检索结果-内蒙古大学图书馆

Suppress Content Shift: Better Diffusion Features via Off-the-Shelf Generation Techniques

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Meng, Benyuan Xu, Qianqian Wang, Zitai Yang, Zhiyong Cao, Xiaochun Huang, Qingming Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS China Peng Cheng Laboratory China School of Computer Science and Tech. University of Chinese Academy of Sciences China Key Laboratory of Big Data Mining and Knowledge Management CAS China School of Cyber Science and Tech. Sun Yat-sen University Shenzhen Campus China

Diffusion models are powerful generative models, and this capability can also be applied to discrimination. The inner activations of a pre-trained diffusion model can serve as features for discriminative tasks, namely, diffusion feature. We discover that diffusion feature has been hindered by a hidden yet universal phenomenon that we call content shift. To be specific, there are content differences between features and the input image, such as the exact shape of a certain object. We locate the cause of content shift as one inherent characteristic of diffusion models, which suggests the broad existence of this phenomenon in diffusion feature. Further empirical study also indicates that its negative impact is not negligible even when content shift is not visually perceivable. Hence, we propose to suppress content shift to enhance the overall quality of diffusion features. Specifically, content shift is related to the information drift during the process of recovering an image from the noisy input, pointing out the possibility of turning off-the-shelf generation techniques into tools for content shift suppression. We further propose a practical guideline named GATE to efficiently evaluate the potential benefit of a technique and provide an implementation of our methodology. Despite the simplicity, the proposed approach has achieved superior results on various tasks and datasets, validating its potential as a generic booster for diffusion features. Our code is availab.e at this url. © 2024, CC BY.

关键词：

The Minority Matters: A Diversity-Promoting Collab.rative Metric Learning Algorithm

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Bao, Shilong Xu, Qianqian Yang, Zhiyong He, Yuan Cao, Xiaochun Huang, Qingming State Key Laboratory of Information Security Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS China School of Computer Science and Tech. University of Chinese Academy of Sciences China Alibaba Group China School of Cyber Science and Technology Shenzhen Campus Sun Yat-sen University China Key Laboratory of Big Data Mining and Knowledge Management CAS China Peng Cheng Laboratory China

Collab.rative Metric Learning (CML) has recently emerged as a popular method in recommendation systems (RS), closing the gap between metric learning and Collab.rative Filtering. Following the convention of RS, existing methods exploit unique user representation in their model design. This paper focuses on a challenging scenario where a user has multiple categories of interests. Under this setting, we argue that the unique user representation might induce preference bias, especially when the item category distribution is imbalanced. To address this issue, we propose a novel method called Diversity-Promoting Collab.rative Metric Learning (DPCML), with the hope of considering the commonly ignored minority interest of the user. The key idea behind DPCML is to include a multiple set of representations for each user in the system. Based on this embedding paradigm, user preference toward an item is aggregated from different embeddings by taking the minimum item-user distance among the user embedding set. Furthermore, we observe that the diversity of the embeddings for the same user also plays an essential role in the model. To this end, we propose a diversity control regularization term to accommodate the multi-vector representation strategy better. Theoretically, we show that DPCML could generalize well to unseen test data by tackling the challenge of the annoying operation that comes from the minimum value. Experiments over a range of benchmark datasets speak to the efficacy of DPCML. © 2022, CC BY-NC-SA.

关键词： Embeddings

DRAUC: An Instance-wise Distributionally Robust AUC Optimization Framework

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Dai, Siran Xu, Qianqian Yang, Zhiyong Cao, Xiaochun Huang, Qingming SKLOIS Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Tech. CAS China School of Computer Science and Tech. University of Chinese Academy of Sciences China School of Cyber Science and Tech. Shenzhen Campus of Sun Yat-sen University China BDKM University of Chinese Academy of Sciences China

The Area Under the ROC Curve (AUC) is a widely employed metric in long-tailed classification scenarios. Nevertheless, most existing methods primarily assume that training and testing examples are drawn i.i.d. from the same distribution, which is often unachievable in practice. Distributionally Robust Optimization (DRO) enhances model performance by optimizing it for the local worst-case scenario, but directly integrating AUC optimization with DRO results in an intractable optimization problem. To tackle this challenge, methodically we propose an instance-wise surrogate loss of Distributionally Robust AUC (DRAUC) and build our optimization framework on top of it. Moreover, we highlight that conventional DRAUC may induce lab.l bias, hence introducing distribution-aware DRAUC as a more suitable metric for robust AUC learning. Theoretically, we affirm that the generalization gap between the training loss and testing error diminishes if the training set is sufficiently large. Empirically, experiments on corrupted benchmark datasets demonstrate the effectiveness of our proposed method. Code is availab.e at: https://***/EldercatSAM/DRAUC. © 2023, CC BY.

