检索结果-内蒙古大学图书馆

Attribute-Driven Feature Disentangling and Temporal Aggregation for Video Person Re-Identification

学校读者我要写书评

暂无评论

Attribute-Driven Feature Disentangling and Temporal Aggregat...

IEEE/CVF Conference on computer Vision and Pattern Recognition

作者： Yiru Zhao Xu Shen Zhongming Jin Hongtao Lu Xian-sheng Hua Key Lab of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Department of Computer Science and Engineering MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University Alibaba Damo Academy Alibaba Group

ISBN: (纸本)9781728132945

Video-based person re-identification plays an important role in surveillance video analysis, expanding image-based methods by learning features of multiple frames. Most existing methods fuse features by temporal average-pooling, without exploring the different frame weights caused by various viewpoints, poses, and occlusions. In this paper, we propose an attribute-driven method for feature disentangling and frame re-weighting. The features of single frames are disentangled into groups of sub-features, each corresponds to specific semantic attributes. The sub-features are re-weighted by the confidence of attribute recognition and then aggregated at the temporal dimension as the final representation. By means of this strategy, the most informative regions of each frame are enhanced and contributes to a more discriminative sequence representation. Extensive ablation studies verify the effectiveness of feature disentangling as well as temporal re-weighting. The experimental results on the iLIDS-VID, PRID-2011 and MARS datasets demonstrate that our proposed method outperforms existing state-of-the-art approaches.

关键词： Frame FRAMES Persons Videotapes Occlusion confidence Single frame

INCREASE APPARENT PUBLIC SPEAKING FLUENCY BY SPEECH AUGMENTATION

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Das, Sagnik Gandhi, Nisha Naik, Tejas Shilkrot, Roy Human Interaction Lab Stony Brook University Department of Computer Science Stony BrookNY United States

Fluent and confident speech is desirable to every speaker. But professional speech delivering requires a great deal of experience and practice. In this paper, we propose a speech stream manipulation system which can help non-professional speakers to produce fluent, professional-like speech content, in turn contributing towards better listener engagement and comprehension. We propose to achieve this task by manipulating the disfluencies in human speech, like the sounds uh and um, the filler words and awkward long silences. Given any unrehearsed speech we segment and silence the filled pauses and doctor the duration of imposed silence as well as other long pauses (disfluent) by a predictive model learned using professional speech dataset. Finally, we output a audio stream in which speaker sounds more fluent, confident and practiced compared to the original speech he/she recorded. According to our quantitative evaluation, we significantly increase the fluency of speech by reducing rate of pauses and fillers. Copyright © 2018, The Authors. All rights reserved.

关键词： Fillers

An incremental turn-taking model for task-oriented dialog systems

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Coman, Andrei C. Yoshino, Koichiro Murase, Yukitoshi Nakamura, Satoshi Riccardi, Giuseppe Signals and Interactive Systems Lab Department of Information Engineering and Computer Science University of Trento Italy Augmented Human Communication Laboratory Graduate School of Information Science Nara Institute of Science and Technology Japan

In a human-machine dialog scenario, deciding the appropriate time for the machine to take the turn is an open research problem. In contrast, humans engaged in conversations are able to timely decide when to interrupt the speaker for competitive or non-competitive reasons. In state-of-the-art turn-by-turn dialog systems the decision on the next dialog action is taken at the end of the utterance. In this paper, we propose a token-by-token prediction of the dialog state from incremental transcriptions of the user utterance. To identify the point of maximal understanding in an ongoing utterance, we a) implement an incremental Dialog State Tracker which is updated on a token basis (iDST) b) re-label the Dialog State Tracking Challenge 2 (DSTC2) dataset and c) adapt it to the incremental turn-taking experimental scenario. The re-labeling consists of assigning a binary value to each token in the user utterance that allows to identify the appropriate point for taking the turn. Finally, we implement an incremental Turn Taking Decider (iTTD) that is trained on these new labels for the turn-taking decision. We show that the proposed model can achieve a better performance compared to a deterministic handcrafted turn-taking algorithm. Copyright © 2019, The Authors. All rights reserved.

