检索结果-南通市图书馆

同方期刊数据库

Jittor-GAN:A fast-training generative adversarial network model zoo based on Jittor

在线全文

学校读者我要写书评

暂无评论

Computational Visual Media 2021年第1期7卷 153-157页

作者： Wen-Yang Zhou Guo-Wei Yang Shi-Min Hu bnrist Tsinghua UniversityBeijing 100084China

With the emergence of CNNs and massive datasets,the performance of many tasks in computer vision has been greatly improved,such as object detection,instance segmentation,and image *** last has many novel applications,... 详细信息

关键词： computer image network

维普期刊数据库

Federated Offline Reinforcement Learning with Proximal Policy Evaluation

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2024年第6期33卷 1360-1372页

作者： Sheng YUE Yongheng DENG Guanbo WANG Ju REN Yaoxue ZHANG Department of Computer Science and Technology BNRist Tsinghua University Zhongguancun Laboratory

Offline reinforcement learning(RL) has gathered increasing attention in recent years, which seeks to learn policies from static datasets without active online exploration. However, the existing offline RL approaches often require a large amount of pre-collected data and hence are hardly implemented by a single agent in practice. Inspired by the advancement of federated learning(FL), this paper studies federated offline reinforcement learning(FORL),whereby multiple agents collaboratively carry out offline policy learning with no need to share their raw ***, a straightforward solution is to simply retrofit the off-the-shelf offline RL methods for FL, whereas such an approach easily overfits individual datasets during local updating, leading to instability and subpar performance. To overcome this challenge, we propose a new FORL algorithm, named model-free(MF)-FORL, that exploits novel“proximal local policy evaluation” to judiciously push up action values beyond local data support, enabling agents to capture the individual information without forgetting the aggregated knowledge. Further, we introduce a model-based variant, MB-FORL, capable of improving the generalization ability and computational efficiency via utilizing a learned dynamics model. We evaluate the proposed algorithms on a suite of complex and high-dimensional offline RL benchmarks, and the results demonstrate significant performance gains over the baselines.

关键词： Federated learning Computational modeling Heuristic algorithms Reinforcement learning Performance gain Benchmark testing Data models Computational efficiency Trajectory Servers

Data-Driven Control of Linear Systems via Quantized Feedback

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Journal of Systems Science & Complexity 2024年第1期37卷 152-168页

作者： LI Xingchen ZHAO Feiran YOU Keyou Department of Automation and bnrist Tsinghua UniversityBeijing 100084China

Quantized feedback control is fundamental to system synthesis with limited communication *** sharp contrast to the existing literature on quantized control which requires an explicit dynamical model,the authors study the quadratic stabilization and performance control problems with logarithmically quantized feedback in a direct data-driven framework,where the system state matrix is not exactly known and instead,belongs to an ambiguity set that is directly constructed from a finite number of noisy system *** this end,the authors firstly establish sufficient and necessary conditions via linear matrix inequalities for the existence of a common quantized controller that achieves our control objectives over the ambiguity ***,the authors provide necessary conditions on the data for the solvability of the LMIs,and determine the coarsest quantization density via semi-definite *** theoretical results are validated through numerical examples.

关键词： Data-driven control linear matrix inequalities linear systems quantized control.

维普期刊数据库

Prospect Theory Based Individual Irrationality Modelling and Behavior Inducement in Pandemic Control

在线全文

学校读者我要写书评

暂无评论

Computer Modeling in Engineering & Sciences 2024年第7期140卷 139-170页

作者： Wenxiang Dong H.Vicky Zhao Department of Automation BNRistTsinghua UniversityBeijing100084China

Understanding and modeling individuals’behaviors during epidemics is crucial for effective epidemic ***,existing research ignores the impact of users’irrationality on decision-making in the ***,existing disease control methods often assume users’full compliance with measures like mandatory isolation,which does not align with the actual *** address these issues,this paper proposes a prospect theorybased framework to model users’decision-making process in epidemics and analyzes how irrationality affects individuals’behaviors and epidemic *** to the analysis results,irrationality tends to prompt conservative behaviors when the infection risk is low but encourages risk-seeking behaviors when the risk is ***,this paper proposes a behavior inducement algorithm to guide individuals’behaviors and control the spread of *** and real user tests validate our analysis,and simulation results show that the proposed behavior inducement algorithm can effectively guide individuals’behavior.

关键词： Disease spread behavior model irrationality prospect theory

Array-level boosting method with spatial extended allocation to improve the accuracy of memristor based computing-in-memory chips

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2021年第6期64卷 80-88页

作者： Wenqiang ZHANG Bin GAO Peng YAO Jianshi TANG He QIAN Huaqiang WU Institute of Microelectronics Beijing National Research Center for Information Science and Technology (BNRist)Tsinghua University Institute of Microelectronics Beijing National Research Center for Information Science and Technology (BNRist)Tsinghua University

Memristor based computing-in-memory chips have shown the potentials to accelerate deep neural networks with high energy efficiency. Due to the inherent filament-based conductive mechanism of the memristor, the reading and writing noises are hard to eliminate. Besides, the precision of the large-scale memristor array is still limited. However, when the noise of the memristor is large, the existing training methods to reduce the accuracy loss of memristor based computing-in-memory chips will face ***, we proposed the array-level boosting method with spatial extended allocation to reduce the accuracy loss induced by the limited precision and large noises. To optimize the spatial allocation number of each layer in the neural network, the greedy spatial extended allocation algorithm is also proposed. The image processing and classification tasks are demonstrated based on fabricated 32×128 memristor arrays to valid the performance of the proposed method. The chip-in-loop results show that the recovered accuracy of ResNet-34 on CIFAR-10 with array-level boosting method is 92.3%, which is closed to software-based accuracy of 93.2%.

