检索结果-内蒙古大学图书馆

Towards Unified Token Learning for Vision-Language Tracking

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zheng, Yaozong Zhong, Bineng Liang, Qihua Li, Guorong Ji, Rongrong Li, Xianxian The Key Laboratory of Education Blockchain and Intelligent Technology Ministry of Education Guangxi Normal University Guilin541004 China The Guangxi Key Laboratory of Multi-Source Information Mining and Security Guangxi Normal University Guilin541004 China The School of Computer Science and Technology Key Laboratory of Big Data Mining and Knowledge Management University of Chinese Academy of Sciences Beijing100049 China Media Analytics and Computing Lab Department of Artificial Intelligence School of Informatics Xiamen University 361005 China

In this paper, we present a simple, flexible and effective vision-language (VL) tracking pipeline, termed MMTrack, which casts VL tracking as a token generation task. Traditional paradigms address VL tracking task indirectly with sophisticated prior designs, making them over-specialize on the features of specific architectures or mechanisms. In contrast, our proposed framework serializes language description and bounding box into a sequence of discrete tokens. In this new design paradigm, all token queries are required to perceive the desired target and directly predict spatial coordinates of the target in an auto-regressive manner. The design without other prior modules avoids multiple sub-tasks learning and hand-designed loss functions, significantly reducing the complexity of VL tracking modeling and allowing our tracker to use a simple cross-entropy loss as unified optimization objective for VL tracking task. Extensive experiments on TNL2K, LaSOT, LaSOText and OTB99-Lang benchmarks show that our approach achieves promising results, compared to other state-of-the-arts. Copyright © 2023, The Authors. All rights reserved.

关键词： Modeling languages

Subspace variational quantum simulator

学校读者我要写书评

暂无评论

Physical Review Research 2023年第2期5卷 023078-023078页

作者： Kentaro Heya Ken M. Nakanishi Kosuke Mitarai Zhiguang Yan Kun Zuo Yasunari Suzuki Takanori Sugiyama Shuhei Tamate Yutaka Tabuchi Keisuke Fujii Yasunobu Nakamura RIKEN Center for Quantum Computing (RQC) Wako Saitama 351-0198 Japan Institute for Physics of Intelligence The University of Tokyo Tokyo 113-0033 Japan Graduate School of Engineering Science Osaka University 1-3 Machikaneyama Toyonaka Osaka 560-8531 Japan Center for Quantum Information and Quantum Biology Osaka University Japan JST PRESTO 4-1-8 Honcho Kawaguchi Saitama 332-0012 Japan NTT Computer and Data Science Laboratories NTT Corporation Musashino 180-8585 Japan Research Center for Advanced Science and Technology (RCAST) The University of Tokyo Meguro-ku Tokyo 153-8904 Japan Department of Applied Physics Graduate School of Engineering The University of Tokyo Bunkyo-ku Tokyo 113-8656 Japan

Quantum simulation is one of the key applications of quantum computing, which accelerates research and development in the fields such as chemistry and material science. The recent development of noisy intermediate-scale quantum (NISQ) devices urges the exploration of applications without the necessity of quantum error correction. In this paper, we propose an efficient method to simulate quantum dynamics driven by a static Hamiltonian on NISQ devices, named subspace variational quantum simulator (SVQS). SVQS employs the subspace-search variational quantum eigensolver (SSVQE) [Phys. Rev. Res. 1, 033062 (2019)] to find a low-lying eigensubspace and extends it to simulate dynamics within the subspace with lower overhead compared to the existing schemes. We experimentally simulate the time-evolution operator in a low-lying eigensubspace of a hydrogen molecule. We also define the subspace process fidelity as a measure between two quantum processes in a subspace. The subspace time evolution mimicked by SVQS shows the subspace process fidelity of 0.896–0.989.

关键词： Quantum algorithms Quantum algorithms for chemical calculations Quantum computation Quantum control Quantum gates Quantum simulation Quantum tomography

ConBERT-RL: A policy-driven deep reinforcement learning based approach for detecting homophobia and transphobia in low-resource languages

学校读者我要写书评

暂无评论

Natural Language Processing Journal

Natural Language Processing Journal 2024年 6卷

作者： Vivek Suresh Raj Chinnaudayar Navaneethakrishnan Subalalitha Lavanya Sambath Frank Glavin Bharathi Raja Chakravarthi University of Calgary Calgary Alberta Canada Department of Computing Technologies SRM Institute of Science and Technology Chennai India Department of Information Technology Madras Institute of Technology Chennai India Insight SFI Research Centre for Data Analytics School of Computer Science University of Galway Galway Ireland School of Computer Science University of Galway Galway Ireland

