In this paper, we propose a defence strategy to improves adversarial robustness incorporating hidden layer representation. The key of this defence strategy aims to compress or filter input’s information including adv...
详细信息
This paper shows a Min-Max property existing in the connection weights of the convolutional layers in a neural network structure, i.e., the LeNet. Specifically, the Min-Max property means that, during the back propaga...
详细信息
With the rapid development of artificial intelligence (AI) in medical image processing, deep learning in color fundus photography (CFP) analysis is also evolving. Although there are some open-source, lab.led datasets ...
详细信息
Generalized zero-shot learning (GZSL) aims to train a model for classifying data samples under the condition that some output classes are unknown during supervised learning. To address this challenging task, GZSL leve...
详细信息
Person search is a challenging task due to the different requirements of annotations between person detection and Re-identification. In general, person search methods use the supervised person Re-identification method...
详细信息
Person search is a challenging task due to the different requirements of annotations between person detection and Re-identification. In general, person search methods use the supervised person Re-identification methods, where abundant identity lab.ls of the bounding boxes are essential. However, most person images are unlab.led in the real-world scenario and it is unpractical to annotate the abundant fine-grained lab.ls for unlab.led images. Obviously, the existing supervised methods are not appropriate with the real-world scenario. Therefore, we propose an unsupervised learning method for person search in this paper, which contacts two parts: one is unsupervised person detection and the other is unsupervised person Re-identification. The experimental results on two well-known datasets, CUHK-SYSU and PRW, indicate that proposed method achieves competitive performance than the state-of-art unsupervised methods. Note that proposed method has greater practical significance even though it does not get the results as good as the general supervised methods.
作者：
Salim RezvaniXizhao WangBig Data Institute
College of Computer Science and Software Engineering Guangdong Key Lab. of Intelligent Information Processing Shenzhen University Shenzhen 518060 Guangdong China
In this note, we show that the calculation of statistics X F 2 and F F in sections 4.5 and 4.6 of the paper (Fan et al., 2017) is not correct. Also, based on the calculation of Critical Difference (CD) of Bonferr...
详细信息
In this note, we show that the calculation of statistics X F 2 and F F in sections 4.5 and 4.6 of the paper (Fan et al., 2017) is not correct. Also, based on the calculation of Critical Difference (CD) of Bonferroni–Dunn test, we show that the conclusion on ”significantly outperforming the compared algorithms”, drawn by the authors using the P -value test, is not convincing.
Image edge detection is an important basis for image recognition extraction. The traditional segmentation algorithm can't effectively extract important edge information of digital image. In view of the low contras...
详细信息
—Uncertainty quantification (UQ) plays a pivotal role in the reduction of uncertainties during both optimization and decision making, applied to solve a variety of real-world applications in science and engineering. ...
详细信息
Feature selection has become a key challenge in machine learning with the rapid growth of data size in real-world applications. However, existing feature selection methods mainly focus on numeric data, which will lead...
详细信息
In the context of online class-incremental continual learning (CIL), adapting to lab.l noise becomes paramount for model success in evolving domains. While some continual learning (CL) methods have begun to address no...
详细信息
In the context of online class-incremental continual learning (CIL), adapting to lab.l noise becomes paramount for model success in evolving domains. While some continual learning (CL) methods have begun to address noisy data streams, most assume that the noise strictly belongs to closed-set noise-i.e., they follow the assumption that noise in the current task originates classes within the same task. This assumption is clearly unrealistic in real-world scenarios. In this paper, we first formulate and analyze the concepts of closed-set and open-set noise, showing that both types can introduce unseen classes for the current training classifier. Then, to effectively handle noisy lab.ls and unknown classes, we present an innovative replay-based method Prototypes as Anchors (PAA), which learns representative and discriminative prototypes for each class, and conducts a similarity-based denoising schema in the representation space to distinguish and eliminate the negative impact of unseen classes. By implementing a dual-classifier architecture, PAA conducts consistency checks between the classifiers to ensure robustness. Extensive experimental results on diverse datasets demonstrate a significant improvement in model performance and robustness compared to existing approaches, offering a promising avenue for continual learning in dynamic, real-world environments.
暂无评论