CounteRGAN: Generating Realistic Counterfactuals with Residual Generative Adversarial Nets

Nemirovsky, Daniel; Thiébaut, Nicolas; Ye, Xu; Gupta, Abhishek

doi:10.48550/arxiv.2009.05199

Cited by 8 publications

(35 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CausalGAN (Kocaoglu et al, 2018) proposes to learn a causal implicit model through adversarial training with a given causal graph for facial attribute disentanglement. CounterGAN (Nemirovsky et al, 2020) employs a residual generator to improve counterfactual realism and actionability compared to regular GANs. Counterfactual Generative Network (CGN) (Sauer & Geiger, 2020) suggests to decouple the ImageNet generation into four aspects of the shape, texture, background, and composer.…”

Section: Related Workmentioning

confidence: 99%

“…Counterfactual synthesis (Kocaoglu et al, 2018;Sauer & Geiger, 2020;Nemirovsky et al, 2020;Yang et al, 2021;Averitt et al, 2020;Thiagarajan et al, 2021) is one of the most promising tasks to achieve the general goal of knowledge extrapolation in GANs. For counterfactual synthesis, if brusquely ignoring differences in details, most existing methods follow the framework of the same fashion -directly modeling a Structural Causal Model (SCM) (Pearl, 2009a) in well-designed architectures of generator networks.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Principled Knowledge Extrapolation with GANs

Feng¹,

Xiao²,

Zheng³

et al. 2022

Preprint

View full text Add to dashboard Cite

Human can extrapolate well, generalize daily knowledge into unseen scenarios, raise and answer counterfactual questions. To imitate this ability via generative models, previous works have extensively studied explicitly encoding Structural Causal Models (SCMs) into architectures of generator networks. This methodology, however, limits the flexibility of the generator as they must be carefully crafted to follow the causal graph, and demands a ground truth SCM with strong ignorability assumption as prior, which is a nontrivial assumption in many real scenarios. Thus, many current causal GAN methods fail to generate high fidelity counterfactual results as they cannot easily leverage state-of-the-art generative models. In this paper, we propose to study counterfactual synthesis from a new perspective of knowledge extrapolation, where a given knowledge dimension of the data distribution is extrapolated, but the remaining knowledge is kept indistinguishable from the original distribution. We show that an adversarial game with a closed-form discriminator can be used to address the knowledge extrapolation problem, and a novel principal knowledge descent method can efficiently estimate the extrapolated distribution through the adversarial game. Our method enjoys both elegant theoretical guarantees and superior performance in many scenarios.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Principled Knowledge Extrapolation with GANs

Feng¹,

Xiao²,

Zheng³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Figure 1: Counterfactuals generated using a state-of-the-art approach [Nemirovsky et al, 2020] and our approach for a multivariate time series. Columns represent features and rows represent time steps.…”

Section: Original Targetmentioning

confidence: 99%

“…To generate more realistic and plausible counterfactuals, while overcoming high computational costs of iterative optimization methods, generative adversarial networks (GANs) have recently been introduced for the generation of counterfactual explanations [Nemirovsky et al, 2020, Van Looveren et al, 2021. GANs have become popular for generating realistic looking fake images by training a generator to create fake samples that a discriminator would erroneously perceive as real samples [Goodfellow et al, 2014].…”

Section: Generative Approachesmentioning

confidence: 99%

Generating Sparse Counterfactual Explanations For Multivariate Time Series

Lang¹,

Giese²,

Ilg³

et al. 2022

Preprint

View full text Add to dashboard Cite

Since neural networks play an increasingly important role in critical sectors, explaining network predictions has become a key research topic. Counterfactual explanations can help to understand why classifier models decide for particular class assignments and, moreover, how the respective input samples would have to be modified such that the class prediction changes. Previous approaches mainly focus on image and tabular data. In this work we propose SPARCE 1 , a generative adversarial network (GAN) architecture that generates SPARse Counterfactual Explanations for multivariate time series. Our approach provides a custom sparsity layer and regularizes the counterfactual loss function in terms of similarity, sparsity, and smoothness of trajectories. We evaluate our approach on real-world human motion datasets as well as a synthetic time series interpretability benchmark. Although we make significantly sparser modifications than other approaches, we achieve comparable or better performance on all metrics. Moreover, we demonstrate that our approach predominantly modifies salient time steps and features, leaving non-salient inputs untouched.

show abstract

“…Olson et al [24] use a combination of such a GAN and a Wasserstein Autoencoder to create counterfactual states to explain Deep Reinforcement Learning algorithms for Atari games. Nemirovsky et al [25] proposed CounterGAN, an architecture in which a generator learns to produce residuals that result in counterfactual images when added to an input image. More recent GAN architectures allow the transformation of images between different image domains [26,27], known as Image-to-Image Translation.…”

Section: Adversarial Approaches To Counterfactual Image Generationmentioning

confidence: 99%

GANterfactual - Counterfactual Explanations for Medical Non-Experts using Generative Adversarial Learning

Mertes,

Huber,

Weitz

et al. 2020

Preprint

View full text Add to dashboard Cite

With the ongoing rise of machine learning, the need for methods for explaining decisions made by artificial intelligence systems is becoming a more and more important topic. Especially for image classification tasks, many state-of-the-art tools to explain such classifiers rely on visual highlighting of important areas of the input data. Contrary, counterfactual explanation systems try to enable a counterfactual reasoning by modifying the input image in a way such that the classifier would have made a different prediction. By doing so, the users of counterfactual explanation systems are equipped with a completely different kind of explanatory information. However, methods for generating realistic counterfactual explanations for image classifiers are still rare. In this work, we present a novel approach to generate such counterfactual image explanations based on adversarial image-to-image translation techniques. Additionally, we conduct a user study to evaluate our approach in a use case which was inspired by a healthcare scenario. Our results show that our approach leads to significantly better results regarding mental models, explanation satisfaction, trust, emotions, and self-efficacy than two state-of-the art systems that work with saliency maps, namely LIME and LRP.

show abstract

CounteRGAN: Generating Realistic Counterfactuals with Residual Generative Adversarial Nets

Cited by 8 publications

References 0 publications

Principled Knowledge Extrapolation with GANs

Principled Knowledge Extrapolation with GANs

Generating Sparse Counterfactual Explanations For Multivariate Time Series

GANterfactual - Counterfactual Explanations for Medical Non-Experts using Generative Adversarial Learning

Contact Info

Product

Resources

About