Interpretation and improvement of deep neural networks relies on better understanding of their underlying mechanisms. In particular, gradients of classes or concepts with respect to the input features (e.g., pixels in images) are often used as importance scores or estimators, which are visualized in saliency maps. Thus, a family of saliency methods provide an intuitive way to identify input features with substantial influences on classifications or latent concepts. Several modifications to conventional saliency maps, such as Rectified Gradients and Layer-wise Relevance Propagation (LRP), have been introduced to allegedly denoise and improve interpretability. While visually coherent in certain cases, Rectified Gradients and other modified saliency maps introduce a strong input bias (e.g., brightness in the RGB space) because of inappropriate uses of the input features. We demonstrate that dark areas of an input image are not highlighted by a saliency map using Rectified Gradients, even if it is relevant for the class or concept. Even in the scaled images, the input bias exists around an artificial point in color spectrum. Our modification, which simply eliminates multiplication with input features, removes this bias. This showcases how a visual criteria may not align with true explainability of deep learning models.
WaterFlow: Explicit Physics-Prior Rectified Flow for Underwater Saliency Mask Generation
Hainan University, China（海南大学）；Huazhong University of Science and Technology, China（华中科技大学）；Shenzhen University, China（深圳大学）；Lingnan University, Hong Kong（岭南大学）
Saliency Map, the gradient of the score function with respect to the input, is the most basic technique for interpreting deep neural network decisions. However, saliency maps are often visually noisy. Although several hypotheses were proposed to account for this phenomenon, there are few works that provide rigorous analyses of noisy saliency maps. In this paper, we firstly propose a new hypothesis that noise may occur in saliency maps when irrelevant features pass through ReLU activation functions. Then, we propose Rectified Gradient, a method that alleviates this problem through layer-wise th…
University of Maryland（马里兰大学）；Genentech, Inc.（基因泰克公司）
Saliency methods have been widely used to highlight important input features in model predictions. Most existing methods use backpropagation on a modified gradient function to generate saliency maps. Thus, noisy gradients can result in unfaithful feature attributions. In this paper, we tackle this issue and introduce a {\it saliency guided training}procedure for neural networks to reduce noisy gradients used in predictions while retaining the predictive performance of the model. Our saliency guided training procedure iteratively masks features with small and potentially noisy gradients while m…
ERA: Entropy-Guided Visual Token Pruning with Rectified Attention for Efficient MLLMs
School of Future Technology, Dalian University of Technology（大连理工大学未来技术学院）；S-Lab, Nanyang Technological University（南洋理工大学S-Lab）；OPPO Research Institute（OPPO研究院）；Department of Computing, The Hong Kong Polytechnic University（香港理工大学计算机系）
In this paper, we propose several novel deep learning methods for object saliency detection based on the powerful convolutional neural networks. In our approach, we use a gradient descent method to iteratively modify an input image based on the pixel-wise gradients to reduce a cost function measuring the class-specific objectness of the image. The pixel-wise gradients can be efficiently computed using the back-propagation algorithm. The discrepancy between the modified image and the original one may be used as a saliency map for the image. Moreover, we have further proposed several new trainin…
Isfahan University of Technology（伊斯法罕理工大学）；McMaster University（麦克马斯特大学）；Seattle University（西雅图大学）
Endoscopy is a valuable tool for the early diagnosis of colon cancer. However, it requires the expertise of endoscopists and is a time-consuming process. In this work, we propose a new multi-label classification method, which considers two aspects of learning approaches (local and global views) for endoscopic image classification. The model consists of a Swin transformer branch and a modified VGG16 model as a CNN branch. To help the learning process of the CNN branch, the model employs saliency maps and endoscopy images and concatenates them. The results demonstrate that this method performed …
Shanghai Jiao Tong University（上海交通大学）；MarkableAI Inc（MarkableAI公司）；Baidu Research（百度研究院）；West Virginia University（西弗吉尼亚大学）；Université de Nantes（南特大学）
Data size is the bottleneck for developing deep saliency models, because collecting eye-movement data is very time consuming and expensive. Most of current studies on human attention and saliency modeling have used high quality stereotype stimuli. In real world, however, captured images undergo various types of transformations. Can we use these transformations to augment existing saliency datasets? Here, we first create a novel saliency dataset including fixations of 10 observers over 1900 images degraded by 19 types of transformations. Second, by analyzing eye movements, we find that observer…
Isfahan University of Technology（伊斯法罕理工大学）；McMaster University（麦克马斯特大学）
Image retargeting is the task of making images capable of being displayed on screens with different sizes. This work should be done so that high-level visual information and low-level features such as texture remain as intact as possible to the human visual system, while the output image may have different dimensions. Thus, simple methods such as scaling and cropping are not adequate for this purpose. In recent years, researchers have tried to improve the existing retargeting methods and introduce new ones. However, a specific method cannot be utilized to retarget all types of images. In other…
Tianjin University（天津大学）；Key Laboratory of Artificial Intelligence, Ministry of Education（教育部人工智能重点实验室）
As moving objects always draw more attention of human eyes, the temporal motive information is always exploited complementarily with spatial information to detect salient objects in videos. Although efficient tools such as optical flow have been proposed to extract temporal motive information, it often encounters difficulties when used for saliency detection due to the movement of camera or the partial movement of salient objects. In this paper, we investigate the complimentary roles of spatial and temporal information and propose a novel dynamic spatiotemporal network (DS-Net) for more effect…
A New One-Shot Federated Learning Framework for Medical Imaging Classification with Feature-Guided Rectified Flow and Knowledge Distillation
Guangdong Provincial Key Laboratory of Ultra High Definition Immersive Media Technology（广东超高清沉浸媒体技术省重点实验室）；Shenzhen Graduate School, Peking University（北京大学深圳研究生院）
ISTI-CNR（意大利国家研究委员会信息科学与技术研究所）；University of Pisa（比萨大学）；Dalhousie University（达尔豪斯大学）
We present an approach to explain the decisions of black box models for image classification. While using the black box to label images, our explanation method exploits the latent feature space learned through an adversarial autoencoder. The proposed method first generates exemplar images in the latent feature space and learns a decision tree classifier. Then, it selects and decodes exemplars respecting local decision rules. Finally, it visualizes them in a manner that shows to the user how the exemplars can be modified to either stay within their class, or to become counter-factuals by "morph…
Department of Engineering Science, University of Oxford（牛津大学工程科学系）
We propose an end-to-end-trainable attention module for convolutional neural network (CNN) architectures built for image classification. The module takes as input the 2D feature vector maps which form the intermediate representations of the input image at different stages in the CNN pipeline, and outputs a 2D matrix of scores for each map. Standard CNN architectures are modified through the incorporation of this module, and trained under the constraint that a convex combination of the intermediate 2D feature vectors, as parameterised by the score matrices, must \textit{alone} be used for class…
