检索结果-内蒙古大学图书馆

Viterbi Decoding of Directed Acyclic Transformer for Non-Autoregressive Machine Translation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Shao, Chenze Ma, Zhengrui Feng, Yang Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China

Non-autoregressive models achieve significant decoding speedup in neural machine translation but lack the ability to capture sequential dependency. Directed Acyclic Transformer (DA-Transformer) was recently proposed to model sequential dependency with a directed acyclic graph. Consequently, it has to apply a sequential decision process at inference time, which harms the global translation accuracy. In this paper, we present a Viterbi decoding framework for DA-Transformer, which guarantees to find the joint optimal solution for the translation and decoding path under any length constraint. Experimental results demonstrate that our approach consistently improves the performance of DA-Transformer while maintaining a similar decoding speedup. Copyright © 2022, The Authors. All rights reserved.

关键词： Decoding

Optically transparent transmissive orbital angular momentum (OAM) metasurface with high transmittance and broadband characteristics

学校读者我要写书评

暂无评论

Optics Express 2025年第12期33卷 24385-24395页

作者： Yong Cai Yuying Dai Yue Ma Gang Wang Guanghui Xu Xingang Ren Hui Wang Xiaohu Zhao Xianliang Wu Information Materials and Intelligent Sensing Laboratory of Anhui Province Anhui University Hefei 230039 China Key Laboratory of Electromagnetic Environmental Sensing of Anhui Higher Education Institutes Anhui University Hefei 230039 China wang_toward@*** ghxu86@*** Key Laboratory of Intelligent Computing & Signal Processing Ministry of Education Institute of Physical Science and Information Technology Anhui University Hefei 230039 China xgren@*** Hefei Normal University Hefei 230009 China

This work proposes a high-performance broadband orbital angular momentum (OAM) beam-generating metasurface based on optically transparent media. A broadband polarization-converting transmissive metasurface, which operates in the 18 - 24 (28.57% fractional bandwidth) GHz range, has been realized by collaboratively designing dual C-split ring resonator structures and bidirectional lattice-patterned surface networks. The array structure demonstrates typical helical phase distribution features in the K-band, and experimental results verify simulation design. Innovatively employing a composite structure of quartz glass substrate and patterned conductive thin films, the design achieves compatible optimization of 77% visible-light transmittance and microwave broadband performance. This approach breaks through the mutually exclusive limitations of traditional metasurfaces’ optical/electromagnetic properties, providing a highly integrated solution for transparent antenna radomes in satellite communication systems and multi-mode beam multiplexing applications.

关键词： Orbital angular momentum multiplexing Phase plates Ring resonators Spatial light modulators Spiral phase Thin films

Improving Fake News Detection of Influential Domain via Domain- and Instance-Level Transfer

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Nan, Qiong Wang, Danding Zhu, Yongchun Sheng, Qiang Shi, Yuhui Cao, Juan Li, Jintao Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China

Both real and fake news in various domains, such as politics, health, and entertainment are spread via online social media every day, necessitating fake news detection for multiple domains. Among them, fake news in specific domains like politics and health has more serious potential negative impacts on the real world (e.g., the infodemic led by COVID-19 misinformation). Previous studies focus on multi-domain fake news detection, by equally mining and modeling the correlation between domains. However, these multi-domain methods suffer from a seesaw problem: the performance of some domains is often improved at the cost of hurting the performance of other domains, which could lead to an unsatisfying performance in specific domains. To address this issue, we propose a Domain- and Instance-level Transfer Framework for Fake News Detection (DITFEND), which could improve the performance of specific target domains. To transfer coarse-grained domain-level knowledge, we train a general model with data of all domains from the meta-learning perspective. To transfer fine-grained instance-level knowledge and adapt the general model to a target domain, we train a language model on the target domain to evaluate the transferability of each data instance in source domains and re-weigh each instance’s contribution. Offline experiments on two datasets demonstrate the effectiveness of DITFEND. Online experiments show that DITFEND brings additional improvements over the base models in a real-world scenario. Copyright © 2022, The Authors. All rights reserved.

关键词： Social networking (online)

SSE-SAM: Balancing Head and Tail Classes Gradually Through Stage-Wise SAM 39

学校读者我要写书评

暂无评论

SSE-SAM: Balancing Head and Tail Classes Gradually Through S...

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Lyu, Xingyu Xu, Qianqian Yang, Zhiyong Lyu, Shaojie Huang, Qingming 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 Tencent Corporate China BDKM University of Chinese Academy of Sciences China

ISBN: (纸本)157735897X

Real-world datasets often exhibit a long-tailed distribution, where vast majority of classes known as tail classes have only few samples. Traditional methods tend to overfit on these tail classes. Recently, a new approach called Imbalanced SAM (ImbSAM) is proposed to leverage the generalization benefits of Sharpness-Aware Minimization (SAM) for long-tailed distributions. The main strategy is to merely enhance the smoothness of the loss function for tail classes. However, we argue that improving generalization in long-tail scenarios requires a careful balance between head and tail classes. We show that neither SAM nor ImbSAM alone can fully achieve this balance. For SAM, we prove that although it enhances the model's generalization ability by escaping saddle point in the overall loss landscape, it does not effectively address this for tail-class losses. Conversely, while ImbSAM is more effective at avoiding saddle points in tail classes, the head classes are trained insufficiently, resulting in significant performance drops. Based on these insights, we propose Stage-wise Saddle Escaping SAM (SSE-SAM), which uses complementary strengths of ImbSAM and SAM in a phased approach. Initially, SSE-SAM follows the majority sample to avoid saddle points of the head-class loss. During the later phase, it focuses on tail-classes to help them escape saddle points. Our experiments confirm that SSE-SAM has better ability in escaping saddles both on head and tail classes, and shows performance improvements. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词：

EDA: Enhanced Domain-Adversarial Training for Anatomical Landmark Detection

学校读者我要写书评

暂无评论

EDA: Enhanced Domain-Adversarial Training for Anatomical Lan...

IEEE International Symposium on Biomedical Imaging

作者： Fan Yang S. Kevin Zhou School of Biomedical Engineering Division of Life Sciences and Medicine University of Science and Technology of China Hefei Anhui China Center for Medical Imaging Robotics Analytic Computing & Learning (MIRACLE) Suzhou Institute for Advanced Research University of Science and Technology of China Suzhou Jiangsu China Key Laboratory of Precision and Intelligent Chemistry USTC Hefei Anhui China Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS) Institute of Computing Technology CAS Beijing China

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

Manually annotating anatomical landmarks in medical images requires experienced clinicians and is a labor-intensive process. However, recent AI-assisted methods for landmark detection often rely on the training and test data originating from the same domain. This work introduces a novel unsupervised domain adaptation (UDA) framework aimed at anatomical landmark detection from a medical image, designed to bridge the gap between a source domain with labels and an unlabeled target domain. Specifically, we have developed a new Domain-Adversarial Network that incorporates skip connections to transfer and fuse high-resolution feature maps. Additionally, we proposed Dynamic Gaussian Learning, which allows the model to escape from local error regions. We carry out experiments on landmark detection for both head and chest, and the results demonstrate that our method achieves state-of-the-art performances in each experiment.

