Efficient and accurate neural field reconstruction using resistive memory
the University of Hong Kong（香港大学）；ACCESS – AI Chip Center for Emerging Smart Systems, InnoHK Centers（ACCESS——新兴智能系统AI芯片中心（InnoHK中心））；Institute of Mind, the University of Hong Kong（香港大学心智研究所）；Institute of Microelectronics, Chinese Academy of Sciences（中国科学院微电子研究所）；University of Chinese Academy of Sciences（中国科学院大学）
Yifei Yu, Shaocong Wang, Woyu Zhang, Xinyuan Zhang, Xiuzhe Wu, Yangu He, Jichang Yang, Yue Zhang, Ning Lin, Bo Wang, Xi Chen, Songqi Wang, Xumeng Zhang, Xiaojua… 展开作者
Yifei Yu, Shaocong Wang, Woyu Zhang, Xinyuan Zhang, Xiuzhe Wu, Yangu He, Jichang Yang, Yue Zhang, Ning Lin, Bo Wang, Xi Chen, Songqi Wang, Xumeng Zhang, Xiaojuan Qi, Zhongrui Wang, Dashan Shang, Qi Liu, Kwang-Ting Cheng, Ming Liu
Human beings construct perception of space by integrating sparse observations into massively interconnected synapses and neurons, offering a superior parallelism and efficiency. Replicating this capability in AI finds wide applications in medical imaging, AR/VR, and embodied AI, where input data is often sparse and computing resources are limited. However, traditional signal reconstruction methods on digital computers face both software and hardware challenges. On the software front, difficulties arise from storage inefficiencies in conventional explicit signal representation. Hardware obstacles include the von Neumann bottleneck, which limits data transfer between the CPU and memory, and the limitations of CMOS circuits in supporting parallel processing. We propose a systematic approach with software-hardware co-optimizations for signal reconstruction from sparse inputs. Software-wise, we employ neural field to implicitly represent signals via neural networks, which is further compressed using low-rank decomposition and structured pruning. Hardware-wise, we design a resistive memory-based computing-in-memory (CIM) platform, featuring a Gaussian Encoder (GE) and an MLP Processing Engine (PE). The GE harnesses the intrinsic stochasticity of resistive memory for efficient input encoding, while the PE achieves precise weight mapping through a Hardware-Aware Quantization (HAQ) circuit. We demonstrate the system's efficacy on a 40nm 256Kb resistive memory-based in-memory computing macro, achieving huge energy efficiency and parallelism improvements without compromising reconstruction quality in tasks like 3D CT sparse reconstruction, novel view synthesis, and novel view synthesis for dynamic scenes. This work advances the AI-driven signal restoration technology and paves the way for future efficient and robust medical AI and 3D vision applications.
Pruning random resistive memory for optimizing analogue AI
The University of Hong Kong（香港大学）；Institute of Microelectronics, Chinese Academy of Sciences（中国科学院微电子研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Institute of Mind, The University of Hong Kong（香港大学心智研究所）；ACCESS – AI Chip Center for Emerging Smart Systems, InnoHK Centers, Hong Kong Science Park（ACCESS – 新兴智能系统AI芯片中心，InnoHK中心，香港科学园）
Resistive Memory-based Neural Differential Equation Solver for Score-based Diffusion Model
the University of Hong Kong（香港大学）；ACCESS – AI Chip Center for Emerging Smart Systems, InnoHK Centers（ACCESS——新兴智能系统AI芯片中心（InnoHK中心））；Institute of the Mind, the University of Hong Kong（香港大学心智研究所）
Ultra Low Power 3D-Embedded Convolutional Neural Network Cube Based on $α$-IGZO Nanosheet and Bi-Layer Resistive Memory
在本文中，我们提出和评估的3D嵌入式神经形态计算块的性能的基础上，铟镓锌氧化物（$α$ -IGZO）的纳米片晶体管和双层电阻存储器件。我们制作了具有Ta $_2$ O $_5$和Al $_2$ O $_3$层的双层阻变随机存取存储器（RRAM）器件。已对器械进行了表征和建模。RRAM和基于$α$ -IGZO的嵌入式纳米片结构的紧凑模型已被用于评估具有用于神经形态应用的RRAM的8个垂直堆叠的基于$α$ -IGZO的纳米片层的系统级性能。该模型考虑了具有均匀位线（BL）、选择线（SL）和字线（WL）电阻的设计空间。最后，我们用我们提出的8层堆叠的基于纳米片的嵌入式存储器模拟了加权求和操作，并评估了VGG-16卷积神经网络（CNN）用于Fashion-MNIST和CIFAR-10数据识别的性能，其在器件变化中分别产生了92%和75%的准确度。
Edge AI without Compromise: Efficient, Versatile and Accurate Neurocomputing in Resistive Random-Access Memory
Stanford University（斯坦福大学）；University of California San Diego（加利福尼亚大学圣地亚哥分校）；Tsinghua University（清华大学）；University of Notre Dame（圣母大学）；University of Pittsburgh（匹兹堡大学）
Minimizing energy dissipation during programming of resistive switching memory devices using their dynamical attractor states
Institute of Physics, Opole University（奥波莱大学物理研究所）；Department of Electronics and Telecommunications Politecnico di Torino（都灵理工大学电子与电信系）；Department of Physics and Astronomy, University of South Carolina（南卡罗来纳大学物理与天文系）
Modern analog computing for solving differential and matrix equations
Institute for Artificial Intelligence, School of Integrated Circuits, Peking University, Beijing Advanced Innovation Center for Integrated Circuits（人工智能研究院，集成电路学院，北京大学，北京集成电路先进创新中心）；Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano（电子、信息与生物工程系，米兰理工大学）；IBM Research Europe（IBM欧洲研究院）
Multi-Level Resistive Synapses for On-Chip Neural Networks: A Physics-Based Design of a Memristive Crossbar Fabric with Quasi-Continuous Conductance States