In this work, we present a novel framework for discriminatory comment classification in targeted low-resource languages thereby enabling identification of discriminatory comments for promoting safer online environment in linguistically diverse contexts. Recently proposed literatures with Bidirectional Encoder Representations from Transformers (BERT) and its variants, have produced promising results, particularly in the case of transliterated Tamil words in English. Such approaches are seen as transfer learning and fine-tuning between a general environment and targeted downstream task. However, for an effective transfer of knowledge, from a source task to a targeted task, their feature space has to be correlated (similarity measure as metric, to measure knowledge transfer), which is unexplored in the previous literature works. In practice, such similarity conditions are often violated. We propose a Con catenated representation, powered with BERT in R einforcement L earning (RL) - ConBERT-RL framework, to capture problem-specific features, as well as to understand nuances in case of transliterated Tamil words into English, to make decision in classifying discriminatory comments. ConBERT-RL incorporates a fusion of learned hidden-state representation from our pre-trained Classifier-Model (CM), along with the broader contextualized representation (pooled output) from BERT. The key idea is to utilize this concatenated representation to drive our policy network for hate comment classification. To effectively learn such a policy, we use the REINFORCE algorithm in a reinforcement learning setting, to guide our ConBERT-RL agent, in making informative decisions. To demonstrate the general aspects of ConBERT-RL, we conduct experiments for offensive comment classification on transliterated Tamil words in English dataset. ConBERT-RL obtains results, where it significantly improved the score of micro average accuracy with 90% ( ≈ 1 . 0 % absolute improvement over BERT+FC), 93% (

关键词： Deep reinforcement learning Low resourced language Homophobia detection Transphobia detection

GradSA: Gradient Sparsification and Accumulation for Communication-Efficient Distributed Deep Learning 15th

学校读者我要写书评

暂无评论

GradSA: Gradient Sparsification and Accumulation for Communi...

15th International Conference on Green, Pervasive, and Cloud computing, GPC 2020

作者： Liu, Bo Jiang, Wenbin Zhao, Shaofeng Jin, Hai He, Bingsheng National Engineering Research Center for Big Data Technology and System Service Computing Technology and System Lab Cluster and Grid Computing Lab Huazhong University of Science and Technology Wuhan China Wuhan National Laboratory for Optoelectronics Key Laboratory of Information Storage System Engineering Research Center of Data Storage Systems and Technology Huazhong University of Science and Technology Wuhan China Library Henan University of Economics and Law Zhengzhou China Department of Computer Science School of Computing National University of Singapore Singapore Singapore

ISBN: (纸本)9783030642426

Large-scale distributed deep learning is of great importance in various applications. For distributed training, the inter-node gradient communication often becomes the performance bottleneck. Gradient sparsification has been proposed to reduce the communication overhead. However, the sparsification only arbitrarily selects a small fraction of gradients, while the efficiency of selection dimension has been overlooked. Furthermore, gradient staleness is inevitable after applying the sparsification which will ultimately lead to model divergence. In this paper, we propose a staleness-compensated sparse stochastic gradient descent algorithm, GradSA, to improve the training efficiency. Layer-level gradients are sparsified to reduce the communication overhead, which conform to the characteristics of the network structure, and historical accumulation of the approximated gradients is utilized to speed up convergence. We demonstrate the model convergence acceleration and the efficiency of our layer-level selection over existing state-of-the-art works, such as DGC, TernGrad, and 8-Bit quantization. © 2020, Springer Nature Switzerland AG.

关键词： Efficiency

Building bilingual and code-switched voice conversion with limited training data using embedding consistency loss

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Yang, Yaogen Zhang, Haozhe Qin, Xiaoyi Liang, Shanshan Cui, Huahua Xu, Mingyang Li, Ming Data Science Research Center Duke Kunshan University Kunshan China School of Electronics and Information Technology Sun Yat-sen University Guangzhou China School of Computer Science Wuhan University Wuhan China Advanced Computing East China Sub-center Su Zhou China

Building cross-lingual voice conversion (VC) systems for multiple speakers and multiple languages has been a challenging task for a long time. This paper describes a parallel nonautoregressive network to achieve bilingual and code-switched voice conversion for multiple speakers when there are only mono-lingual corpora for each language. We achieve crosslingual VC between Mandarin speech with multiple speakers and English speech with multiple speakers by applying bilingual bottleneck features. To boost voice cloning performance, we use an adversarial speaker classifier with a gradient reversal layer to reduce the source speaker's information from the output of encoder. Furthermore, in order to improve speaker similarity between reference speech and converted speech, we adopt an embedding consistency loss between the synthesized speech and its natural reference speech in our network. Experimental results show that our proposed method can achieve high quality converted speech with mean opinion score (MOS) around 4. The conversion system performs well in terms of speaker similarity for both in-set speaker conversion and out-set-of one-shot conversion. Audio samples are available online1. Copyright © 2021, The Authors. All rights reserved.

