Visualizing the Emergence of Intermediate Visual Patterns in DNNs

Authors: Mingjie Li, Shaobo Wang, Quanshi Zhang

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we used our method to visualize sample features and regional features in VGG-16 [49], Res Net-34/50 [21], Mobile Net-V2 [44], which were learned for object classification, based on the Tiny Image Net dataset [26], the MS COCO 2014 dataset [30], and the CUB-200-2011 dataset [55].
Researcher Affiliation Academia Mingjie Li Shanghai Jiao Tong University limingjie0608@sjtu.edu.cn Shaobo Wang Harbin Institute of Technology 181110315@stu.hit.edu.cn Quanshi Zhang Shanghai Jiao Tong University zqs1022@sjtu.edu.cn
Pseudocode No The paper describes its algorithms in prose and mathematical formulations but does not contain any explicitly labeled 'Pseudocode' or 'Algorithm' blocks.
Open Source Code No The paper does not provide any explicit statement about open-source code availability for the methodology described, nor does it include links to a code repository.
Open Datasets Yes In this section, we used our method to visualize sample features and regional features in VGG-16 [49], Res Net-34/50 [21], Mobile Net-V2 [44], which were learned for object classification, based on the Tiny Image Net dataset [26], the MS COCO 2014 dataset [30], and the CUB-200-2011 dataset [55].
Dataset Splits No The paper mentions using images for both training and testing for certain datasets but does not provide specific details on dataset splits (e.g., percentages, sample counts, or explicit validation sets).
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or detailed computer specifications) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details, such as library names with version numbers, required to replicate the experiment.
Experiment Setup Yes We set α = 0.1. The attack was iterated for 20 steps with the step size of 0.1 255. We selected the feature before the last fully-connected layer as the sample feature f. We estimated regional importance w(r) with κ set to 1000. We could simply set τ = 0.4.