关键词： Recurrent neural networks

High-quality voice conversion using spectrogram-based WaveNet Vocoder 19

学校读者我要写书评

暂无评论

High-quality voice conversion using spectrogram-based WaveNe...

19th Annual Conference of the International Speech Communication, INTERSPEECH 2018

作者： Chen, Kuan Chen, Bo Lai, Jiahao Yu, Kai Key Lab. of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering SpeechLab Department of Computer Science and Engineering Brain Science and Technology Research Center Shanghai Jiao Tong University Shanghai China

Waveform generator is a key component in voice conversion. Recently, WaveNet waveform generator conditioned on the Mel-cepstrum (Mcep) has shown better quality over standard vocoder. In this paper, an enhanced WaveNet model based on spectrogram is proposed to further improve voice conversion performance. Here, Mel-frequency spectrogram is converted from source speaker to target speaker using an LSTM-RNN based frame-to-frame feature mapping. To evaluate the performance, the proposed approach is compared to an Mcep based LSTM-RNN voice conversion system. Both STRAIGHT vocoder and Mcep-based WaveNet vocoder are elected to produce the converted speech for Mcep conversion system. The fundamental frequency (F0) of the converted speech in different systems is analyzed. The naturalness, similarity and intelligibility are evaluated in subjective measures. Results show that the spectrogram based WaveNet waveform generator can achieve better voice conversion quality compared to traditional WaveNet approaches. The Mel-spectrogram based voice conversion can achieve significant improvement in speaker similarity and inherent F0 conversion. © 2018 International Speech Communication Association. All rights reserved.

关键词： Vocoders

Structured dialogue policy with graph neural networks 27

学校读者我要写书评

暂无评论

Structured dialogue policy with graph neural networks

27th International Conference on Computational Linguistics, COLING 2018

作者： Chen, Lu Tan, Bowen Long, Sishan Yu, Kai Key Lab. of Shanghai Education Commission for Intelligent Interaction and Cognitive Eng. SpeechLab Department of Computer Science and Engineering Brain Science and Technology Research Center Shanghai Jiao Tong University Shanghai China

ISBN: (纸本)9781948087506

Recently, deep reinforcement learning (DRL) has been used for dialogue policy optimization. However, many DRL-based policies are not sample-efficient. Most recent advances focus on improving DRL optimization algorithms to address this issue. Here, we take an alternative route of designing neural network structure that is better suited for DRL-based dialogue management. The proposed structured deep reinforcement learning is based on graph neural networks (GNN), which consists of some sub-networks, each one for a node on a directed graph. The graph is defined according to the domain ontology and each node can be considered as a sub-agent. During decision making, these sub-agents have internal message exchange between neighbors on the graph. We also propose an approach to jointly optimize the graph structure as well as the parameters of GNN. Experiments show that structured DRL significantly outperforms previous state-of-the-art approaches in almost all of the 18 tasks of the PyDial benchmark. © 2018 COLING 2018 - 27th International Conference on Computational Linguistics, Proceedings. All rights reserved.

关键词： Reinforcement learning

How End Users Express Conditionals in Programming by Demonstration for Mobile Apps

学校读者我要写书评

暂无评论

How End Users Express Conditionals in Programming by Demonst...

