检索结果-内蒙古大学图书馆

Instrumental Variable Value Iteration for Causal Offline Reinforcement Learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Liao, Luofeng Fu, Zuyue Yang, Zhuoran Wang, Yixin Ma, Dingli Kolar, Mladen Wang, Zhaoran Department of Industrial Engineering and Operations Research Columbia University New YorkNY10027 United States Department of Industrial Engineering and Management Sciences Northwestern University EvanstonIL60208 United States Department of Statistics and Data Science Yale University New HavenCT06520 United States Department of Statistics University of Michigan Ann ArborMI48109 United States Department of Information Systems and Operations Management Michael G. Foster School of Business University of Washington SeattleWA98195 United States Department of Data Sciences Operations Marshall School of Business University of Southern California Los AngelesCA90089 United States

In offline reinforcement learning (RL) an optimal policy is learned solely from a priori collected observational data. However, in observational data, actions are often confounded by unobserved variables. Instrumental variables (IVs), in the context of RL, are the variables whose influence on the state variables is all mediated by the action. When a valid instrument is present, we can recover the confounded transition dynamics through observational data. We study a confounded Markov decision process where the transition dynamics admit an additive nonlinear functional form. Using IVs, we derive a conditional moment restriction through which we can identify transition dynamics based on observational data. We propose a provably efficient IV-aided Value Iteration (IVVI) algorithm based on a primal-dual reformulation of the conditional moment restriction. To our knowledge, this is the first provably efficient algorithm for instrument-aided offline RL. Copyright © 2021, The Authors. All rights reserved.

关键词： Reinforcement learning

Low-complexity hybrid beamforming for multi-cell mmWave massive MIMO: A primitive Kronecker decomposition approach

学校读者我要写书评

暂无评论

Signal Processing 2026年 238卷

作者： Teng Sun Guangxu Zhu Xiaofan Li Jiancun Fan Minghua Xia School of Electronic and Information Engineering Xi’an Jiaotong University No. 28 Xianning West Rd. Xi’an 710049 Shaanxi China The 54th Research Institute of CETC No. 580 Zhongshan West Rd. Shijiazhuang 050081 Hebei China Shenzhen Research Institute of Big Data 2001 Longxiang Avenue Shenzhen 518172 Guangdong China School of Intelligent Systems Science and Engineering Jinan University 206 Qianshan Rd. Zhuhai 519070 Guangdong China School of Electronics and Information Technology Sun Yat-sen University No. 132 Waihan East Rd. Guangzhou 510006 Guangdong China

To circumvent the high path loss of mmWave propagation and reduce the hardware cost of massive multiple-input multiple-output antenna systems, full-dimensional hybrid beamforming is critical in 5G and beyond wireless communications. Concerning an uplink multi-cell system with a large-scale uniform planar antenna array, this paper designs an efficient hybrid beamformer using primitive Kronecker decomposition and dynamic factor allocation, where the analog beamformer applies to null the inter-cell interference and simultaneously enhances the desired signals. In contrast, the digital beamformer mitigates the intra-cell interference using the minimum mean square error (MMSE) criterion. Then, due to the low accuracy of phase shifters inherent in the analog beamformer, a low-complexity hybrid beamformer is developed to slow its adjustment speed. Next, an optimality analysis from a subspace perspective is performed, and a sufficient condition for optimal antenna configuration is established. Finally, simulation results demonstrate that the achievable sum rate of the proposed beamformer approaches that of the optimal pure digital MMSE scheme, yet with much lower computational complexity and hardware cost.

