检索结果-内蒙古大学图书馆

28th International Conference on Computational Linguistics, COLING 2020

作者： Liu, Junhao Shou, Linjun Pei, Jian Gong, Ming Yang, Min Jiang, Daxin Shenzhen Key Laboratory for High Performance Data Mining Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China Microsoft STCA NLP Group Beijing China School of Computing Science Simon Fraser University Canada

ISBN: (纸本)9781952148279

Cross-lingual Machine Reading Comprehension (CLMRC) remains a challenging problem due to the lack of large-scale annotated datasets in low-source languages, such as Arabic, Hindi, and Vietnamese. Many previous approaches use translation data by translating from a rich-source language, such as English, to low-source languages as auxiliary supervision. However, how to effectively leverage translation data and reduce the impact of noise introduced by translation remains onerous. In this paper, we tackle this challenge and enhance the cross-lingual transferring performance by a novel augmentation approach named Language Branch Machine Reading Comprehension (LBMRC). A language branch is a group of passages in one single language paired with questions in all target languages. We train multiple machine reading comprehension (MRC) models proficient in individual language based on LBMRC. Then, we devise a multilingual distillation approach to amalgamate knowledge from multiple language branch models to a single model for all target languages. Combining the LBMRC and multilingual distillation can be more robust to the data noises, therefore, improving the model’s cross-lingual ability. Meanwhile, the produced single multilingual model can apply to all target languages, which saves the cost of training, inference, and maintenance for multiple models. Extensive experiments on two CLMRC benchmarks clearly show the effectiveness of our proposed method. © 2020 COLING 2020 - 28th International Conference on Computational Linguistics, Proceedings of the Conference. All rights reserved.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

Dual-View Pyramid Pooling in Deep Neural Networks for Improved Medical Image Classification and Confidence Calibration

arXiv

引用

arXiv 2024年

作者： Zhang, Xiaoqing Nie, Qiushi Xiao, Zunjie Zhao, Jilu Wu, Xiao Guo, Pengxin Li, Runzhi Liu, Jin Wei, Yanjie Pan, Yi Center for High Performance Computing Shenzhen Key Laboratory of Intelligent Bioinformatics Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Guangdong Shenzhen518055 China Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen518055 China Department of Statistics and Actuarial Science The University of Hong Kong 999077 Hong Kong Cooperative Innovation Center of Internet Healthcare Zhengzhou University Zhengzhou450001 China Hunan Provincial Key Lab on Bioinformatics School of Computer Science and Engineering Central South University Changsha410083 China Xinjiang Engineering Research Center of Big Data and Intelligent Software School of software Xinjiang University Wulumuqi830046 China Faculty of Computer Science and Control Engineering Shenzhen University of Advanced Technology Shenzhen China

Spatial pooling (SP) and cross-channel pooling (CCP) operators have been applied to aggregate spatial features and pixel-wise features from feature maps in deep neural networks (DNNs), respectively. Their main goal is to reduce computation and memory overhead without visibly weakening the performance of DNNs. However, SP often faces the problem of losing the subtle feature representations, while CCP has a high possibility of ignoring salient feature representations, which may lead to both miscalibration of confidence issues and suboptimal medical classification results. To address these problems, we propose a novel dual-view framework, the first to systematically investigate the relative roles of SP and CCP by analyzing the difference between spatial features and pixel-wise features. Based on this framework, we propose a new pooling method, termed dual-view pyramid pooling (DVPP), to aggregate multi-scale dual-view features. DVPP aims to boost both medical image classification and confidence calibration performance by fully leveraging the merits of SP and CCP operators from a dual-axis perspective. Additionally, we discuss how to fulfill DVPP with five parameter-free implementations. Extensive experiments on six 2D/3D medical image classification tasks show that our DVPP surpasses state-of-the-art pooling methods in terms of medical image classification results and confidence calibration across different DNNs. © 2024, CC BY.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Redshift-Space Distortions in Massive Neutrinos Cosmologies

arXiv

引用

arXiv 2025年

作者： Verdiani, Francesco Bellini, Emilio Moretti, Chiara Sefusatti, Emiliano Carbone, Carmelita Viel, Matteo SISSA - International School for Advanced Studies Via Bonomea 265 Trieste34136 Italy Istituto Nazionale di Astrofisica Osservatorio Astronomico di Trieste via Tiepolo 11 Trieste34143 Italy Institute for Fundamental Physics of the Universe Via Beirut 2 Trieste34151 Italy Istituto Nazionale di Fisica Nucleare Sezione di Trieste via Valerio 2 Trieste34127 Italy Center for Astrophysics and Cosmology University of Nova Gorica Nova Gorica Slovenia Istituto Nazionale di Astrofisica Istituto di Astrofisica Spaziale e Fisica cosmica di Milano Via Alfonso Corti 12 Milano20133 Italy ICSC - Centro Nazionale di Ricerca in High Performance Computing Big Data e Quantum Computing Via Magnanelli 2 Bologna Italy

