GAN Dissection: Visualizing and Understanding Generative Adversarial Networks

Authors: David Bau, Jun-Yan Zhu, Hendrik Strobelt, Bolei Zhou, Joshua B. Tenenbaum, William T. Freeman, Antonio Torralba

ICLR 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this work, we present an analytic framework to visualize and understand GANs at the unit-, object-, and scene-level. We first identify a group of interpretable units that are closely related to object concepts using a segmentation-based network dissection method. Then, we quantify the causal effect of interpretable units by measuring the ability of interventions to control objects in the output. We examine the contextual relationship between these units and their surroundings by inserting the discovered object concepts into new images. We show several practical applications enabled by our framework, from comparing internal representations across different layers, models, and datasets, to improving GANs by locating and removing artifact-causing units, to interactively manipulating objects in a scene.
Researcher Affiliation Collaboration 1Massachusetts Institute of Technology, 2MIT-IBM Watson AI Lab, 3IBM Research, 4The Chinese University of Hong Kong
Pseudocode No The paper provides mathematical formulations and figures to describe its methods but does not include structured pseudocode or algorithm blocks.
Open Source Code No The abstract mentions 'open source interpretation tools' but does not provide a concrete link to the source code repository or an explicit statement that the code itself is released.
Open Datasets Yes We study three variants of Progressive GANs (Karras et al., 2018) trained on LSUN scene datasets (Yu et al., 2015).
Dataset Splits No The paper mentions using 'a separate validation set' for choosing a threshold (tu,c) but does not provide specific details on dataset splits (e.g., percentages or sample counts) for training, validation, or testing.
Hardware Specification No The paper does not provide specific hardware details such as GPU or CPU models used for running its experiments.
Software Dependencies No The paper does not provide specific software dependencies or version numbers needed to replicate the experiments.
Experiment Setup No While the paper mentions optimization using stochastic gradient descent, it does not provide specific experimental setup details such as hyperparameter values (e.g., learning rate, batch size, number of epochs) or specific optimizer configurations.