• Home
  • Search
  • Information Bottleneck Approach to Spatial Attention Learning
  • Cite Icon7
  • https://doi.org/10.24963/ijcai.2021/108Copy DOI Icon

Information Bottleneck Approach to Spatial Attention Learning

  • Aug 1, 2021
  • Qiuxia Lai +5 more
Show More
  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

The selective visual attention mechanism in the human visual system (HVS) restricts the amount of information to reach visual awareness for perceiving natural scenes, allowing near real-time information processing with limited computational capacity. This kind of selectivity acts as an ‘Information Bottleneck (IB)’, which seeks a trade-off between information compression and predictive accuracy. However, such information constraints are rarely explored in the attention mechanism for deep neural networks (DNNs). In this paper, we propose an IB-inspired spatial attention module for DNN structures built for visual recognition. The module takes as input an intermediate representation of the input image, and outputs a variational 2D attention map that minimizes the mutual information (MI) between the attention-modulated representation and the input, while maximizing the MI between the attention-modulated representation and the task label. To further restrict the information bypassed by the attention map, we quantize the continuous attention scores to a set of learnable anchor values during training. Extensive experiments show that the proposed IB-inspired spatial attention mechanism can yield attention maps that neatly highlight the regions of interest while suppressing backgrounds, and bootstrap standard DNN structures for visual recognition tasks (e.g., image classification, fine-grained recognition, cross-domain classification). The attention maps are interpretable for the decision making of the DNNs as verified in the experiments. Our code is available at this https URL.

Similar Papers
  • Research Article
  • Citations1

Spatial attention for human-centric visual understanding: An Information Bottleneck method

  • Sep 24, 2024
  • Computer Vision and Image Understanding
  • Qiuxia Lai +4
  • Research Article
  • Citations15

SIAM: A parameter-free, Spatial Intersection Attention Module

  • Apr 21, 2024
  • Pattern Recognition
  • Gaoge Han +3
  • Research Article
  • Citations383

The role of neural mechanisms of attention in solving the binding problem.

  • Sep 01, 1999
  • Neuron
  • John H Reynolds +1
  • Research Article
  • Citations175

Deep distributed convolutional neural networks: Universality

  • Nov 01, 2018
  • Analysis and Applications
  • Ding-Xuan Zhou
  • Research Article
  • Citations1

Super-resolution reconstruction of binocular image based on multi-level fusion attention network

  • Jan 01, 2023
  • Journal of Image and Graphics
  • Xu Lei +2
  • Conference Article
  • Citations550

Attention Branch Network: Learning of Attention Mechanism for Visual Explanation

  • Jun 01, 2019
  • Hiroshi Fukui +3
  • PDF
  • Research Article
  • Citations6

Deterioration Level Estimation Based on Convolutional Neural Network Using Confidence-Aware Attention Mechanism for Infrastructure Inspection

  • Jan 05, 2022
  • Sensors (Basel, Switzerland)
  • Naoki Ogawa +3
  • Research Article
  • Citations117

Visual vs internal attention mechanisms in deep neural networks for image classification and object detection

  • Nov 03, 2021
  • Pattern Recognition
  • Abraham Montoya Obeso +3
  • Research Article
  • Citations40

Deep encoder–decoder-NN: A deep learning-based autonomous vehicle trajectory prediction and correction model

  • Jan 10, 2022
  • Physica A: Statistical Mechanics and its Applications
  • Fei Hui +4
  • Dissertation

Face Recognition with Attention Mechanisms

  • Jan 01, 2022
  • Qiangchang Wang
  • Research Article
  • Citations25

Multi-attention-based soft partition network for vehicle re-identification

  • Feb 23, 2023
  • Journal of Computational Design and Engineering
  • Sangrok Lee +2
  • PDF
  • Research Article
  • Citations45

Attention Map-Guided Visual Explanations for Deep Neural Networks

  • Apr 11, 2022
  • Applied Sciences
  • Junkang An +1
  • Research Article
  • Citations15

Brain tumor classification by using a novel convolutional neural network structure

  • Jun 02, 2022
  • International Journal of Imaging Systems and Technology
  • Özlem Polat +2
  • Conference Article
  • Citations21

Human-AI Interactive and Continuous Sensemaking: A Case Study of Image Classification using Scribble Attention Maps

  • May 08, 2021
  • Haifeng Shen +6
  • Conference Article

Analysis of the Information Plane for Deep Reinforcement Learning Using Proximal Policy Optimization

  • Dec 01, 2025
  • Arthur Fernandes +1
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.