D
Die Zhang
Researcher at Shanghai Jiao Tong University
Publications - 8
Citations - 37
Die Zhang is an academic researcher from Shanghai Jiao Tong University. The author has contributed to research in topics: Inference & Computer science. The author has an hindex of 4, co-authored 8 publications receiving 28 citations.
Papers
More filters
Posted Content
Interpreting Multivariate Interactions in DNNs
TL;DR: This paper defines and quantify the significance of interactions among multiple input variables of the DNN based on the Shapley value, which is designed to assign the attribution value of each input variable to the inference.
Proceedings Article
Building Interpretable Interaction Trees for Deep NLP Models.
Die Zhang,Hao Zhang,Huilin Zhou,Xiaoyi Bao,Da Huo,Ruizhao Chen,Xu Cheng,Mengyue Wu,Quanshi Zhang +8 more
TL;DR: This paper proposed a method to disentangle and quantify interactions among words that are encoded inside a DNN for natural language processing, and constructed a tree to encode salient interactions extracted by the DNN.
Posted Content
Interpreting Multivariate Shapley Interactions in DNNs
TL;DR: In this paper, the significance of multivariate interactions among multiple input variables of the DNN is defined and quantified based on the Shapley value, which is designed to assign the attribution value of each input variable to the inference.
Posted Content
Game-theoretic Understanding of Adversarially Learned Features.
Jie Ren,Die Zhang,Yisen Wang,Lu Chen,Zhanpeng Zhou,Xu Cheng,Xin Wang,Yiting Chen,Jie Shi,Quanshi Zhang +9 more
TL;DR: In this article, the authors define the multi-order interaction in game theory, which satisfies six properties: adversarial attacks mainly affect high-order interactions to fool the DNN, and the robustness of adversarially trained DNNs comes from category-specific loworder interactions.
Posted Content
Interpreting Hierarchical Linguistic Interactions in DNNs.
Die Zhang,Huilin Zhou,Xiaoyi Bao,Da Huo,Ruizhao Chen,Xu Cheng,Hao Zhang,Mengyue Wu,Quanshi Zhang +8 more
TL;DR: This paper proposes a method to disentangle and quantify interactions among words that are encoded inside a DNN for natural language processing, and constructs a tree to encode salient interactions extracted by the DNN.