namely the multi-order interaction and the multivariate interaction. More crucially, we use these interaction indexes to explain feature representations encoded in a DNN from the following four aspects: 1) Quantifying knowledge concepts encoded by a DNN; 2) Exploring how a DNN encodes visual concepts, and extracting prototypical concepts encoded in the DNN; 3) Learning optimal baseline values for the Shapley value, and providing a unified perspective to compare fourteen different attribution methods; 4) Theoretically explaining the representation bottleneck of DNNs. Furthermore, we prove the relationship between the interaction encoded in a DNN and the representation power of a DNN (e.g., generalization power, adversarial transferability, and adversarial robustness). In this way, game-theoretic interactions successfully bridge the gap between “the explanation of knowledge concepts encoded in a DNN” and “the explanation of the representation capacity of a DNN” as a unified explanation. |
| 源URL | [http://ir.ia.ac.cn/handle/173211/58569]  |
| 专题 | 自动化研究所_学术期刊_International Journal of Automation and Computing |
| 作者单位 | 1.School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China 2.XLAB, The Second Academy of China Aerospace Science and Industry Corporation, Beijing 100854, China |
| 推荐引用方式 GB/T 7714 | Huilin Zhou, Jie Ren, Huiqi Deng,et al. Interpretability of Neural Networks Based on Game-theoretic Interactions[J]. Machine Intelligence Research,2024,21(4):718-739. |
| APA | Huilin Zhou, Jie Ren, Huiqi Deng, Xu Cheng,Jinpeng Zhang,& Quanshi Zhang.(2024).Interpretability of Neural Networks Based on Game-theoretic Interactions.Machine Intelligence Research,21(4),718-739. |
| MLA | Huilin Zhou,et al."Interpretability of Neural Networks Based on Game-theoretic Interactions".Machine Intelligence Research 21.4(2024):718-739. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。