关键词： memristor computing-in-memory array-level boosting neuromorphic computing RRAM

View planning in robot active vision: A survey of systems,algorithms, and applications

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Computational Visual Media 2020年第3期6卷 225-245页

作者： Rui Zeng Yuhui Wen Wang Zhao Yong-Jin Liu bnrist Department of Computer Science and TechnologyTsinghua UniversityBeijingChina

Rapid development of artificial intelligence motivates researchers to expand the capabilities of intelligent and autonomous robots. In many robotic applications, robots are required to make planning decisions based on perceptual information to achieve diverse goals in an efficient and effective way. The planning problem has been investigated in active robot vision, in which a robot analyzes its environment and its own state in order to move sensors to obtain more useful information under certain constraints. View planning,which aims to find the best view sequence for a sensor,is one of the most challenging issues in active robot vision. The quality and efficiency of view planning are critical for many robot systems and are influenced by the nature of their tasks, hardware conditions, scanning states, and planning strategies. In this paper, wefirst summarize some basic concepts of active robot vision, and then review representative work on systems,algorithms and applications from four perspectives:object reconstruction, scene reconstruction, object recognition, and pose ***, some potential directions are outlined for future work.

关键词： robot hardware planning

维普期刊数据库

Optoelectronic convolutional neural networks based on time-stretch method

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2021年第2期64卷 180-191页

作者： Yubin ZANG Minghua CHEN Sigang YANG Hongwei CHEN Beijing National Research Center for Information Science and Technology (bnrist) Department of Electronic EngineeringTsinghua University

In this paper, a new architecture of optoelectronic convolutional neural networks(CNNs) based on time-stretch method is proposed. In this loop-shaped structure mainly composed of fiber optical and electronic devices, computations of data from each layer of CNN which are carried by light pulses with high repetition rate can be accomplished in a serial way. Therefore, a 5-layer CNN with two convolution layers,two mean pooling layers and one fully-connected layer are implemented. Under the test of handwriting digit recognition, its accuracy can reach up to 95% under ideal circumstances. Tests under different relative noise levels have been conducted and analyzed as well.

关键词： convolutional neural networks time-stretch method artificial intelligence

Recent progress in single crystal perovskite X-ray detectors

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第3期67卷 5-19页

作者： Xiao LIU Jun REN Yu-Ang CHEN Xiangshun GENG Dan XIE Tian-Ling REN School of Integrated Circuits Tsinghua University Beijing National Research Center for Information Science and Technology (bnrist)

Perovskites have attracted extensive attention as radiation detection material due to their long carrier diffusion length and lifetime, high absorption coefficient, and flexible manufacturing process. Compared with polycrystalline structures, single crystal perovskites improve the performance of optoelectronic devices due to their low defect state density, better photoelectric characteristics, and chemical ***, we review the recent progress of single crystal perovskite X-ray detectors. First, we briefly introduced the basic concepts, detection mechanisms, figure of merits of perovskite X-ray detectors, and the preparation methods of single crystal perovskites. Then, we summarized the significant advancements in single crystal perovskite X-ray detectors in recent times. Finally, we discussed the critical challenges and some practicable solutions for developing high-performance X-ray detectors.

关键词： single crystal perovskites carrier transport radiation detection X-ray detectors X-ray imaging

同方期刊数据库

Near-field wideband beam training for ELAA with uniform circular array

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2024年第6期67卷 370-383页

作者： Yuhao CHEN Linglong DAI Department of Electronic Engineering Tsinghua UniversityBeijing National Research Center for Information Science and Technology (BNRist)

Extremely large-scale antenna array(ELAA) at millimeter wave(mmWave) and Terahertz(THz)band has been considered a key technology for combating high attenuation in high-frequency bands in future6G communications. Uniform circular arrays(UCAs) have attracted much attention because of their ability to provide flat beamforming gain at all angles. To realize efficient beamforming, beam training is widely used to acquire channel state information. However, with a large antenna number, the beam training overhead in ELAA systems becomes overwhelming. Moreover, with a large bandwidth, the beam defocus effect severely degrades beam training accuracy. To address these issues, this paper proposes a frequency-dependent focusing(FDF)-based beam training scheme to realize effective beam training in near-field wideband ELAA systems with UCA. Specifically, we first analyze the FDF property of UCA, where signals at different subcarriers can simultaneously focus on different distances. Then, by exploiting the FDF property to search different distances using different subcarriers simultaneously, we design a hierarchical codebook and propose an FDFbased beam training scheme. To reveal the effectiveness of the proposed scheme, we compare its necessary beam training overhead with that of existing schemes. Finally, the simulation results demonstrate that the proposed scheme can achieve accurate beam training in near-field wideband UCA systems with a low beam training overhead.

关键词： extremely large-scale antenna array (ELAA) uniform circular array (UCA) near-field wideband beam training