Generating visual explanations with natural language

Hendricks, Lisa Anne; Rohrbach, Anna; Schiele, Bernt; Darrell, Trevor; Akata, Zeynep

doi:10.1002/ail2.55

Cited by 14 publications

(14 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…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].…”

Section: Related Workmentioning

confidence: 99%

Explaining Classifiers with Causal Concept Effect (CaCE)

Goyal¹,

Feder²,

Shalit³

et al. 2019

Preprint

View full text Add to dashboard Cite

How can we understand classification decisions made by deep neural nets? We propose answering this question by using ideas from causal inference. We define the "Causal Concept Effect" (CaCE) as the causal effect that the presence or absence of a concept has on the prediction of a given deep neural net. We then use this measure as a mean to understand what drives the network's prediction and what does not. Yet many existing interpretability methods rely solely on correlations, resulting in potentially misleading explanations. We show how CaCE can avoid such mistakes. In high-risk domains such as medicine, knowing the root cause of the prediction is crucial. If we knew that the network's prediction was caused by arbitrary concepts such as the lighting conditions in an X-ray room instead of medically meaningful concept, this would prevent us from disastrous deployment of such model. Estimating CaCE is difficult in situations where we cannot easily simulate the do-operator. As a simple solution, we propose learning a generative model, specifically a Variational AutoEncoder (VAE) on image pixels or image embeddings extracted from the classifier to measure VAE-CaCE. We show that VAE-CaCE is able to correctly estimate the true causal effect as compared to other baselines in controlled settings with synthetic and semi-natural high dimensional images. * This work was done while Yash Goyal was interning at Google Brain.Preprint. Under review.

show abstract

Section: Related Workmentioning

confidence: 99%

Explaining Classifiers with Causal Concept Effect (CaCE)

Goyal¹,

Feder²,

Shalit³

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…In [18], a textual explanation model is proposed to investigate why the model predicted a class x instead of class X based on counterfactuals. They offered a "phrasecritic model" using explanation annotations and its counterparts.…”

Section: Bmentioning

confidence: 99%

“…Lack of alignment in the image-text pairs emphasizes that mere accuracy is not enough if it is not accompanied by a valid justification, i.e., to be right for the right reason [16]. Explainability efforts are also carried out by generating visual [17] and textual explanations [18], but they can result in multiple explanations with varied evaluation metrics. Modular approaches make the system interpretable by design [19], [20] but are majorly tested on synthetic datasets such as CLEVR [21].…”

Section: Introductionmentioning

confidence: 99%

A Review on Explainability in Multimodal Deep Neural Nets

2021

View full text Add to dashboard Cite

Artificial Intelligence techniques powered by deep neural nets have achieved much success in several application domains, most significantly and notably in the Computer Vision applications and Natural Language Processing tasks. Surpassing human-level performance propelled the research in the applications where different modalities amongst language, vision, sensory, text play an important role in accurate predictions and identification. Several multimodal fusion methods employing deep learning models are proposed in the literature. Despite their outstanding performance, the complex, opaque and black-box nature of the deep neural nets limits their social acceptance and usability. This has given rise to the quest for model interpretability and explainability, more so in the complex tasks involving multimodal AI methods. This paper extensively reviews the present literature to present a comprehensive survey and commentary on the explainability in multimodal deep neural nets, especially for the vision and language tasks. Several topics on multimodal AI and its applications for generic domains have been covered in this paper, including the significance, datasets, fundamental building blocks of the methods and techniques, challenges, applications, and future trends in this domain.INDEX TERMS deep multimodal learning, explainable AI, interpretability, survey, trends, vision and language research, XAI.

show abstract

“…The explanation provides only the most critical changes (i.e., adding wool and removing beard and horns) required to alter the model's prediction from Goat to Sheep, though several other changes may be necessary. While there are recent works on generating pixel-level counter-factual and contrastive explanations [41], [42], [43], to the best of our knowledge, this is the first work to propose a method for generating explanations that are iterative, counter-factual as well as conceptual.…”

Section: ) Explanation Is An Interactive Communication Processmentioning

confidence: 99%

CX-ToM: Counterfactual Explanations with Theory-of-Mind for Enhancing Human Trust in Image Recognition Models

Akula¹,

Wang²,

Liu³

et al. 2021

Preprint

View full text Add to dashboard Cite

We propose CX-ToM, short for counterfactual explanations with theory-of mind, a new explainable AI (XAI) framework for explaining decisions made by a deep convolutional neural network (CNN). In contrast to the current methods in XAI that generate explanations as a single shot response, we pose explanation as an iterative communication process, i.e. dialog, between the machine and human user. More concretely, our CX-ToM framework generates sequence of explanations in a dialog by mediating the differences between the minds of machine and human user. To do this, we use Theory of Mind (ToM) which helps us in explicitly modeling human's intention, machine's mind as inferred by the human as well as human's mind as inferred by the machine. Moreover, most state-of-the-art XAI frameworks provide attention (or heat map) based explanations. In our work, we show that these attention based explanations are not sufficient for increasing human trust in the underlying CNN model. In CX-ToM, we instead use counterfactual explanations called fault-lines which we define as follows: given an input image I for which a CNN classification model M predicts class c pred , a fault-line identifies the minimal semantic-level features (e.g., stripes on zebra, pointed ears of dog), referred to as explainable concepts, that need to be added to or deleted from I in order to alter the classification category of I by M to another specified class c alt . We argue that, due to the iterative, conceptual and counterfactual nature of CX-ToM explanations, our framework is practical and more natural for both expert and non-expert users to understand the internal workings of complex deep learning models. Extensive quantitative and qualitative experiments verify our hypotheses, demonstrating that our CX-ToM significantly outperforms the state-of-the-art explainable AI models.

show abstract

Generating visual explanations with natural language

Cited by 14 publications

References 36 publications

Explaining Classifiers with Causal Concept Effect (CaCE)

Explaining Classifiers with Causal Concept Effect (CaCE)

A Review on Explainability in Multimodal Deep Neural Nets

CX-ToM: Counterfactual Explanations with Theory-of-Mind for Enhancing Human Trust in Image Recognition Models

Contact Info

Product

Resources

About