IEEE Symposium on Visual Languages and human Centric Computing (VL/HCC)

作者： Marissa Radensky Toby Jia-Jun Li Brad A. Myers Computer Science Department Amherst College Amherst MA Human-Computer Interaction Institute Carnegie Mellon University Pittsburgh PA

Though conditionals are an integral component of programming, providing an easy means of creating conditionals remains a challenge for programming-by-demonstration (PBD) systems for task automation. We hypothesize that a promising method for implementing conditionals in such systems is to incorporate the use of verbal instructions. Verbal instructions supplied concurrently with demonstrations have been shown to improve the generalizability of PBD. However, the challenge of supporting conditional creation using this multi-modal approach has not been addressed. In this extended abstract, we present our study on understanding how end users describe conditionals in natural language for mobile app tasks. We conducted a formative study of 56 participants asking them to verbally describe conditionals in different settings for 9 sample tasks and to invent conditional tasks. Participant responses were analyzed using open coding and revealed that, in the context of mobile apps, end users often omit desired else statements when explaining conditionals, sometimes use ambiguous concepts in expressing conditionals, and often desire to implement complex conditionals. Based on these findings, we discuss the implications for designing a multimodal PBD interface to support the creation of conditionals.

关键词： Task analysis Programming Natural languages Automation Encoding Public transportation Visualization

Predicting risk of dyslexia with an online gamified test

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Rello, Luz Baeza-Yates, Ricardo Ali, Abdullah Bigham, Jeffrey P. Serra, Miquel Department of Information Systems and Technology IE University Madrid Spain KCCIS Northeastern University at Silicon Valley San JoseCA United States School of Computer Science University of Washington SeattleWA United States Human-Computer Interaction Institute Carnegie Mellon University PittsburghPA United States Department of Cognition and Development University of Barcelona Barcelona Spain

Dyslexia is a specific learning disorder related to school failure. Detection is both crucial and challenging, especially in languages with transparent orthographies, such as Spanish. To make detecting dyslexia easier, we designed an online gamified test and associated predictive machine learning model. In a study with more than 4,300 participants, our model correctly detected nearly 80% of participants with dyslexia. Our work shows that dyslexia can be screened using a computational approach. An online screening tool based on our methods has already been used by more than 200,000 people. Copyright © 2019, The Authors. All rights reserved.

关键词：

New Interfaces for Musical Expression

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Poupyrev, Ivan Lyons, Michael J. Fels, Sidney Bean, Tina Blaine Interaction Lab Sony CSL 3-14-13 Higashi-Gotanda Tokyo141-0022 Japan ATR MIC 2-2 Hikafidai Seika Souraku-gun Kyoto619-02 Japan Department of ECE UBC VancouverBCV6T 1Z4 Canada School of Computer Science Carnegie Mellon University 5000 Forbes Avenue PittsburghPA15213 United States

The rapid evolution of electronics, digital media, advanced materials, and other areas of technology, is opening up unprecedented opportunities for musical interface inventors and designers. The possibilities afforded by these new technologies carry with them the challenges of a complex and often confusing array of choices for musical composers and performers. New musical technologies are at least partly responsible for the current explosion of new musical forms, some of which are controversial and challenge traditional definitions of music. Alternative musical controllers, currently the leading edge of the ongoing dialogue between technology and musical culture, involve many of the issues covered at past CHI meetings. This workshop brings together interface experts interested in musical controllers and musicians and composers involved in the development of new musical interfaces. © 2020, CC-BY.

关键词： Controllers

The state-of-the-art in Cardiac MRI Reconstruction: Results of the CMRxRecon Challenge in MICCAI 2023

