Timezone: »

Quantification and Analysis of Layer-wise and Pixel-wise Information Discarding
Haotian Ma · Hao Zhang · Fan Zhou · Yinqing Zhang · Quanshi Zhang

Tue Jul 19 03:30 PM -- 05:30 PM (PDT) @ Hall E #909

This paper presents a method to explain how the information of each input variable is gradually discarded during the forward propagation in a deep neural network (DNN), which provides new perspectives to explain DNNs. We define two types of entropy-based metrics, i.e. (1) the discarding of pixel-wise information used in the forward propagation, and (2) the uncertainty of the input reconstruction, to measure input information contained by a specific layer from two perspectives. Unlike previous attribution metrics, the proposed metrics ensure the fairness of comparisons between different layers of different DNNs. We can use these metrics to analyze the efficiency of information processing in DNNs, which exhibits strong connections to the performance of DNNs. We analyze information discarding in apixel-wise manner, which is different from the information bottleneck theory measuring feature information w.r.t. the sample distribution. Experiments have shown the effectiveness of our metrics in analyzing classic DNNs and explaining existing deep-learning techniques. The code is available at https://github.com/haotianSustc/deepinfo.

Author Information

Haotian Ma (Southern University of Science and Technology)
Hao Zhang (Shanghai Jiao Tong University)
Fan Zhou (Shanghai Jiao Tong University)
Yinqing Zhang (Shanghai Jiao Tong University)
Quanshi Zhang (Shanghai Jiao Tong University)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors