检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Xu, Yi Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University

Pre-trained language models (PrLM) has been shown powerful in enhancing a broad range of downstream tasks including various dialogue related ones. However, PrLMs are usually trained on general plain text with common language model (LM) training objectives, which cannot sufficiently capture dialogue exclusive features due to the limitation of such training setting, so that there is an immediate need to fill the gap between a specific dialogue task and the LM task. As it is unlikely to collect huge dialogue data for dialogue-oriented pre-training, in this paper, we propose three strategies to simulate the conversation features on general plain text. Our proposed method differs from existing post-training methods that it may yield a general-purpose PrLM and does not individualize to any detailed task while keeping the capability of learning dialogue related features including speaker awareness, continuity and consistency. The resulted Dialog-PrLM is fine-tuned on three public multi-turn dialogue datasets and helps achieve significant and consistent improvement over the plain PrLMs. Copyright © 2021, The Authors. All rights reserved.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

DVP: Data visualization platform

arXiv

引用

arXiv 2019年

作者： Yousefa, Waleed A. Abouelkahire, Ahmed A. Marzouk, Omar S. Mohame, Sameh K Alaggan, Mohammad Human Computer Interaction Laboratory HCI Lab Computer Science Department Faculty of Computers and Information Helwan University

We identify two major steps in data analysis, data exploration for understanding and observing patterns/relationships in data;and construction, design and assessment of various models to formalize these relationships. For each step, there exists a large set of tools and software. For the first step, many visualization tools exist, such as, GGobi, Parallax, and Crystal Vision, and most recently tableau and plottly. For the second step, many Scientific Computing Environments (SCEs) exist, such as, Matlab, Mathematica, R and Python. However, there does not exist a tool which allows for seamless two-way interaction between visualization tools and SCEs. We have designed and implemented a data visualization platform (DVP) with an architecture and design that attempts to bridge this gap. DVP connects seamlessly to SCEs to bring the computational capabilities to the visualization methods in a single coherent platform. DVP is designed with two interfaces, the desktop stand alone version and the online interface. To illustrate the power of DVP design, a free demo for the online interface of DVP is available [1] and very low-level design details are explained in this article. Since DVP was launched, circa 2012, the present manuscript was not published since today for commercialization and patent considerations. Copyright © 2019, The Authors. All rights reserved.

关键词： Data visualization

来源：评论

学校读者我要写书评

暂无评论

Multilingual pre-training with universal dependency learning 21

Multilingual pre-training with universal dependency learning

引用

Proceedings of the 35th International Conference on Neural Information Processing Systems

作者： Kailai Sun Zuchao Li Hai Zhao Department of Computer Science and Engineering Shanghai Jiao Tong University Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University

ISBN: (纸本)9781713845393

The pre-trained language model (PrLM) demonstrates domination in downstream natural language processing tasks, in which multilingual PrLM takes advantage of language universality to alleviate the issue of limited resources for low-resource languages. Despite its successes, the performance of multilingual PrLM is still unsatisfactory, when multilingual PrLMs only focus on plain text and ignore obvious universal linguistic structure clues. Existing PrLMs have shown that monolingual linguistic structure knowledge may bring about better performance. Thus we propose a novel multilingual PrLM that supports both explicit universal dependency parsing and implicit language modeling. Syntax in terms of universal dependency parse serves as not only pre-training objective but also learned representation in our model, which brings unprecedented PrLM interpretability and convenience in downstream task use. Our model outperforms two popular multilingual PrLM, multilingual-BERT and XLM-R, on cross-lingual natural language understanding (NLU) benchmarks and linguistic structure parsing datasets, demonstrating the effectiveness and stronger cross-lingual modeling capabilities of our approach.

关键词：

来源：评论

学校读者我要写书评

暂无评论

DATA-DEPENDENT GAUSSIAN PRIOR OBJECTIVE FOR LANGUAGE GENERATION 8

DATA-DEPENDENT GAUSSIAN PRIOR OBJECTIVE FOR LANGUAGE GENERAT...

引用

8th International Conference on Learning Representations, ICLR 2020

作者： Li, Zuchao Wang, Rui Chen, Kehai Utiyama, Masao Sumita, Eiichiro Zhang, Zhuosheng Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University China Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University China Kyoto Japan

For typical sequence prediction problems such as language generation, maximum likelihood estimation (MLE) has commonly been adopted as it encourages the predicted sequence most consistent with the ground-truth sequence to have the highest probability of occurring. However, MLE focuses on once-to-all matching between the predicted sequence and gold-standard, consequently treating all incorrect predictions as being equally incorrect. We refer to this drawback as negative diversity ignorance in this paper. Treating all incorrect predictions as equal unfairly downplays the nuance of these sequences' detailed token-wise structure. To counteract this, we augment the MLE loss by introducing an extra Kullback-Leibler divergence term derived by comparing a data-dependent Gaussian prior and the detailed training prediction. The proposed data-dependent Gaussian prior objective (D2GPo) is defined over a prior topological order of tokens and is poles apart from the data-independent Gaussian prior (L2 regularization) commonly adopted in smoothing the training of MLE. Experimental results show that the proposed method makes effective use of a more detailed prior in the data and has improved performance in typical language generation tasks, including supervised and unsupervised machine translation, text summarization, storytelling, and image captioning. © 2020 8th International Conference on Learning Representations, ICLR 2020. All rights reserved.

关键词： Maximum likelihood estimation

来源：评论

学校读者我要写书评

暂无评论

Structural pre-training for dialogue comprehension

arXiv

引用

arXiv 2021年

作者： Zhang, Zhuosheng Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University

Pre-trained language models (PrLMs) have demonstrated superior performance due to their strong ability to learn universal language representations from self-supervised pre-training. However, even with the help of the powerful PrLMs, it is still challenging to effectively capture task-related knowledge from dialogue texts which are enriched by correlations among speaker-aware utterances. In this work, we present SPIDER, Structural Pre-traIned DialoguE Reader, to capture dialogue exclusive features. To simulate the dialogue-like features, we propose two training objectives in addition to the original LM objectives: 1) utterance order restoration, which predicts the order of the permuted utterances in dialogue context;2) sentence backbone regularization, which regularizes the model to improve the factual correctness of summarized subject-verb-object triplets. Experimental results on widely used dialogue benchmarks verify the effectiveness of the newly introduced self-supervised tasks. Copyright © 2021, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Fully implantable and battery-free wireless optoelectronic system for modulable cancer therapy and real-time monitoring

引用

npj Flexible Electronics 2023年第1期7卷 119-130页

作者： Kiho Kim In Sik Min Tae Hee Kim Do Hyeon Kim Seungwon Hwang Kyowon Kang Kyubeen Kim Sangun Park Jongmin Lee Young Uk Cho Jung Woo Lee Woon-Hong Yeo Young Min Song Youngmee Jung Ki Jun Yu Functional Bio-integrated Electronics and Energy Management Lab School of Electrical and Electronic EngineeringYonsei University50 Yonsei-roSeodaemun-guSeoul 03722Republic of Korea Center for Biomaterials Biomedical Research InstituteKorea Institute of Science and Technology(KIST)Seoul 02792Republic of Korea Department of Fusion Research and Collaboration Biomedical Research InstituteSeoul National University HospitalSeoul 03080Republic of Korea School of Electrical Engineering and Computer Science(EECS) Gwangju Institute of Science and Technology(GIST)Gwangju 61005Republic of Korea PA1 Foundry BusinessSamsung Electronics1 Samsungjeonja-roHwaseong-siGyeonggi-doRepublic of Korea R&D Center DNA Lab PharmaresearchSeongnam 13486Republic of Korea KU-KIST Graduate School of Converging Science and Technology Korea UniversitySeoul 136-705Republic of Korea School of Materials Science and Engineering Energy Materials for Soft Electronics LaboratoryPusan National UniversityBusan 46241Republic of Korea IEN Center for Human-Centric Interfaces and Engineering and George W.Woodruff School of Mechanical Engineering Georgia Institute of TechnologyAtlantaGA 30332USA Wallace H.Coulter Department of Biomedical Engineering Georgia Institute of Technology and Emory University School of MedicineAtlantaGA 30332USA Institute for Robotics and Intelligent Machines Parker H.Petit Institute for Bioengineering and BiosciencesInstitute for MaterialsNeural Engineering CenterGeorgia Institute of TechnologyAtlantaGA 30332USA School of Electrical and Electronic Engineering YU-Korea Institute of Science and Technology(KIST)InstituteYonsei University50Yonsei-roSeodaemun-guSeoul 03722Republic of Korea

Photodynamic therapy(PDT)is attracting attention as a next-generation cancer treatment that can selectively destroy malignant tissues,exhibit fewer side effects,and lack pain during *** PDT systems have recently been developed to resolve the issues of bulky and expensive conventional PDT systems and to implement continuous and repetitive *** implantable PDT systems,however,are not able to perform multiple functions simultaneously,such as modulating light intensity,measuring,and transmitting tumor-related data,resulting in the complexity of cancer ***,we introduce a flexible and fully implantable wireless optoelectronic system capable of continuous and effective cancer treatment by fusing PDT and hyperthermia and enabling tumor size monitoring in *** system exploits micro inorganic light-emitting diodes(μ-LED)that emit light with a wavelength of 624 nm,designed not to affect surrounding normal tissues by utilizing a fully programmable light intensity ofμ-LED and precisely monitoring the tumor size by Si phototransistor during a long-term implantation(2–3 weeks).The superiority of simultaneous cancer treatment and tumor size monitoring capabilities of our system operated by wireless power and data transmissions with a cell phone was confirmed through in vitro experiments,ray-tracing simulation results,and a tumor xenograft mouse model in *** all-in-one single system for cancer treatment offers opportunities to not only enable effective treatment of tumors located deep in the tissue but also enable precise and continuous monitoring of tumor size in real-time.

关键词： optoelectronic enable battery

来源：评论

学校读者我要写书评

暂无评论

Comparative Methodological Approaches for Analyzing Margin of Stability During Normal Walking

Comparative Methodological Approaches for Analyzing Margin o...

引用

3D Analysis of human Movement (3DAHM), International Symposium on

作者： Magalí Sganga Federico Villagra Otar Akanyeti Emiliano Pablo Ravera Departamento de Ciencias de la Salud Laboratorio de Investigación del Movimiento Humano Universidad Maimónides Universidad Nacional de La Matanza Buenos Aires Argentina Department of Life Sciences Carwyn James Building Penglais Campus Aberystwyth University Aberystwyth Wales UK Department of Computer Science Aberystwyth University Aberystwyth Wales UK Group of Analysis Modeling Processing and Clinician Implementation of Biomechanical Signals and Systems. IBB CONICET-UNER. Human Movement Research Laboratory National University of Entre Rios Oro Verde Argentina

ISBN: (数字)9798331521738

ISBN: (纸本)9798331509842

A comprehensive evaluation of pathological gait patterns is essential for understanding the underlying pathology and its progression to deliver personalized treatments. One quantitative indicator for assessing fall risks during walking is margin of stability (MoS), a measure of instantaneous mechanical stability derived from the dynamic relationship between the extrapolated center of mass (CoM) and the center of pressure. However, extrapolating CoM during movement is not straightforward, and there is an undesired variability induced by the selected measurement protocol, which often depends on the available technology. Up until now, the extent of this variability has not been investigated systematically. This study compares four different CoM estimation methods. Based on the Conventional Gait Model, the first two methods estimate CoM using virtual sacrum (1) and pelvis centroid (2). The third method estimates CoM from ground reaction force (3), and the last method uses the accelerometer and gyroscope data from a smartphone mounted on the lower back. 21 healthy subjects (10 female, 11 male, age: 31.7 ± 9.0 years; BMI: 24.7 ± 3.5 kg m- 2 ) walked on an instrumented treadmill at a fixed speed at 1.3 m s -1 for 3 minutes. Descriptive statistical analysis was performed, including temporal and normalized gait cycle analysis, focusing on the minimum and maximum MoS, and its ranges during walking. SPMld with paired t-tests was employed for the analysis. Our results suggest that CoM estimation varies significantly depending on the selected measurement protocol and technology which should be considered while reporting and

关键词： Legged locomotion Pathology Protocols Three-dimensional displays Statistical analysis Estimation Position measurement Stability analysis Pressure measurement Pelvis

来源：评论

学校读者我要写书评

暂无评论

Weakly Supervised Visual-Auditory Fixation Prediction with Multigranularity Perception

arXiv

引用

arXiv 2021年

作者： Wang, Guotao Chen, Chenglizhao Fan, Deng-Ping Hao, Aimin Qin, Hong The State Key Laboratory of Virtual Reality Technology and Systems Beihang University China The College of Computer Science and Technology China University of Petroleum China ETH Zurich Zurich Switzerland The State Key Laboratory of Virtual Reality Technology and Systems Beihang University Research Unit of Virtual Human and Virtual Surgery Chinese Academy of Medical Sciences and with Pengcheng Laboratory China The Computer Science Department Stony Brook University United States

Thanks to the rapid advances in deep learning techniques and the wide availability of large-scale training sets, the performance of video saliency detection models has been improving steadily and significantly. However, deep learning-based visual-audio fixation prediction is still in its infancy. At present, only a few visual-audio sequences have been furnished, with real fixations being recorded in real visual-audio environments. Hence, it would be neither efficient nor necessary to recollect real fixations under the same visual-audio circumstances. To address this problem, this paper promotes a novel approach in a weakly supervised manner to alleviate the demand of large-scale training sets for visual-audio model training. By using only the video category tags, we propose the selective class activation mapping (SCAM) and its upgrade (SCAM+). In the spatial-temporal-audio circumstance, the former follows a coarse-to-fine strategy to select the most discriminative regions, and these regions are usually capable of exhibiting high consistency with the real human-eye fixations. The latter equips the SCAM with an additional multi-granularity perception mechanism, making the whole process more consistent with that of the real human visual system. Moreover, we distill knowledge from these regions to obtain complete new spatial-temporal-audio (STA) fixation prediction (FP) networks, enabling broad applications in cases where video tags are not available. Without resorting to any real human-eye fixation, the performances of these STA FP networks are comparable to those of fully supervised networks. The code and results are publicly available at https://***/guotaowang/STANet. Copyright © 2021, The Authors. All rights reserved.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

Attention Based Hierarchical Aggregation Network for 3D Left Atrial Segmentation 9th

Attention Based Hierarchical Aggregation Network for 3D Left...

引用

9th International Workshop on Statistical Atlases and Computational Models of the Heart: Atrial Segmentation and LV Quantification Challenges, STACOM 2018, held in conjunction with Medical Image Computing and computer-Assisted Intervention, MICCAI 2018

作者： Li, Caizi Tong, Qianqian Liao, Xiangyun Si, Weixin Sun, Yinzi Wang, Qiong Heng, Pheng-Ann School of Computer Science Wuhan University Wuhan430072 China Shenzhen Key Laboratory of Virtual Reality and Human Interaction Technology Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen518000 China Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong China

ISBN: (纸本)9783030120283

Atrial fibrillation (AF) is the most common type of cardiac arrhythmia. The atrial segmentation is essential for the understanding of the human atria structure which is vital to the AF treatment. In this paper, we propose a novel three-dimensional (3D) segmentation network combining hierarchical aggregation and attention mechanism for 3D left atrial segmentation, named attention based hierarchical aggregation network (HAANet). In our network, the shallow and deep feature fusion capability of encoder-decoder convolutional neural networks is enhanced through hierarchical aggregation. Besides, attention mechanism is adopted to promote the ability of extracting efficient features. Experimental results demonstrate the HAANet can produce good results for 3D left atrial segmentation and the dice score of our HAANet reaches 92.30. © 2019, Springer Nature Switzerland AG.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

Modeling a target-selection motion by leveraging an optimal feedback control mechanism

引用

science China(Information sciences) 2018年第4期61卷 209-211页

作者： Jin HUANG Xiaolan PENG Feng TIAN Hongan WANG Guozhong DAI Beijing Key Laboratory of Human-Computer Interaction Institute of Software Chinese Academy of Sciences State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences School of Computer and Control Engineering University of Chinese Academy of Sciences

Target selection has always been a popular research topic in the human-computer interaction(HCI)*** with continuous interactive spaces,such as video games,augmented reality(AR),and virtual reality(VR),are becom-

关键词： Modeling a target-selection motion by leveraging an optimal feedback control mechanism

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：