关键词： Embeddings

Assessing responsible innovation training

学校读者我要写书评

暂无评论

Journal of Responsible Technology 2023年 16卷

作者： Stahl, Bernd Carsten Aicardi, Christine Brooks, Laurence Craigon, Peter J. Cunden, Mayen Burton, Saheli Datta Heaver, Martin De Saille, Stevienna De Dolby, Serena Dowthwaite, Liz Eke, Damian Hughes, Stephen Keene, Paul Kuh, Vivienne Portillo, Virginia Shanley, Danielle Smallman, Melanie Smith, Michael Stilgoe, Jack Ulnicane, Inga Wagner, Christian Webb, Helena School of Computer Science University of Nottingham United Kingdom Dept. of Informatics and School of Biomedical Engineering & Imaging Sciences King's College London United Kingdom Information School University of Sheffield Sheffield United Kingdom Horizon Digital Economy Research School of Computer Science University of Nottingham United Kingdom Independent Researcher Dept of Science and Technology Studies University College London United Kingdom ORBIT RRI limited Leicester United Kingdom Dept of Sociological Studies University of Sheffield Sheffield United Kingdom Centre for Computing and Social Responsibility De Montfort University Leicester United Kingdom School of Chemistry University of Bristol United Kingdom Department of Philosophy Faculty of Arts and Social Sciences Maastricht University Netherlands Biodiversity and Conservation Science Department of Biodiversity Conservation and Attractions Wildlife Research Centre Western Australia De Montfort University United Kingdom Lab for Uncertainty in Data and Decision Making (LUCID) School of Computer Science University of Nottingham United Kingdom

There is broad agreement that one important aspect of responsible innovation (RI) is to provide training on its principles and practices to current and future researchers and innovators, notably including doctoral students. Much less agreement can be observed concerning the question of what this training should consist of, how it should be delivered and how it could be assessed. The increasing institutional embedding of RI leads to calls for the alignment of RI training with training in other subjects. One can therefore observe a push towards the official assessment of RI training, for example in the recent call for proposals for centres for doctoral training by UK Research and Innovation. This editorial article takes its point of departure from the recognition that the RI community will need to react to the call for assessment of RI training. It provides an overview of the background and open questions around RI training and assessment as a background of examples of RI training assessment at doctoral level. There is unlikely to be one right way of assessing RI training across institutions and disciplines, but we expect that the examples provided in this article can help RI scholars and practitioners orient their training and its assessment in ways that are academically viable as well as supportive of the overall aims of RI. © 2023

关键词： Assessment Doctoral training Education Responsible innovation

Towards Tailored Models on Private AIoT Devices: Federated Direct Neural Architecture Search

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Chunhui Yuan, Xiaoming Zhang, Qianyun Zhu, Guangxu Cheng, Lei Zhang, Ning The Michtom School of Computer Science Brandeis University WalthamMA United States The Qinhuangdao Branch Campus Northeastern University Qinhuangdao066004 China The School of Cyber Science and Technology Beihang University Beijing China The Shenzhen Research Institute of Big Data Guangdong Shenzhen China The College of Information Science and Electronic Engineering Zhejiang University China Hangzhou310027 China The Department of Electrical and Computing Engineering University of Windsor WindsorONN9B 3P4 Canada

Neural networks often encounter various stringent resource constraints while deploying on edge devices. To tackle these problems with less human efforts, automated machine learning becomes popular in finding various neural architectures that fit diverse Artificial Intelligence of Things (AIoT) scenarios. Recently, to prevent the leakage of private information while enable automated machine intelligence, there is an emerging trend to integrate federated learning and neural architecture search (NAS). Although promising as it may seem, the coupling of difficulties from both tenets makes the algorithm development quite challenging. In particular, how to efficiently search the optimal neural architecture directly from massive non-independent and identically distributed (non-IID) data among AIoT devices in a federated manner is a hard nut to crack. In this paper, to tackle this challenge, by leveraging the advances in ProxylessNAS, we propose a Federated Direct Neural Architecture Search (FDNAS) framework that allows for hardware-friendly NAS from non-IID data across devices. To further adapt to both various data distributions and different type of devices with heterogeneous embedded hardware platforms, inspired by meta-learning, a Cluster Federated Direct Neural Architecture Search (CFDNAS) framework is proposed to achieve device-aware NAS, in the sense that each device can learn a tailored deep learning model for its particular data distribution and hardware constraint. Extensive experiments on non-IID datasets have shown the state-of-the-art accuracy-efficiency trade-offs achieved by the proposed solution in the presence of both data and device heterogeneity. © 2022, CC BY.