关键词： Full-dimensional beamforming Hybrid beamformer Massive MIMO mmWave communication Primitive Kronecker decomposition

Predicting violent behavior using language agnostic models 10

学校读者我要写书评

暂无评论

Predicting violent behavior using language agnostic models

10th International Joint Conference on Knowledge Discovery, Knowledge engineering and Knowledge Management, IC3K 2018

作者： Liu, Yingjie Wert, Gregory Greenawald, Benjamin Boni, Mohammad Al Brown, Donald E. Data Science Institute University of Virginia United States Department of Systems and Information Engineering University of Virginia United States

ISBN: (纸本)9789897583308

Groups advocating violence have caused significant destruction to individuals and societies. To combat this, governmental and non-governmental organizations must quickly identify violent groups and limit their exposure. While some groups are well-known for their violence, smaller, less recognized groups are difficult to classify. However, using texts from these groups, we may be able to identify them. This paper applies text analysis techniques to differentiate violent and non-violent groups using discourses from various value-motivated groups. Significantly, the algorithms are constructed to be language-agnostic. The results show that deep learning models outperform traditional models. Our models achieve high accuracy when fairly trained only on data from other groups. Additionally, the results indicate that the models achieve better performance by removing groups with a large amount of documents that can bias the classification. This study shows promise in using scalable, language-independent techniques to effectively identify violent value-motivated groups. Copyright 2018 by SCITEPRESS – science and Technology Publications, Lda. All rights reserved

关键词： information retrieval systems

Social media in product development

学校读者我要写书评

暂无评论

Social media in product development

2018 systems and information engineering Design Symposium, SIEDS 2018

作者： Sandor, Logan Midura, Justis Abedin, Sabreena Ingber, George Pederson, Maxwell Sander, Tim Wolf, Liam Bailey, Reid Flower, Abigail Department of Systems and Information Engineering University of Virginia United States Data Science Institute University of Virginia United States

ISBN: (纸本)9781538663431

Social media offers a unique and wide-reaching view into consumers' thoughts, opinions, recommendations, and personal experiences. This kind of information can be valuable in the product development process. The focus of this project is to examine if and how unprompted social media data can inform new product development using a case study with a Fortune 100 consumer products company. In the case study, a text analytics tool was created to allow discovery and exploration of potential new trends in skincare ingredients and application methods using data obtained from Reddit, Facebook, and skincare blogs. The tool identifies and monitors trends over time for popularity and impact to inform decision making in new product development. Based on the case study, important factors in using social media to identify trends include 1) finding social media sites with relevant content to meet the goals of the project, 2) using term lists to target specific areas of interest, and 3) considering time with respect to frequency of postings, setting time periods of trend measurement, and ensuring the data includes timestamps. Setting the time period over which trends are assessed involves a trade-off between the ability to identify early trends (shorter time periods are better) and sufficient data (longer time periods are better). In the case study, sites where people come for help from a community (e.g., Reddit) and DIY/maker blogs proved to be good sources of emerging trends. The use of term lists proved key in targeting analyses on ingredients and application methods in the case. © 2018 IEEE.

关键词： Product development

Retraction notice to “Neural network-based optimization of hydrogen fuel production energy system with proton exchange electrolyzer supported nanomaterial” [Fuel 332 (2023) 125827]

学校读者我要写书评

暂无评论

Fuel 2025年 391卷

作者： Tao Hai Kosar Hikmat Hama Aziz Jincheng Zhou Hayder A. Dhahad Kamal Sharma Sattam Fahad Almojil Abdulaziz Ibrahim Almohana Abdulrhman Fahmi Alali Teeba Ismail Kh Sadok Mehrez Anas Abdelrahman School of Computer and Information Qiannan Normal University for Nationalities Duyun Guizhou 558000 China Key Laboratory of Complex Systems and Intelligent Optimization of Guizhou Duyun Guizhou 558000 China Institute for Big Data Analytics and Artificial Intelligence (IBDAAI) Universiti Teknologi MARA Shah Alam Selangor 40450 Malaysia Department of Chemistry College of Science University of Sulaimani Qlyasan Street Kurdistan Region 46001 Iraq Department of Medical Laboratory of Science College of Health Sciences University of Human Development Sulaimaniyah Iraq Key Laboratory of Complex Systems and Intelligent Optimization of Qiannan Duyun 558000 China Mechanical Engineering Department University of Technology Baghdad Iraq Institute of Engineering and Technology GLA University Mathura Uttar Pradesh 281406 India Department of Civil Engineering College of Engineering King Saud University P.O. Box 800 Riyadh 11421 Saudi Arabia Department of Computer Engineering College of Engineering and Computer Science Lebanese French University Kurdistan Region Iraq Department of Mechanical Engineering College of Engineering at Al Kharj Prince Sattam bin Abdulaziz University 16273 Saudi Arabia Department of Mechanical Engineering University of Tunis El Manar ENIT BP 37 Le Belv’ed‘ere Tunis 1002 Tunisia Department of Chemical Engineering Faculty of Engineering & Technology Future University in Egypt New Cairo 11500 Tunisia

