Quantification and Analysis of Layer-wise and Pixel-wise Information Discarding

Authors: Haotian Ma, Hao Zhang, Fan Zhou, Yinqing Zhang, Quanshi Zhang

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments have shown the effectiveness of our metrics in analyzing classic DNNs and explaining existing deep-learning techniques.
Researcher Affiliation Academia 1 Shanghai Jiao Tong University, Shanghai, China 2 Southern University of Science and Technology, Shenzhen, China.
Pseudocode No The paper describes mathematical formulations and procedures (e.g., Equation (3) and (4)), but does not include any clearly labeled 'Pseudocode' or 'Algorithm' block or figure.
Open Source Code Yes The code is available at https://github.com/haotianSustc/deepinfo.
Open Datasets Yes CUB200-2011 dataset (Wah et al., 2011), CIFAR-10 dataset (Krizhevsky, 2009), ISBI cell tracking challenge (WWW, 2012), Image Net dataset (Russakovsky et al., 2015).
Dataset Splits No No explicit training/validation/test splits (e.g., percentages, sample counts) are provided for the datasets used in the experiments. The paper states 'we used object images cropped by object bounding boxes for both training and testing' but lacks specific split details.
Hardware Specification No No specific hardware details (such as GPU models, CPU types, or cloud instance specifications) used for running the experiments are mentioned in the paper.
Software Dependencies No No specific software dependencies with version numbers (e.g., Python 3.8, PyTorch 1.9) are mentioned in the paper.
Experiment Setup Yes In order to learn the parameter σ, we used the learning rate 1 10 4, and learned σ for 100 epochs. ... In the following experiments, we set β = 1.5 10 4.