检索结果-南通市图书馆

SWG: an architecture for sparse weight gradient computation

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2期67卷 302-321页

作者： Weiwei WU Fengbin TU Xiangyu LI Shaojun WEI Shouyi YIN School of Integrated Circuits Tsinghua University department of electronic and Computer Engineering The Hong Kong University of Science and Technology

On-device training for deep neural networks(DNN) has become a trend due to various user preferences and scenarios. The DNN training process consists of three phases, feedforward(FF), backpropagation(bP), and weight gradient(WG) update. WG takes about one-third of the computation in the whole training process. Current training accelerators usually ignore the special computation property of WG and process it in a way similar to FF/bP. besides, the extensive data sparsity existing in WG, which brings opportunities to save computation, is not well explored. Nevertheless, exploiting the optimization opportunities would meet three underutilization problems, which are caused by(1) the mismatch between WG data dimensions and hardware parallelism,(2) the full sparsity, i.e., the sparsity of feature map(Fmap),error map(Emap), and gradient, and(3) the workload imbalance resulting from irregular sparsity. In this paper, we propose a specific architecture for sparse weight gradient(SWG) computation. The architecture is designed based on hierarchical unrolling and sparsity-aware(HUSA) dataflow to exploit the optimization opportunities of the special computation property and full data sparsity. In HUSA dataflow, the data dimensions are unrolled hierarchically on the hardware architecture. A valid-data trace(VDT) mechanism is embedded in the dataflow to avoid the underutilization caused by the two-sided input sparsity. The gradient is unrolled in PE to alleviate the underutilization induced by output sparsity while maintaining the data reuse opportunities. besides, we design an intra-and inter-column balancer(IIbLC) to dynamically tackle the workload imbalance problem resulting from the irregular sparsity. Experimental results show that with HUSA dataflow exploiting the full sparsity, SWG achieves a speedup of 12.23× over state-of-the-art gradient computation architecture, Train Ware. SWG helps to improve the energy efficiency of the state-of-the-art training accelerator LNPU from

关键词： CNN training gradient computation sparsity architecture

Implantable probe with integrated reference electrode for in situ neural signal and calcium ion monitoring

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

bio-design and Manufacturing 2024年第4期7卷 591-595页

作者： Junyu Xiao Mengfei Xu Longchun Wang bin Yang Jingquan Liu National Key Laboratory of Advanced Micro and Nano manufacture Technology Shanghai Jiao Tong UniversityShanghai 200240China department of Micro/Nano electronics Shanghai Jiao Tong UniversityShanghai 200240China

Monitoring the electrophysiology activity of neurons and blood calcium signals can enable a better understanding of disease-related neural system ***,currently,in situ calcium ion monitoring tools are scarce and exhibit low integration and limited *** this letter,we propose an implantable probe with an integrated in situ Ag/AgCl reference electrode(ISA/ARE)that can monitor action potential(AP)and Ca^(2+) concentrations.

关键词： neural electrode enable

Dissolvable temporary barrier:a novel paradigm for flexible hydrogel patterning in organ-on-a-chip models

在线全文

学校读者我要写书评

暂无评论

bio-design and Manufacturing 2024年第2期7卷 153-166页

作者： Ding Wang Qinyu Li Chenyang Zhou Zhangjie Li Kangyi Lu Yijun Liu Lian Xuan Xiaolin Wang department of Micro/Nano electronics School of Electronic Information and Electrical EngineeringShanghai Jiao Tong UniversityShanghai 200240China department of Ophthalmology LKS Faculty of MedicineThe University of Hong KongHong Kong 999077China Institute of Medical Robotics Shanghai Jiao Tong UniversityShanghai 200240China National Key Laboratory of Advanced Micro and Nano manufacture Technology Shanghai Jiao Tong UniversityShanghai 200240China National Center for Translational Medicine(Shanghai)SHU branch Shanghai UniversityShanghai 200444China