关键词： Optimization

A Novel Tri-band Frequency Selective Surface

学校读者我要写书评

暂无评论

Journal of Physics: Conference Series 2020年第1期1651卷

作者： Shenbing Wu Mingsheng Chen Chao Wang Xianliang Wu Key Lab of Ministry of Education of Intelligent Computing & Signal Processing Anhui University Hefei Anhui 230601 China School of Electronic Engineering Huainan Normal University Huainan Anhui 232001 China

This paper presents a novel tri-band frequency miniaturized frequency selection surface (FSS). The proposed FSS consists of single-node folding and single-node folding bending. The single-layer FSS exhibits three stop bands centered at 2.45, 3.61 and 5.79 GHz, with bandwidths of 0.42, 0.48 and 0.70 GHz, respectively. The proposed FSS has better miniaturization characteristics, and its cell size is 0.126 λ×0.126 λ. The three stop band frequencies of the FSS can be independently controlled by simply changing the length of the corresponding branch. In addition, the proposed FSS is symmetrical about the center and exhibits good resonance stability in both TE and TM polarization. The prototype of the proposed FSS was fabricated and tested, compared with the simulation results, the FSS showed stable performance. This method provides ideas for designing multiband FSS in the future.

关键词：

A Unified Generalization Analysis of Re-Weighting and Logit-Adjustment for Imbalanced Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wang, Zitai Xu, Qianqian Yang, Zhiyong He, Yuan Cao, Xiaochun Huang, Qingming SKLOIS Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Tech. CAS China School of Computer Science and Tech. University of Chinese Academy of Sciences China Alibaba Group China School of Cyber Science and Tech. Shenzhen Campus Sun Yat-sen University China BDKM University of Chinese Academy of Sciences China

Real-world datasets are typically imbalanced in the sense that only a few classes have numerous samples, while many classes are associated with only a few samples. As a result, a naïve ERM learning process will be biased towards the majority classes, making it difficult to generalize to the minority classes. To address this issue, one simple but effective approach is to modify the loss function to emphasize the learning on minority classes, such as re-weighting the losses or adjusting the logits via class-dependent terms. However, existing generalization analysis of such losses is still coarse-grained and fragmented, failing to explain some empirical results. To bridge this gap, we propose a novel technique named data-dependent contraction to capture how these modified losses handle different classes. On top of this technique, a fine-grained generalization bound is established for imbalanced learning, which helps reveal the mystery of re-weighting and logit-adjustment in a unified manner. Furthermore, a principled learning algorithm is developed based on the theoretical insights. Finally, the empirical results on benchmark datasets not only validate the theoretical results but also demonstrate the effectiveness of the proposed method. © 2023, CC BY.

关键词： Learning algorithms

Generalized Face Forgery Detection via Adaptive Learning for Pre-trained Vision Transformer

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Luo, Anwei Cai, Rizhao Kong, Chenqi Ju, Yakun Kang, Xiangui Huang, Jiwu Kot, Alex C. The School of Information Technology Jiangxi University of Finance and Economics Nanchang330013 China The School of Computer Science and Engineering Sun Yat-Sen University Guangzhou510006 China Lab. School of Electrical and Electronic Engineering Nanyang Technology University Singapore The Guangdong Key Laboratory of Intelligent Information Processing National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Shenzhen518060 China The China-Singapore International Joint Research Institute Singapore

With the rapid progress of generative models, the current challenge in face forgery detection is how to effectively detect realistic manipulated faces from different unseen domains. Though previous studies show that pre-trained Vision Transformer (ViT) based models can achieve some promising results after fully fine-tuning on the Deepfake dataset, their generalization performances are still unsatisfactory. One possible reason is that fully fine-tuned ViT-based models may disrupt the pre-trained features [1], [2] and overfit to some data-specific patterns [3]. To alleviate this issue, we present a Forgery-aware Adaptive Vision Transformer (FA-ViT) under the adaptive learning paradigm, where the parameters in the pre-trained ViT are kept fixed while the designed adaptive modules are optimized to capture forgery features. Specifically, a global adaptive module is designed to model long-range interactions among input tokens, which takes advantage of self-attention mechanism to mine global forgery clues. To further explore essential local forgery clues, a local adaptive module is proposed to expose local inconsistencies by enhancing the local contextual association. In addition, we introduce a fine-grained adaptive learning module that emphasizes the common compact representation of genuine faces through relationship learning in fine-grained pairs, driving these proposed adaptive modules to be aware of fine-grained forgery-aware information. Extensive experiments demonstrate that our FA-ViT achieves state-of-the-arts results in the cross-dataset evaluation, and enhances the robustness against unseen perturbations. Particularly, FA-ViT achieves 93.83% and 78.32% AUC scores on Celeb-DF and DFDC datasets in the cross-dataset evaluation. The code and trained model have been released at: https://***/LoveSiameseCat/FAViT. Copyright © 2023, The Authors. All rights reserved.

关键词： Contrastive Learning

Bilingual Word Embedding with Sentence Combination CNN for 1-to-N Sentence Alignment 20

学校读者我要写书评

暂无评论

Bilingual Word Embedding with Sentence Combination CNN for 1...

