Counterfactual Visual Explanations

By Yash Goyal, Ziyan Wu, Jan Ernst, Dhruv Batra, Devi Parikh, Stefan Lee. 2019

596 citations recorded by the source corpus.

In this work, we develop a technique to produce counterfactual visual explanations. Given a 'query' image $I$ for which a vision system predicts class $c$, a counterfactual visual explanation identifies how $I$ could change such that the system would output a different specified class $c'$. To do this, we select a 'distractor' image $I'$ that the system predicts as class $c'$ and identify spatial regions in $I$ and $I'$ such that replacing the identified region in $I$ with the identified region in $I'$ would push the system towards classifying $I$ as $c'$. We apply our approach to multiple image classification datasets generating qualitative results showcasing the interpretability and discriminativeness of our counterfactual explanations. To explore the effectiveness of our explanations in teaching humans, we present machine teaching experiments for the task of fine-grained bird classification. We find that users trained to distinguish bird species fare better when given access to counterfactual explanations in addition to training examples.

Subfields: Artificial Intelligence, Biophysics, Computer Vision and Pattern Recognition, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Biophysics

In Computer Vision and Pattern Recognition

In AI Ethics, Fairness & Society

Citations in the corpus

CausaLM: Causal Model Explanation Through Counterfactual Language Models (2020, methodology): “Using generative models has been recently discussed in the case of images (Goyal et al. 2019a).”
Generative Counterfactual Introspection for Explainable Deep Learning (2019, background): “In the counterfactual visual explanation [22] work, a patch based editing of input image is optimize in order to satisfy the intended changes in the prediction.”
EDUCE: Explaining model Decisions through Unsupervised Concepts Extraction (2019, background): “[7] propose visual explanations of a classiﬁer’s decision, while [1] use an architecture composed of an observer and a classiﬁer, in which the classiﬁer’s prediction can be exposed as a binary tree structure.”
Causal Interpretability for Machine Learning - Problems, Methods and Evaluation (2020, background): “nference have gained a lot of attention from the interpretable machine learning ﬁeld. Research in this area has mainly focused on generating counterfactual explanations from both the data perspective [34; 76] as well as the components of a model [77; 38]. Existing surveys on interpretable machine le”
Explaining Classifiers with Causal Concept Effect (CaCE) (2019, neutral): “Many interpretability methods developed to have causal-ﬂavor are for local explanations, such as removing and adding pixels to generate counterfactual explanations for images [5, 1] or for texts [7].”
An Empirical Study on Robustness to Spurious Correlations using Pre-trained Language Models (2020, background): “Similarly, counterfactual augmentation aims to identify parts of the input that impact the label when intervened upon, thus avoiding learning spurious features (Goyal et al., 2019; Kaushik et al., 2020).”
A Survey on Neural Network Interpretability (2020, background): “[79] try to find the minimum-edit on the input image which can result in a different predicted class c′.”
Counterfactual Vision-and-Language Navigation via Adversarial Path Sampling (2019, methodology): “In the field of data science, counterfactual thinking has been used to make trained models explainable and more robust [24, 12, 14].”
Explaining in Style: Training a GAN to explain a classifier in StyleSpace (2021, methodology): “Among the different forms of explanations, counterfactual explanations are gaining increasing attention [18, 11, 32].”
Towards causal benchmarking of bias in face analysis algorithms (2020, background): “ally, there is research into interpreting neural networks. One strategy is to determine regions of the input that are salient, either through analysis of gradients or perturbations of the input image [13, 16, 24, 26, 69, 72, 74]. Network dissection approaches explore how particular neurons within a ”
On Interpretability of Artificial Neural Networks: A Survey (2020, background): “Y. Goyal et al. [62] explored an alternative way to derive a counterfactual visual explanation.”
On Completeness-aware Concept-Based Explanations in Deep Neural Networks (2019, background): “…14], (b) sample-based explanation methods, that attribute the decision to previously observed samples [15, 16, 17, 18], and (c) counterfactual-based explanation methods, which answers the question: “what to alter in the current input to change the outcome of the model” [19, 20, 21, 22, 23, 24, 25].”
Distilling Model Failures as Directions in Latent Space (2022, unknown): “They include gradient-(Simonyan et al., 2013; Dabkowski & Gal, 2017; Sundararajan et al., 2017) and perturbation-based (Ribeiro et al., 2016a; Goyal et al., 2019; Fong & Vedaldi, 2017; Dabkowski & Gal, 2017; Zintgraf et al., 2017; Dhurandhar et al., 2018; Chang et al., 2019; Hendricks et al., 2018;…”
Explaining Visual Models by Causal Attribution (2019, background): “[7], on the other hand, consider a counterfactual as an image similar to the original one but resulting on a different prediction.”
Empowering Things With Intelligence: A Survey of the Progress, Challenges, and Opportunities in Artificial Intelligence of Things (2020, unknown): “Moreover, causal inference can be used to unfold the “black-box” decision process of DNNs to address the interpretability issue, including model-based interpretation methods [232] and example-based interpretation methods [233].”
A Survey of Contrastive and Counterfactual Explanation Generation Methods for Explainable Artificial Intelligence (2021, background): “report how often the discriminative regions lie inside the test data example segmentations as well as relevant specific key regions [139].”
Explaining NLP Models via Minimal Contrastive Editing (MiCE) (2020, background): “The problem of generating minimal contrastive edits, also called counterfactual explanations (Wachter et al., 2017), 12 has previously been explored for tabular data (Karimi et al., 2020) and images (Hendricks et al., 2018; Goyal et al., 2019; Looveren and Klaise, 2019) but less for language.”
SurvNAM: The machine learning survival model explanation (2021, background): “Due to intuitive and human-friendly explanations provided by this family of methods, several its modifications has been proposed [35, 36, 37, 38].”
Counterfactual VQA: A Cause-Effect Look at Language Bias (2020, background): “Recently, counterfactual thinking has been exploited in various computer vision problems, including visual explanations [48, 49], scene graph generation [50], video analysis [51, 52], and vision-language tasks [29, 27, 32].”