We study the Kaiser formula for biased tracers in massive neutrino cosmologies by comparing its predictions with a large set of N-body simulations. In particular, we examine the ambiguity in the definition of the peculiar velocity contribution at linear level, whether it should be expressed in terms of the total matter velocity field or only the cold-matter component, comprised of cold dark matter and baryons. We revisit and extend previous qualitative studies on the topic with larger statistics and including a full fit of halo power spectrum and halo-matter cross-power spectrum measurements. We find clear evidence that the clustering of halos in redshift space is correctly described assuming that halo velocity is tracing the velocity field of cold-matter. The opposite assumption provides a worse fit to the simulated data and can lead to a spurious running of nuisance parameters in the perturbative galaxy power spectrum model. © 2025, CC BY.

关键词： Neutrons

来源：评论

学校读者我要写书评

暂无评论

Spintronic Terahertz Polarization Programmable System for Information Encoding

SSRN

引用

SSRN 2023年

作者： Li, Hongqing Jiang, Yunqing Zhang, Xiaoqiang Zhang, Fan Xiao, Yongguang Tang, Minghua Zhao, Zhenyang Xu, Yong Liu, Fengguang Zhao, Weisheng School of Materials Science and Engineering Xiangtan University Hunan Xiangtan411105 China School of Integrated Circuit Science and Engineering Hefei Innovation Research Insititute Beihang University Beijing100191 China Anhui High Reliability Chips Engineering Laboratory Hefei230013 China Shandong Dongyi Photoelectric Instruments Co. Ltd Yantai264670 China Shandong Dongyi Institute of Optoelectronic Technology for Industry Yantai264670 China Beijing Advanced Innovation Center for Big Data and Brain Computing China

Terahertz (THz) wave as an important electromagnetic wave has been increasingly applied in wireless communication and imaging sensing technology. It has been an increasingly practical application. Ultra-high data rates in communication technology push toward a broadband, controllable polarization, and multiple transmission channels signal sources. Here, we propose a programmable polarization spintronics terahertz source (PP-STS), in which contains a controllable polarization spintronic THz emitter and a microcontroller unit. The PP-STS not only can generate THz waves with wide frequency bandwidth, but also has the function of switching the orthogonal linear polarization state of the generated THz waves flexibly. By coding the microcontroller unit, the designed PP-SPS can directly program the orthogonal linear polarization states as a series of information to transmit. To achieve the interconversions between digital codes and the THz polarization information, an optimum matching THz polarization beam splitter (THz-PBS) is also designed via simulations and experiments. The THz-PBS, as a decoder, can accurately split the transverse electric (TE) and transverse magnetic (TM) modes into two THz detectors. At last, a coding and decoding prototype system of spintronic THz polarization is built by integrating the PP-STS, the THz-PBS and two THz detectors. The send and receive information with high speed and accuracy are verified in this prototype system as well. We believe that our work takes THz applications for multiplexing communications one step further and sheds light on commercializing the PP-STS with exciting manufacturing lines for efficient and economical THz polarization technology. © 2023, The Authors. All rights reserved.

关键词： Microcontrollers

来源：评论

学校读者我要写书评

暂无评论

Spintronic Terahertz Polarization Programmable System for Information Encoding

SSRN

引用

SSRN 2023年

关键词： Microcontrollers

来源：评论

学校读者我要写书评

暂无评论

Spintronic Terahertz Polarization Programmable System for Information Encoding

SSRN

引用

SSRN 2023年

关键词： Microcontrollers

来源：评论

学校读者我要写书评

暂无评论

Spintronic Terahertz Polarization Programmable System for Information Encoding

SSRN

引用

SSRN 2023年

作者： Li, Hongqing Jiang, Yunqing Zhang, Xiaoqiang Zhang, Fan Xiao, Yongguang Tang, Minghua Zhao, Zhenyang Xu, Yong Liu, Fengguang Zhao, Weisheng Beijing Advanced Innovation Center for Big Data and Brain Computing China School of Materials Science and Engineering Xiangtan University Hunan Xiangtan411105 China School of Integrated Circuit Science and Engineering Hefei Innovation Research Insititute Beihang University Beijing100191 China Anhui High Reliability Chips Engineering Laboratory Hefei230013 China Shandong Dongyi Photoelectric Instruments Co. Ltd Yantai264670 China Shandong Dongyi Institute of Optoelectronic Technology for Industry Yantai264670 China