DiffGradCAM: A Universal Class Activation Map Resistant to Adversarial Training

• URL: http://arxiv.org/abs/2506.08514v2
• Date: Fri, 18 Jul 2025 10:39:51 GMT
• Title: DiffGradCAM: A Universal Class Activation Map Resistant to Adversarial Training
• Authors: Jacob Piland, Chris Sweet, Adam Czajka,
• Abstract summary: Class Activation Mapping (CAM) and its gradient-based variants (e.g., GradCAM) have become standard tools for explaining Convolutional Neural Network (CNN) predictions.<n>We introduce textbfSalience-Hoax Activation Maps (SHAMs), an emphentropy-aware form of passive fooling.<n>We then propose textbfDiffGradCAM, a novel, lightweight, and contrastive approach to class activation mapping.
• Score: 3.858607108771203
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Class Activation Mapping (CAM) and its gradient-based variants (e.g., GradCAM) have become standard tools for explaining Convolutional Neural Network (CNN) predictions. However, these approaches typically focus on individual logits, while for neural networks using softmax, the class membership probability estimates depend \textit{only} on the \textit{differences} between logits, not on their absolute values. This disconnect leaves standard CAMs vulnerable to adversarial manipulation, such as passive fooling, where a model is trained to produce misleading CAMs without affecting decision performance. We introduce \textbf{Salience-Hoax Activation Maps (SHAMs)}, an \emph{entropy-aware form of passive fooling} that serves as a benchmark for CAM robustness under adversarial conditions. To address the passive fooling vulnerability, we then propose \textbf{DiffGradCAM}, a novel, lightweight, and contrastive approach to class activation mapping that is both non-suceptible to passive fooling, but also matches the output of standard CAM methods such as GradCAM in the non-adversarial case. Together, SHAM and DiffGradCAM establish a new framework for probing and improving the robustness of saliency-based explanations. We validate both contributions across multi-class tasks with few and many classes.

Related papers

• PseudoNeg-MAE: Self-Supervised Point Cloud Learning using Conditional Pseudo-Negative Embeddings [55.55445978692678]
  PseudoNeg-MAE enhances global feature representation of point cloud masked autoencoders by making them both discriminative and sensitive to transformations.<n>We propose a novel loss that explicitly penalizes invariant collapse, enabling the network to capture richer transformation cues while preserving discriminative representations.
  arXiv  Detail & Related papers  (2024-09-24T07:57:21Z)
• DecomCAM: Advancing Beyond Saliency Maps through Decomposition and Integration [25.299607743268993]
  Class Activation Map (CAM) methods highlight regions revealing the model's decision-making basis but lack clear saliency maps and detailed interpretability. We propose DecomCAM, a novel decomposition-and-integration method that distills shared patterns from channel activation maps. Experiments reveal that DecomCAM not only excels in locating accuracy but also achieves an optimizing balance between interpretability and computational efficiency.
  arXiv  Detail & Related papers  (2024-05-29T08:40:11Z)
• BroadCAM: Outcome-agnostic Class Activation Mapping for Small-scale Weakly Supervised Applications [69.22739434619531]
  We propose an outcome-agnostic CAM approach, called BroadCAM, for small-scale weakly supervised applications. By evaluating BroadCAM on VOC2012 and BCSS-WSSS for WSSS and OpenImages30k for WSOL, BroadCAM demonstrates superior performance.
  arXiv  Detail & Related papers  (2023-09-07T06:45:43Z)
• Attention-based Class Activation Diffusion for Weakly-Supervised Semantic Segmentation [98.306533433627]
  extracting class activation maps (CAM) is a key step for weakly-supervised semantic segmentation (WSSS) This paper proposes a new method to couple CAM and Attention matrix in a probabilistic Diffusion way, and dub it AD-CAM. Experiments show that AD-CAM as pseudo labels can yield stronger WSSS models than the state-of-the-art variants of CAM.
  arXiv  Detail & Related papers  (2022-11-20T10:06:32Z)
• Recipro-CAM: Gradient-free reciprocal class activation map [0.0]
  We propose a lightweight architecture and gradient free Reciprocal CAM (Recipro-CAM) to exploit the correlation between activation maps and network outputs. With the proposed method, we achieved the gains of 1:78 - 3:72% in the ResNet family compared to Score-CAM. In addition, Recipro-CAM exhibits a saliency map generation rate similar to Grad-CAM and approximately 148 times faster than Score-CAM.
  arXiv  Detail & Related papers  (2022-09-28T13:15:03Z)
• Class Re-Activation Maps for Weakly-Supervised Semantic Segmentation [88.55040177178442]
  Class activation maps (CAM) is arguably the most standard step of generating pseudo masks for semantic segmentation. Yet, the crux of the unsatisfactory pseudo masks is the binary cross-entropy loss (BCE) widely used in CAM. We introduce an embarrassingly simple yet surprisingly effective method: Reactivating the converged CAM with BCE by using softmax cross-entropy loss (SCE) The evaluation on both PASCAL VOC and MSCOCO shows that ReCAM not only generates high-quality masks, but also supports plug-and-play in any CAM variant with little overhead.
  arXiv  Detail & Related papers  (2022-03-02T09:14:58Z)
• SSA: Semantic Structure Aware Inference for Weakly Pixel-Wise Dense Predictions without Cost [36.27226683586425]
  The semantic structure aware inference (SSA) is proposed to explore the semantic structure information hidden in different stages of the CNN-based network to generate high-quality CAM in the model inference. The proposed method has the advantage of no parameters and does not need to be trained. Therefore, it can be applied to a wide range of weakly-supervised pixel-wise dense prediction tasks.
  arXiv  Detail & Related papers  (2021-11-05T11:07:21Z)
• Towards Learning Spatially Discriminative Feature Representations [26.554140976236052]
  We propose a novel loss function, termed as CAM-loss, to constrain the embedded feature maps with the class activation maps (CAMs) CAM-loss drives the backbone to express the features of target category and suppress the features of non-target categories or background. Experimental results show that CAM-loss is applicable to a variety of network structures and can be combined with mainstream regularization methods to improve the performance of image classification.
  arXiv  Detail & Related papers  (2021-09-03T08:04:17Z)
• Joint Noise-Tolerant Learning and Meta Camera Shift Adaptation for Unsupervised Person Re-Identification [60.36551512902312]
  unsupervised person re-identification (re-ID) aims to learn discriminative models with unlabeled data. One popular method is to obtain pseudo-label by clustering and use them to optimize the model. In this paper, we propose a unified framework to solve both problems.
  arXiv  Detail & Related papers  (2021-03-08T09:13:06Z)
• Use HiResCAM instead of Grad-CAM for faithful explanations of convolutional neural networks [89.56292219019163]
  Explanation methods facilitate the development of models that learn meaningful concepts and avoid exploiting spurious correlations. We illustrate a previously unrecognized limitation of the popular neural network explanation method Grad-CAM. We propose HiResCAM, a class-specific explanation method that is guaranteed to highlight only the locations the model used to make each prediction.
  arXiv  Detail & Related papers  (2020-11-17T19:26:14Z)
• Eigen-CAM: Class Activation Map using Principal Components [1.2691047660244335]
  This paper builds on previous ideas to cope with the increasing demand for interpretable, robust, and transparent models. The proposed Eigen-CAM computes and visualizes the principle components of the learned features/representations from the convolutional layers.
  arXiv  Detail & Related papers  (2020-08-01T17:14:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.