Activation Map

Activation maps (AMs) visualize the internal workings of deep neural networks, revealing which parts of the input data most strongly influence a model's predictions. Current research focuses on improving AM generation techniques, particularly for weakly supervised learning and applications where labeled data is scarce, often employing architectures like U-Nets and CNNs, and exploring methods like Class Activation Maps (CAMs) and their variants. These advancements enhance the interpretability of complex models, leading to improved model understanding, more reliable predictions, and facilitating applications in diverse fields such as medical image analysis, autonomous driving, and face recognition. Furthermore, research is exploring the use of AMs for model debugging, detecting data set shifts, and improving model efficiency.

Papers

November 5, 2024

SurfGNN: A robust surface-based prediction model with interpretability for coactivation maps of spatial and cortical features
Zhuoshuo Li (1), Jiong Zhang (2), Youbing Zeng (1), Jiaying Lin (1), Dan Zhang (3), Jianjia Zhang (1), Duan Xu (4), Hosung Kim (5), Bingguang Liu (6), Mengting Liu (1) ((1) Department of Biomedical Engineering, Sun Yat-sen University, Shenzhen, China, (2) Institute of Biomedical Engineering, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, China, (3) School of Cyber Science and Engineering, Ningbo University of Technology, Ningbo, China, (4) Department of Radiology, School of Medicine, University of California San Francisco, San Francisco, CA, USA, (5) USC Mark and Mary Stevens Neuroimaging and Informatics Institute, Keck School of Medicine of USC, University of Southern California, Los Angeles, CA, USA, (6) Department of Radiology, Shenzhen Maternity and Child Healthcare Hospital, Shenzhen, China)
Inherent Interpretability Predictive Model Brain Age Activation Map Cortical Surface Interpretable Graph Neural Network

October 24, 2024

Equitable Federated Learning with Activation Clustering
Antesh Upadhyay, Abolfazl Hashemi
Algorithmic Bias Unobserved Heterogeneity Activation Map Similarity Matrix Clustering Based Approach

July 8, 2024

LaFAM: Unsupervised Feature Attribution with Label-free Activation Maps
Aray Karjauv, Sahin Albayrak
Convolutional Neural Network Feature Attribution Class Activation Activation Map

April 30, 2024

Guiding Attention in End-to-End Driving Models
Diego Porres, Yi Xiao, Gabriel Villalonga, Alexandre Levy, Antonio M. López
Autonomous Driving Imitation Learning Semantic Map Activation Map Autonomous Driving Model End to End Driving

January 2, 2024

Integrating Edges into U-Net Models with Explainable Activation Maps for Brain Tumor Segmentation using MR Images
Subin Sahayam, Umarani Jayaraman
Tumor Segmentation Brain Tumor Segmentation Magnetic Resonance Image Manipulation Edge Detection Module Activation Map U Net Model

December 19, 2023

CAManim: Animating end-to-end network activation maps
Emily Kaczmarek, Olivier X. Miguel, Alexa C. Bowie, Robin Ducharme, Alysha L. J. Dingwall-Harvey, Steven Hawken, Christine M. Armour, Mark C. Walker, Kevin Dick
Convolutional Neural Network Explainable Artificial Intelligence Activation Map XAI Model

November 29, 2023

CRAFT: Contextual Re-Activation of Filters for face recognition Training
Aman Bhatta, Domingo Mery, Haiyu Wu, Kevin W. Bowyer
Face Recognition Convolutional Filter Max Filter Activation Map Traditional CrAFT CNN Backbone

November 17, 2023

On Functional Activations in Deep Neural Networks
Andrew S. Nencka, L. Tugan Muftuler, Peter LaViolette, Kevin M. Koch
Deep Neural Network Activation Map Functional Network Prompt Framework Structure Function

September 21, 2023

Activation Compression of Graph Neural Networks using Block-wise Quantization with Improved Variance Minimization
Sebastian Eliassen, Raghavendra Selvan
Graph Neural Network Efficient Training Large Scale Graph Activation Map Activation Compression Block Quantization

June 20, 2023

UM-CAM: Uncertainty-weighted Multi-resolution Class Activation Maps for Weakly-supervised Fetal Brain Segmentation
Jia Fu, Tao Lu, Shaoting Zhang, Guotai Wang
Weakly Supervised Fetal Brain Activation Map Weakly Supervised Segmentation Fm G Cam

June 1, 2023

Analyzing the Internals of Neural Radiance Fields
Lukas Radl, Andreas Kurz, Michael Steiner, Markus Steinberger
Neural Radiance Field Activation Map ReLU Neural Network