Proceedings of the 4th International Conference on Natural Language processing and Information Retrieval

作者： Xinyuan Ren Xiangling Fu Xuesi Zhou Chunsheng Liu Songfeng Gao Lei Peng School of Software Engineering Beijing University of Posts and Telecommunications Beijing China and Key Laboratory of Trustworthy Distributed Computing and Service Ministry of Education Beijing University of Posts and Telecommunications Beijing China Department of Electronic Engineering Tsinghua University Hai Dian Beijing China and Multimedia Signal and Intelligent Information Processing Lab Tsinghua University Hai Dian Beijing China BUPT and Huarong Joint lab of Smart Finance Beijing P. R. China and HuaRong RongTong (Beijing) Technology Co. Ltd Beijing P. R. China

ISBN: (纸本)9781450377607

Sentence alignment, as one of the most active and fundamental tasks in the field of natural language processing (NLP), is usually realized in two categories of methods. One is traditional methods which are firstly proposed, the other, which are adopted later, is based on the Neural Network method. Presently, under the limitation that the existing mainstream data corpora are mostly in the form of 1-to-1, the alignment models with relatively good performance mainly apply to the cases of 1-to-1 sentence alignment. However, under the circumstance that a sentence contains too much information, 1-to-N sentence alignment can actually have a better effect on sentence translation tasks, compared with the 1-to-1 form, since it is more flexible and can reduce the complexity of the original sentence. As a result, we attempt to exploit neural networks with relatively good performance in the cases of 1-to-1 to fit in the cases of 1-to-N. In this paper, a novel 1-N Bilingual word Embedding with Sentence Combination CNN Improved Framework (1-NBESCC) is proposed in order to align 1-to-N sentences more precisely. Experiments show that our proposed model performs as good as the traditional methods such as BLEUALIGN in 1-to-1 situation, but much better in 1-to-N situation.

关键词： Convolutional Neural Network

A Low SAR, Small size, High gain, Circularly polarized Antenna for Wearable Application

学校读者我要写书评

暂无评论

Journal of Physics: Conference Series 2020年第1期1651卷

作者： Chao Wang Liang Zhang Yi Cheng XueMei Li Xianliang Wu Key Lab of Ministry of Education of Intelligent Computing & Signal Processing Anhui University Hefei Anhui 230601 China AnHui Province Key Laboratory of simulation and design for Electronic information system Hefei Normal university Hefei Anhui 230601 China

A planar, small size, high gain, low specific absorption rate (SAR) and circularly-polarized (CP) wearable antenna is proposed in this paper. The antenna is fabricated on a substrate with a dielectric constant of 3.5 with a thickness of 0.5 mm and sized at 42×42 mm only. Both the simulation results show that the overlap bandwidth of impedance and axial ratio of the proposed antenna is 13.7% (5.15-5.91 GHz). To isolate the antenna against the influence of the human body, a artificial magnetic conductor (AMC) plane is added on the reverse side of the radiator. The AMC structure effectively isolates the human body from the radiation of the antenna and reduces the specific absorption rate (SAR) of it by more than 96.1% at 5.5 GHz, which improves the antenna gain and the peak gain reaches 5.09 dBi at 5.5 GHz. The wearable performance of the antenna was analyzed, and it is observed that it can sustain good performance even under human body loading. Beside, the antenna has a small size, which makes it ideal for wearable applications.

关键词：

OpenAUC: Towards AUC-Oriented Open-Set Recognition

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wang, Zitai Xu, Qianqian Yang, Zhiyong He, Yuan Cao, Xiaochun Huang, Qingming SKLOIS Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China Key Lab. of Intelligent Information Processing Institute of Computing Tech. CAS China School of Computer Science and Tech. University of Chinese Academy of Sciences China Alibaba Group China School of Cyber Science and Tech. Sun Yat-sen University Shenzhen Campus China BDKM University of Chinese Academy of Sciences China Peng Cheng Laboratory China

Traditional machine learning follows a close-set assumption that the training and test set share the same lab.l space. While in many practical scenarios, it is inevitable that some test samples belong to unknown classes (open-set). To fix this issue, Open-Set Recognition (OSR), whose goal is to make correct predictions on both close-set samples and open-set samples, has attracted rising attention. In this direction, the vast majority of literature focuses on the pattern of open-set samples. However, how to evaluate model performance in this challenging task is still unsolved. In this paper, a systematic analysis reveals that most existing metrics are essentially inconsistent with the aforementioned goal of OSR: (1) For metrics extended from close-set classification, such as Open-set F-score, Youden’s index, and Normalized Accuracy, a poor open-set prediction can escape from a low performance score with a superior close-set prediction. (2) Novelty detection AUC, which measures the ranking performance between close-set and open-set samples, ignores the close-set performance. To fix these issues, we propose a novel metric named OpenAUC. Compared with existing metrics, OpenAUC enjoys a concise pairwise formulation that evaluates open-set performance and close-set performance in a coupling manner. Further analysis shows that OpenAUC is free from the aforementioned inconsistency properties. Finally, an end-to-end learning method is proposed to minimize the OpenAUC risk, and the experimental results on popular benchmark datasets speak to its effectiveness. Project Page: https://***/wang22ti/OpenAUC. © 2022, CC BY.

关键词： Forecasting