Web Intelligence (WI) 3.0: in search of a better-connected world to create a future intelligent society

学校读者我要写书评

暂无评论

Artificial Intelligence Review 2025年第9期58卷

作者： Kuai, Hongzhi Huang, Jimmy X. Tao, Xiaohui Pasi, Gabriella Yao, Yiyu Liu, Jiming Zhong, Ning School of Computer Science and Technology/School of Artificial Intelligence Chongqing University of Posts and Telecommunications Chongqing China Key Laboratory of Big Data Intelligent Computing Chongqing University of Posts and Telecommunications Chongqing China School of Information Technology York University North YorkON Canada School of Mathematics Physics and Computing University of Southern Queensland ToowoombaQLD Australia Department of Informatics Systems and Communication University of Milano-Bicocca Milano Italy Department of Computer Science University of Regina Regina Canada Department of Computer Science Hong Kong Baptist University Kowloon Tong Hong Kong Faculty of Engineering Maebashi Institute of Technology Gunma Maebashi Japan

Over the past two decades, Web Intelligence (WI) has emerged as a key field driving the evolution of AI in the connected world, addressing the demands of a future intelligent society. This paper provides a comprehensive review of WI’s contributions since its inception in 2000, spanning three distinct phases: Wisdom World Wide Web (WI 1.0, 2000–2009), Wisdom Web of Things (WI 2.0, 2010–2017), and Wisdom Web of Everything (WI 3.0, since 2018). For each phase, we examine key advancements, challenges, and future directions from the perspectives of both intelligent machines and human experts, highlighting significant societal impacts. To advance WI research, we propose a large language model-based learning framework for topic analysis and trend prediction. Moving beyond single-perspective approaches, we emphasize the Connected Intelligence Ecosystem defined by the HIGH5 scheme comprising one goal, two twins, three fundamentals, four functions, and five services that are realized through WI 3.0. This vision serves as a bridge from localized models to a global reference framework for addressing sustainability challenges in future societies. To illustrate the real-world implications of WI 3.0, we present case studies focusing on brain-inspired research, particularly in the intersection of brain intelligence, brain health, and brainternet-fostering interdisciplinary collaboration across diverse research communities. © The Author(s) 2025.

关键词： Intelligent computing

Real-time Semantic Segmentation via Spatial-detail Guided Context Propagation

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Hao, Shijie Zhou, Yuan Guo, Yanrong Hong, Richang Cheng, Jun Wang, Meng Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology Ministry of Education School of Computer Science and Information Engineering Hefei University of Technology Hefei230009 China CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Chinese University of Hong Kong Hong Kong

Nowadays, vision-based computing tasks play an important role in various real-world applications. However, many vision computing tasks, e.g. semantic segmentation, are usually computationally expensive, posing a challenge to the computing systems that are resource-constrained but require fast response speed. Therefore, it is valuable to develop accurate and real-time vision processing models that only require limited computational resources. To this end, we propose the Spatial-detail Guided Context Propagation Network (SGCPNet) for achieving real-time semantic segmentation. In SGCPNet, we propose the strategy of spatial-detail guided context propagation. It uses the spatial details of shallow layers to guide the propagation of the low-resolution global contexts, in which the lost spatial information can be effectively reconstructed. In this way, the need for maintaining high-resolution features along the network is freed, therefore largely improving the model efficiency. On the other hand, due to the effective reconstruction of spatial details, the segmentation accuracy can be still preserved. In the experiments, we validate the effectiveness and efficiency of the proposed SGCPNet model. On the Citysacpes dataset, for example, our SGCPNet achieves 69.5% mIoU segmentation accuracy, while its speed reaches 178.5 FPS on 768×1536 images on a GeForce GTX 1080 Ti GPU card. In addition, SGCPNet is very lightweight and only contains 0.61 M parameters. The code will be released at https://***/zhouyuan888888/SGCPNet. Copyright © 2020, The Authors. All rights reserved.