May 22, 2023

MAGDiff: Covariate Data Set Shift Detection via Activation Graphs of Deep Neural Networks
Charles Arnal, Felix Hensel, Mathieu Carrière, Théo Lacombe, Hiroaki Kurihara, Yuichi Ike, Frédéric Chazal
Deep Neural Network Covariate Shift Task Adaptation Activation Map Novel Representation Sequential Change Point Detection

March 27, 2023

Identifying Suspicious Regions of Covid-19 by Abnormality-Sensitive Activation Mapping
Ryo Toda, Hayato Itoh, Masahiro Oda, Yuichiro Hayashi, Yoshito Otake, Masahiro Hashimoto, Toshiaki Akashi, Shigeki Aoki, Kensaku Mori
Covid 19 COVID 19 Diagnosis Chest CT Activation Map CT Volume

December 15, 2022

CLAM: Selective Clarification for Ambiguous Questions with Generative Language Models
Lorenz Kuhn, Yarin Gal, Sebastian Farquhar
Language Model Dialogue System Generative Language Model Multi Turn Dialogue Ambiguous Question Activation Map

December 13, 2022

Interpretable Diabetic Retinopathy Diagnosis based on Biomarker Activation Map
Pengxiao Zang, Tristan T. Hormel, Jie Wang, Yukun Guo, Steven T. Bailey, Christina J. Flaxel, David Huang, Thomas S. Hwang, Yali Jia
Optical Coherence Tomography Diabetic Retinopathy Biomarker Selection Activation Map

October 22, 2022

SLAMs: Semantic Learning based Activation Map for Weakly Supervised Semantic Segmentation
Junliang Chen, Xiaodong Zhao, Minmin Liu, Linlin Shen
Semantic Segmentation Weakly Supervised Semantic Segmentation Pin Slam Semantic Learning Activation Map Image Level Classification

July 24, 2022

A Transformer-based Neural Language Model that Synthesizes Brain Activation Maps from Free-Form Text Queries
Gia H. Ngo, Minh Nguyen, Nancy F. Chen, Mert R. Sabuncu
Transformer Based Language Model Natural Language Query Neuroscience Research Activation Pattern Activation Map

June 2, 2022

A Deep Learning Network for the Classification of Intracardiac Electrograms in Atrial Tachycardia
Zerui Chen, Sonia Xhyn Teo, Andrie Ochtman, Shier Nee Saw, Nicholas Cheng, Eric Tien Siang Lim, Murphy Lyu, Hwee Kuan Lee
Classification Code Long Short Term Memory Deep Learning Network Activation Map Ventricular Tachycardia

May 27, 2022

CREAM: Weakly Supervised Object Localization via Class RE-Activation Mapping
Jilan Xu, Junlin Hou, Yuejie Zhang, Rui Feng, Rui-Wei Zhao, Tao Zhang, Xuequan Lu, Shang Gao
Class Activation Weakly Supervised Object Localization Activation Map Context Embeddings

May 2, 2022

Zebra: Memory Bandwidth Reduction for CNN Accelerators With Zero Block Regularization of Activation Maps
Hsu-Tung Shih, Tian-Sheuan Chang
Pruning Method Tiny ImageNet Major Challenge Bottleneck Activation Map CNN Accelerator

Activation Map

Papers

SurfGNN: A robust surface-based prediction model with interpretability for coactivation maps of spatial and cortical features

Equitable Federated Learning with Activation Clustering

LaFAM: Unsupervised Feature Attribution with Label-free Activation Maps

Guiding Attention in End-to-End Driving Models

Integrating Edges into U-Net Models with Explainable Activation Maps for Brain Tumor Segmentation using MR Images

CAManim: Animating end-to-end network activation maps

CRAFT: Contextual Re-Activation of Filters for face recognition Training

On Functional Activations in Deep Neural Networks

Activation Compression of Graph Neural Networks using Block-wise Quantization with Improved Variance Minimization

UM-CAM: Uncertainty-weighted Multi-resolution Class Activation Maps for Weakly-supervised Fetal Brain Segmentation

Analyzing the Internals of Neural Radiance Fields

MAGDiff: Covariate Data Set Shift Detection via Activation Graphs of Deep Neural Networks

Identifying Suspicious Regions of Covid-19 by Abnormality-Sensitive Activation Mapping

CLAM: Selective Clarification for Ambiguous Questions with Generative Language Models

Interpretable Diabetic Retinopathy Diagnosis based on Biomarker Activation Map

SLAMs: Semantic Learning based Activation Map for Weakly Supervised Semantic Segmentation

A Transformer-based Neural Language Model that Synthesizes Brain Activation Maps from Free-Form Text Queries

A Deep Learning Network for the Classification of Intracardiac Electrograms in Atrial Tachycardia

CREAM: Weakly Supervised Object Localization via Class RE-Activation Mapping

Zebra: Memory Bandwidth Reduction for CNN Accelerators With Zero Block Regularization of Activation Maps