关键词： Training Measurement Adaptation models Three-dimensional displays Smoothing methods Head Kernel Biomedical imaging

Yolov9-Spe: A Novel Deep Learning Method For Classroom Student Behavior Detection

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Liu, Junxiu Wu, Fengxiang Lu, Baoshan Qin, Sheng Fu, Qiang Luo, Yuling Guangxi Key Lab of Brain-inspired Computing and Intelligent Chips School of Electronic and Information Engineering Guangxi Normal University Guilin541004 China Guangxi Wireless Broadband Communication and Signal Processing Key Laboratory Guilin University of Electronic Technology Guilin541004 China

Deep learning can significantly enhance student behavior detection in classrooms. However, challenges such as small target recognition, blurry data, occlusion, and multi-scale detection persist. To address these, we propose YOLOv9-SPE, an improved model based on YOLOv9, incorporating Space-to-Depth Convolution (SPD-Conv), Partial Convolution (PConv), and an Efficient Multi-Scale Attention (EMA) Module. SPD-Conv enhances small target recognition and improves feature extraction for blurry data, while PConv, adapted from FasterNet, addresses occlusion issues. EMA is integrated to handle multi-scale targets in classroom student behavior detection. Experimental results on a student behavior dataset show that YOLOv9-SPE achieves a 2.1% improvement in mean average precision over the original YOLOv9. These enhancements provide a more accurate and efficient method for monitoring and analyzing student behavior in educational settings. © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

Non-Monotonic Latent Alignments for CTC-Based Non-Autoregressive Machine Translation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Shao, Chenze Feng, Yang Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China

Non-autoregressive translation (NAT) models are typically trained with the cross-entropy loss, which forces the model outputs to be aligned verbatim with the target sentence and will highly penalize small shifts in word positions. Latent alignment models relax the explicit alignment by marginalizing out all monotonic latent alignments with the CTC loss. However, they cannot handle non-monotonic alignments, which is non-negligible as there is typically global word reordering in machine translation. In this work, we explore non-monotonic latent alignments for NAT. We extend the alignment space to non-monotonic alignments to allow for the global word reordering and further consider all alignments that overlap with the target sentence. We non-monotonically match the alignments to the target sentence and train the latent alignment model to maximize the F1 score of non-monotonic matching. Extensive experiments on major WMT benchmarks show that our method substantially improves the translation performance of CTC-based models. Our best model achieves 30.06 BLEU on WMT14 En-De with only one-iteration decoding, closing the gap between non-autoregressive and autoregressive models.2 Copyright © 2022, The Authors. All rights reserved.

关键词： Alignment

A Unified Generalization Analysis of Re-Weighting and Logit-Adjustment for Imbalanced Learning 37

学校读者我要写书评

暂无评论

A Unified Generalization Analysis of Re-Weighting and Logit-...

37th Conference on Neural information processing Systems, NeurIPS 2023

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 of Sun Yat-sen University China BDKM University of Chinese Academy of Sciences China

ISBN: (纸本)9781713899921

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 Neural information processing systems foundation. All rights reserved.

关键词： Contrastive Learning

SEMANTIC LATENT DECOMPOSITION WITH NORMALIZING FLOWS FOR FACE EDITING

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Binglei Huang, Zhizhong Shan, Hongming Zhang, Junping Shanghai Key Lab of Intelligent Information Processing School of Computer Science China Institute of Science and Technology for Brain-inspired Intelligence Fudan University Shanghai200433 China

Navigating in the latent space of StyleGAN has shown effectiveness for face editing. However, the resulting methods usually encounter challenges in complicated navigation due to the entanglement among different attributes in the latent space. To address this issue, this paper proposes a novel framework, termed SDFlow, with a semantic decomposition in original latent space using continuous conditional normalizing flows. Specifically, SDFlow decomposes the original latent code into different irrelevant variables by jointly optimizing two components: (i) a semantic encoder to estimate semantic variables from input faces and (ii) a flow-based transformation module to map the latent code into a semantic-irrelevant variable in Gaussian distribution, conditioned on the learned semantic variables. To eliminate the entanglement between variables, we employ a disentangled learning strategy under a mutual information framework, thereby providing precise manipulation controls. Experimental results demonstrate that SDFlow outperforms existing state-of-the-art face editing methods both qualitatively and quantitatively. The source code is made available at https://***/phil329/SDFlow. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics