Generic Attention-model Explainability by Weighted Relevance Accumulation

by   Yiming Huang, et al.

Attention-based transformer models have achieved remarkable progress in multi-modal tasks, such as visual question answering. The explainability of attention-based methods has recently attracted wide interest as it can explain the inner changes of attention tokens by accumulating relevancy across attention layers. Current methods simply update relevancy by equally accumulating the token relevancy before and after the attention processes. However, the importance of token values is usually different during relevance accumulation. In this paper, we propose a weighted relevancy strategy, which takes the importance of token values into consideration, to reduce distortion when equally accumulating relevance. To evaluate our method, we propose a unified CLIP-based two-stage model, named CLIPmapper, to process Vision-and-Language tasks through CLIP encoder and a following mapper. CLIPmapper consists of self-attention, cross-attention, single-modality, and cross-modality attention, thus it is more suitable for evaluating our generic explainability method. Extensive perturbation tests on visual question answering and image captioning validate that our explainability method outperforms existing methods.


page 1

page 4

page 6

page 8

page 9


Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformers

Transformers are increasingly dominating multi-modal reasoning tasks, su...

Multi-Layer Attention-Based Explainability via Transformers for Tabular Data

We propose a graph-oriented attention-based explainability method for ta...

A Novel Attention-based Aggregation Function to Combine Vision and Language

The joint understanding of vision and language has been recently gaining...

Compound Tokens: Channel Fusion for Vision-Language Representation Learning

We present an effective method for fusing visual-and-language representa...

Predicting Token Impact Towards Efficient Vision Transformer

Token filtering to reduce irrelevant tokens prior to self-attention is a...

Simple is not Easy: A Simple Strong Baseline for TextVQA and TextCaps

Texts appearing in daily scenes that can be recognized by OCR (Optical C...

Auto-Parsing Network for Image Captioning and Visual Question Answering

We propose an Auto-Parsing Network (APN) to discover and exploit the inp...

Please sign up or login with your details

Forgot password? Click here to reset