关键词： Semantic Segmentation

Multi-node optical frequency dissemination with post automatic phase correction

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Hu, Liang Tian, Xueyang Wu, Guiling Kong, Mengya Shen, Jianguo Chen, Jianping State Key Laboratory of Advanced Optical Communication Systems and Networks Department of Electronic Engineering Shanghai Jiao Tong University Shanghai200240 China Shanghai Institute for Advanced Communication and Data Science Shanghai Jiao Tong University Shanghai200240 China the College of Physics and electronic information Engineering Zhejiang Normal University Jinhua321004 China

—We report a technique for coherence transfer of laser light through a fiber link, where the optical phase noise induced by environmental perturbation via the fiber link is compensated by remote users with passive phase noise correction, rather than at the input as is conventional. Neither phase discrimination nor active phase tracking is required due to the open-loop design, mitigating some technical problems such as the limited compensation speed and the finite compensation precision as conventional active phase noise cancellation. We theoretically analyze and experimentally demonstrate that the delay effect introduced residual fiber phase noise after noise compensation is a factor of 7 higher than the conventional techniques. Using this technique, we demonstrate the transfer laser light through a 145-km-long, lab-based spooled fiber. After being compensated, the relative frequency instability in terms of overlapping Allan deviation is 1.9 × 10−15 at 1s averaging time and scales down 5.3×10−19 at 10,000 s averaging time. The frequency uncertainty of the light after transferring through the fiber relative to that of the input light is (−0.36 ± 2.6) × 10−18. As the transmitted optical signal remains unaltered until it reaches the remote sites, it can be transmitted simultaneously to multiple remote sites on an arbitrarily complex fiber network, paving a way to develop a multi-node optical frequency dissemination system with post automatic phase noise correction for a number of end users. Copyright © 2020, The Authors. All rights reserved.

关键词： Natural frequencies

Fundamental limitations of Rayleigh backscattering noise on fiber-based multiple-access optical frequency transfer

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Hu, Liang Tian, Xueyang Wu, Guiling Shen, Jianguo Chen, Jianping State Key Laboratory of Advanced Optical Communication Systems and Networks Department of Electronic Engineering Shanghai Jiao Tong University Shanghai200240 China Shanghai Institute for Advanced Communication and Data Science Shanghai Jiao Tong University Shanghai200240 China College of Physics and Electronic Information Engineering Zhejiang Normal University Jinhua321004 China

Ultimate limits of performance for single-source-bidirectional architecture e.g., optical frequency transfer are set by delay fluctuations within the host material, e.g., optical fiber or atmospheric channel, which is dynamically induced. The most common manifestation of such phase disorder in the optical fiber is backscattering lights coming from Rayleigh backscattering and fresnel reflection, which is observed in nearly all single-source-bidirectional architectures and can lead to both irreversible coherence losses as well as undesirable interference coupling. While it has been shown that backscattering induced phase noise can be suppressed by adopting acoustic-optic-modulators (AOMs) at the local and remote sites to break the frequency symmetry in both directions. However, this issue can not be avoided for conventional fiber-optic multiple-access coherent optical phase dissemination in which the interference of the signal light with the Rayleigh backscattered light will probably destroy the coherence of the stabilized optical signal. We suppress the backscattering effect by locally breaking the frequency symmetry at the extraction point by inserting an additional AOM. Here, we theoretically analyze and experimentally demonstrate an add-drop one more AOM approach for suppressing the Rayleigh backscattering within the fiber link. Near-complete suppression of backscattering noise is experimentally confirmed through the measurementthe elimination of a common interference term of the signal light and the Rayleigh backscattered light. The results demonstrate that the Rayleigh backscattering light has a limited effect compared to the residual delay-limited fiber phase noise on the system’s performance. Our results also provide new evidence that it is possible to largely suppress Rayleigh and other backscattering noise within a long optical fiber link, where the accumulated phase noise could be large, by using frequency symmetry breaking at each access node to achieve ro

关键词： Natural frequencies