A combination of hydrogels and microfluidics allows the construction of biomimetic three-dimensional(3D)tissue models in vitro,which are also known as *** hydrogel patterningwith awell-controlled spatial distribution is typically achieved by embedding sophisticated microstructures to act as a ***,these physical barriers inevitably expose cells/tissues to a less physiologically relevant microenvironment than in vivo ***,we present a novel dissolvable temporary barrier(DTb)strategy that allows robust and flexible hydrogel patterning with great freedom of design and desirable flow stimuli for cellular *** key aspect of this approach is the patterning of a water-soluble rigid barrier as a guiding path for the hydrogel using stencil printing technology,followed by a barrier-free medium perfusion after the dissolution of the *** and multiple tissue compartments with different geometries can be established using either straight or curved DTb *** effectiveness of this strategy is further validated by generating a 3D vascular network through vasculogenesis and angiogenesis using a vascularized microtumor *** a new proof-of-concept in vasculature-on-a-chip,DTb enables seamless contact between the hydrogel and the culture medium in closed microdevices,which is an improved protocol for the fabrication ofmultiorgan ***,we expect it to serve as a promising paradigm for organ-on-a-chip devices for the development of tumor vascularization and drug evaluation in the future preclinical studies.

关键词： Dissolvable temporary barrier Hydrogel patterning Microfluidics Organ-on-a-chip Vascularization

RSPSSL:A novel high-fidelity Raman spectral preprocessing scheme to enhance biomedical applications and chemical resolution visualization

在线全文

学校读者我要写书评

暂无评论

Light(Science & Applications) 2024年第2期13卷 366-386页

作者： Jiaqi Hu Gina Jinna Chen Chenlong Xue Pei Liang Yanqun Xiang Chuanlun Zhang Xiaokeng Chi Guoying Liu Yanfang Ye Dongyu Cui De Zhang Xiaojun yu Hong Dang Wen Zhang Junfan Chen Quan Tang Penglai Guo Ho-Pui Ho Yuchao Li Longqing Cong Perry Ping Shum State Key Laboratory of Optical Fiber and Cable manufacture Technology Guangdong Key Laboratory of Integrated Optoelectronics IntellisenseDepartment of EEESouthern University of Science and TechnologyShenzhen 518055China College of Optical and electronic Technology China Jiliang UniversityHangzhou 310018China department of Nasopharyngeal Carcinoma Sun Yat-sen University Cancer CenterState Key Laboratory of Oncology in South ChinaCollaborative Innovation Center for Cancer MedicineGuangdong Key Laboratory of Nasopharyngeal Carcinoma Diagnosis and TherapyGuangzhou 510060China department of Ocean Science and Engineering Southern University of Science and TechnologyShenzhen 518055China department of Nephrology Chaozhou People’s HospitalChaozhou 521011China Clinical Research design Division Sun Yatsen Memorial HospitalGuangzhouGuangdong 510120China School of Automation Northwestern Polytechnical UniversityXi’anShaanxi 710072China department of biomedical Engineering The Chinese University of Hong KongHong KongChina Guangdong Provincial Key Laboratory of Nanophotonic Manipulation Institute of NanophotonicsJinan UniversityGuangzhou 511443China

Raman spectroscopy has tremendous potential for material analysis with its molecular fingerprinting capability in many branches of science and *** is also an emerging omics technique for metabolic profiling to shape precision ***,precisely attributing vibration peaks coupled with specific environmental,instrumental,and specimen noise is *** Raman spectral preprocessing to remove statistical bias noise and sample-related errors should provide a powerful tool for valuable information ***,we propose a novel Raman spectral preprocessing scheme based on self-supervised learning(RSPSSL)with high capacity and spectral *** can preprocess arbitrary Raman spectra without further training at a speed of~1900 spectra per second without human *** experimental data preprocessing trial demonstrated its excellent capacity and signal fidelity with an 88%reduction in root mean square error and a 60%reduction in infinite norm(L__(∞))compared to established *** this advantage,it remarkably enhanced various biomedical applications with a 400%accuracy elevation(ΔAUC)in cancer diagnosis,an average 38%(few-shot)and 242%accuracy improvement in paraquat concentration prediction,and unsealed the chemical resolution of biomedical hyperspectral images,especially in the spectral fingerprint *** precisely preprocessed various Raman spectra from different spectroscopy devices,laboratories,and diverse *** scheme will enable biomedical mechanism screening with the label-free volumetric molecular imaging tool on organism and disease metabolomics profiling with a scenario of high throughput,cross-device,various analyte complexity,and diverse applications.

关键词： spectral scheme resolution

A graph neural network approach to the inverse design for thermal transparency with periodic interparticle system

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

Chinese Physics b 2024年第8期33卷 295-303页

作者：刘斌王译浠 department of electronic Information and Artificial Intelligence LeShan Normal UniversityLeShan 614099China department of Physics State Key Laboratory of Surface Physicsand Key Laboratory of Micro and Nano Photonic Structures(MOE)Fudan UniversityShanghai 200438China

Recent years have witnessed significant advances in utilizing machine learning-based techniques for thermal metamaterial-based structures and devices to attain favorable thermal transport *** the various thermal transport behaviors,achieving thermal transparency stands out as particularly desirable and *** earlier work demonstrated the use of a thermal metamaterial-based periodic interparticle system as the underlying structure for manipulating thermal transport behavior and achieving thermal *** this paper,we introduce an approach based on graph neural network to address the complex inverse design problem of determining the design parameters for a thermal metamaterial-based periodic interparticle system with the desired thermal transport *** work demonstrates that combining graph neural network modeling and inference is an effective approach for solving inverse design problems associated with attaining desirable thermal transport behaviors using thermal metamaterials.

关键词： thermal metamaterial thermal transparency inverse design machine learning graph neural net-work

A Novel Separability Objective Function in CNN for Feature Extraction of SAR Images

在线全文

学校读者我要写书评

暂无评论

Chinese Journal of electronics 2019年第2期28卷 423-429页

作者： GAO Fei WANG Meng WANG Jun YANG Erfu ZHOU Huiyu department of electronic and Information Engineering Beihang University department of design Manufacture and Engineering Management (DMEM) University of Strathclyde department of Informatics University of Leicester

Convolutional neural network(CNN) has become a promising method for Synthetic aperture radar(SAR) target recognition. Existing CNN models aim at seeking the best separation between classes, but rarely care about the separability of them. We performs a separability measure by analyzing the property of linear separability, and proposes an objective function for CNN to extract linearly separable features. The experimental results indicate the output features are linearly separable,and the classification results are comparable with the other state of the art techniques.

关键词： Synthetic aperture radar(SAR) Convolution neural network(CNN) Classification Linear separability Objective function

Physical mechanism of secondary-electron emission in Si wafers

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Chinese Physics b 2024年第4期33卷 677-681页

作者：赵亚楠孟祥兆彭淑婷苗光辉高玉强彭斌崔万照胡忠强 State Key Laboratory for Manufacturing Systems Engineering Electronic Materials Research LaboratoryKey Laboratory of the Ministry of EducationEngineering Research Center of Spin Quantum Sensor ChipsUniversities of Shaanxi ProvinceSchool of Electronic Science and EngineeringXi'an Jiaotong UniversityXi'an 710049China College of Aeronautics Nanjing University of Aeronautics and AstronauticsNanjing 210016China National Key Laboratory of Science and Technology on Space Microwave China Academy of Space TechnologyXi'an 710100China department of Physics School of Physics and Electronic InformationAnhui Normal UniversityWuhu 241000China

CMOS-compatible RF/microwave devices,such as filters and amplifiers,have been widely used in wireless communication ***,secondary-electron emission phenomena often occur in RF/microwave devices based on silicon(Si)wafers,especially in the high-frequency *** this paper,we have studied the major factors that influence the secondary-electron yield(SEY)in commercial Si wafers with different doping *** show that the SEY is suppressed as the doping concentration increases,corresponding to a relatively short effective escape depthλ.Meanwhile,the reduced narrow band gap is beneficial in suppressing the SEY,in which the absence of a shallow energy band below the conduction band will easily capture electrons,as revealed by first-principles ***,the new physical mechanism combined with the effective escape depth and band gap can provide useful guidance for the design of integrated RF/microwave devices based on Si wafers.