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lyu, Jun Qin, Chen Wang, Shuo Wang, Fanwen Li, Yan Wang, Zi Guo, Kunyuan Ouyang, Cheng Tänzer, Michael Liu, Meng Sun, Longyu Sun, Mengting Li, Qin Shi, Zhang Hua, Sha Li, Hao Chen, Zhensen Zhang, Zhenlin Xin, Bingyu Metaxas, Dimitris N. Yiasemis, George Teuwen, Jonas Zhang, Liping Chen, Weitian Pang, Yanwei Liu, Xiaohan Razumov, Artem Dylov, Dmitry V. Dou, Quan Yan, Kang Xue, Yuyang Du, Yuning Dietlmeier, Julia Garcia-Cabrera, Carles Hemidi, Ziad Al-Haj Vogt, Nora Xu, Ziqiang Zhang, Yajing Chu, Ying-Hua Chen, Weibo Bai, Wenjia Zhuang, Xiahai Qin, Jing Wu, Lianmin Yang, Guang Qu, Xiaobo Wang, He Wang, Chengyan Psychiatry Neuroimaging Laboratory Brigham and Women’s Hospital Harvard Medical School 399 Revolution Drive BostonMA02215 United States Department of Electrical and Electronic Engineering & I-X Imperial College London United Kingdom Digital Medical Research Center School of Basic Medical Sciences Fudan University Shanghai China Department of Bioengineering & I-X Imperial College London LondonW12 7SL United Kingdom Cardiovascular Magnetic Resonance Unit Royal Brompton Hospital Guy’s and St Thomas’ NHS Foundation Trust LondonSW3 6NP United Kingdom Department of Radiology Ruijin Hospital Shanghai Jiao Tong University School of Medicine Shanghai China Department of Electronic Science Fujian Provincial Key Laboratory of Plasma and Magnetic Resonance National Institute for Data Science in Health and Medicine Institute of Artificial Intelligence Xiamen University Xiamen361102 China Department of Computing Department of Brain Sciences Imperial College London LondonSW7 2AZ United Kingdom Human Phenome Institute Fudan University 825 Zhangheng Road Pudong New District Shanghai201203 China Department of Radiology Zhongshan Hospital Fudan University Shanghai China Department of Cardiovascular Medicine Ruijin Hospital Lu Wan Branch Shanghai Jiao Tong University School of Medicine Shanghai China Institute of Science and Technology for Brain-Inspired Intelligence Fudan University Shanghai200433 China Department of Computer Science Rutgers University PiscatawayNJ08854 United States AI for Oncology Netherlands Cancer Institute Plesmanlaan 121 Amsterdam1066 CX Netherlands Department of Imaging and Interventional Radiology The Chinese University of Hong Kong Hong Kong TJK-BIIT Lab School of Electrical and Information Engineering Tianjin University Tianjin300072 China Skolkovo Institute Of Science And Technology Center for Artificial Intelligence Technology 30/1 Bolshoy blvd. Moscow121205 Russia Department of Biomedical Engineering University of Virginia

Cardiac magnetic resonance imaging (MRI) provides detailed and quantitative evaluation of the heart’s structure, function, and tissue characteristics with high-resolution spatial-temporal imaging. However, its slow imaging speed and motion artifacts are notable limitations. Undersampling reconstruction, especially data-driven algorithms, has emerged as a promising solution to accelerate scans and enhance imaging performance using highly under-sampled data. Nevertheless, the scarcity of publicly available cardiac k-space datasets and evaluation platform hinder the development of data-driven reconstruction algorithms. To address this issue, we organized the Cardiac MRI Reconstruction Challenge (CMRxRecon) in 2023, in collaboration with the 26th International Conference on Medical Image Computing and computer-Assisted Intervention (MICCAI). CMRxRecon presented an extensive k-space dataset comprising cine and mapping raw data, accompanied by detailed annotations of cardiac anatomical structures. With overwhelming participation, the challenge attracted more than 285 teams and over 600 participants. Among them, 22 teams successfully submitted Docker containers for the testing phase, with 7 teams submitted for both cine and mapping tasks. All teams use deep learning based approaches, indicating that deep learning has predominately become a promising solution for the problem. The first-place winner of both tasks utilizes the E2E-VarNet architecture as backbones. In contrast, U-Net is still the most popular backbone for both multi-coil and single-coil reconstructions. This paper provides a comprehensive overview of the challenge design, presents a summary of the submitted results, reviews the employed methods, and offers an in-depth discussion that aims to inspire future advancements in cardiac MRI reconstruction models. The summary emphasizes the effective strategies observed in Cardiac MRI reconstruction, including backbone architecture, loss function, pre-processing techn

关键词： Mapping