关键词： Network architecture

Exploring bilingual parallel corpora for syntactically controllable paraphrase generation 29

学校读者我要写书评

暂无评论

Exploring bilingual parallel corpora for syntactically contr...

29th International Joint Conference on Artificial Intelligence, IJCAI 2020

作者： Liu, Mingtong Yang, Erguang Xiong, Deyi Zhang, Yujie Sheng, Chen Hu, Changjian Xu, Jinan Chen, Yufeng School of Computer Science and Information Technology & Beijing Key Lab of Traffic Data Analysis and Mining Beijing Jiaotong University Beijing China College of Intelligence and Computing Tianjin University Tianjin China Lenovo Research AI Lab Beijing China

ISBN: (纸本)9780999241165

Paraphrase generation is of great importance to many downstream tasks in natural language processing. Recent efforts have focused on generating paraphrases in specific syntactic forms, which, generally, heavily relies on manually annotated paraphrase data that is not easily available for many languages and domains. In this paper, we propose a novel end-to-end framework to leverage existing large-scale bilingual parallel corpora to generate paraphrases under the control of syntactic exemplars. In order to train one model over the two languages of parallel corpora, we embed sentences of them into the same content and style spaces with shared content and style encoders using cross-lingual word embeddings. We propose an adversarial discriminator to disentangle the content and style space, and employ a latent variable to model the syntactic style of a given exemplar in order to guide the two decoders for generation. Additionally, we introduce cycle and masking learning schemes to efficiently train the model. Experiments and analyses demonstrate that the proposed model trained only on bilingual parallel data is capable of generating diverse paraphrases with desirable syntactic styles. Fine-tuning the trained model on a small paraphrase corpus makes it substantially outperform state-of-the-art paraphrase generation models trained on a larger paraphrase dataset. © 2020 Inst. Sci. inf., Univ. Defence in Belgrade. All rights reserved.

关键词： Syntactics

Deep Reinforcement Learning Guided Graph Neural Networks for Brain Network Analysis

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhao, Xusheng Wu, Jia Peng, Hao Beheshti, Amin Monaghan, Jessica J.M. McAlpine, David Hernandez-Perez, Heivet Dras, Mark Dai, Qiong Li, Yangyang Yu, Philip S. He, Lifang Institute of Information Engineering Chinese Academy of Sciences Beijing China School of Cyber Security University of Chinese Academy of Sciences Beijing China School of Computing Macquarie University Sydney Australia Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University Beijing China National Acoustic Laboratories Sydney Australia Department of Linguistics The Australian Hearing Hub Macquarie University Sydney Australia CAEIT Beijing China Department of Computer Science University of Illinois ChicagoIL United States Computer Science & Engineering Lehigh University PA United States

Modern neuroimaging techniques enable us to construct human brains as brain networks or connectomes. Capturing brain networks' structural information and hierarchical patterns is essential for understanding brain functions and disease states. Recently, the promising network representation learning capability of graph neural networks (GNNs) has prompted related methods for brain network analysis to be proposed. Specifically, these methods apply feature aggregation and global pooling to convert brain network instances into vector representations encoding brain structure induction for downstream brain network analysis tasks. However, existing GNN-based methods often neglect that brain networks of different subjects may require various aggregation iterations and use GNN with a fixed number of layers to learn all brain networks. Therefore, how to fully release the potential of GNNs to promote brain network analysis is still non-trivial. In our work, a novel brain network representation framework, BN-GNN, is proposed to solve this difficulty, which searches for the optimal GNN architecture for each brain network. Concretely, BN-GNN employs deep reinforcement learning (DRL) to automatically predict the optimal number of feature propagations (reflected in the number of GNN layers) required for a given brain network. Furthermore, BN-GNN improves the upper bound of traditional GNNs' performance in eight brain network disease analysis tasks. © 2022, CC BY.

关键词： Reinforcement learning