关键词： secondary-electron yield doping concentration escape depth Si wafer

Layer Hall effect induced by hidden berry curvature in antiferromagnetic insulators

在线全文

学校读者我要写书评

暂无评论

National Science Review 2024年第2期11卷 15-23页

作者： Rui Chen Hai-Peng Sun Mingqiang Gu Chun-bo Hua Qihang Liu Hai-Zhou Lu X.C.Xie Shenzhen Institute for Quantum Science and Engineering and department of Physics Southern University of Science and Technology(SUSTech) International Quantum Academy Institute for Theoretical Physics and Astrophysics University of Würzburg School of electronic and Information Engineering Hubei University of Science and Technology Guangdong Provincial Key Laboratory of Computational Science and Material design Southern University of Science and Technology Shenzhen Key Laboratory of Quantum Science and Engineering International Center for Quantum Materials School of PhysicsPeking University Collaborative Innovation Center of Quantum Matter CAS Center for Excellence in Topological Quantum Computation University of Chinese Academy of Sciences

The layer Hall effect describes electrons spontaneously deflected to opposite sides at different layers, which has been experimentally reported in the Mnbi2Te4thin films under perpendicular electric fields. Here, we reveal a universal origin of the layer Hall effect in terms of the so-called hidden berry curvature, as well as material design principles. Hence, it gives rise to zero berry curvature in momentum space but non-zero layer-locked hidden berry curvature in real space. We show that, compared to that of a trivial insulator, the layer Hall effect is significantly enhanced in antiferromagnetic topological insulators. Our universal picture provides a paradigm for revealing the hidden physics as a result of the interplay between the global and local symmetries, and can be generalized in various scenarios.

关键词： layerHall effect hidden berry curvature antiferromagnetic insulator axion insulator, quantum transport

Machine-learned wearable sensors for real-time hand-motion recognition: toward practical 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

National Science Review 2024年第2期11卷 231-263页

作者： Kyung Rok Pyun Kangkyu Kwon Myung Jin Yoo Kyun Kyu Kim Dohyeon Gong Woon-Hong Yeo Seungyong Han Seung Hwan Ko department of Mechanical Engineering Seoul National University IEN Center for Human-Centric Interfaces and Engineering Institute for Electronics and NanotechnologyGeorgia Institute of Technology School of Electrical and Computer Engineering Georgia Institute of Technology department of Chemical Engineering Stanford University department of Mechanical Engineering Ajou University George W.Woodruff School of Mechanical Engineering Georgia Institute of Technology Institute of Advanced Machinery and design(SNU-IAMD) Seoul National University

Soft electromechanical sensors have led to a new paradigm of electronic devices for novel motion-based wearable applications in our daily lives. However, the vast amount of random and unidentified signals generated by complex body motions has hindered the precise recognition and practical application of this technology. Recent advancements in artificial-intelligence technology have enabled significant strides in extracting features from massive and intricate data sets, thereby presenting a breakthrough in utilizing wearable sensors for practical applications. beyond traditional machine-learning techniques for classifying simple gestures, advanced machine-learning algorithms have been developed to handle more complex and nuanced motion-based tasks with restricted training data sets. Machine-learning techniques have improved the ability to perceive, and thus machine-learned wearable soft sensors have enabled accurate and rapid human-gesture recognition, providing real-time feedback to users. This forms a crucial component of future wearable electronics, contributing to a robust human–machine interface. In this review, we provide a comprehensive summary covering materials, structures and machine-learning algorithms for hand-gesture recognition and possible practical applications through machine-learned wearable electromechanical sensors.

关键词： wearable sensor soft electronics artificial intelligence machine learning human–machine interfaces gesture recognition