A5008650478- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Neural dynamics and brain function
- Neural Networks and Reservoir Computing
- Ferroelectric and Piezoelectric Materials
- Microwave Dielectric Ceramics Synthesis
- Transition Metal Oxide Nanomaterials
- Electronic and Structural Properties of Oxides
- Multiferroics and related materials
- Machine Learning and ELM
- Neuroscience and Neural Engineering
Chinese Academy of Sciences
2020-2024
University of Chinese Academy of Sciences
2022-2024
Institute of Microelectronics
2022-2024
ShanghaiTech University
2020-2021
Shanghai Institute of Ceramics
2020-2021
Intrinsic plasticity of neurons, such as spontaneous threshold lowering (STL) to modulate neuronal excitability, is key spatial attention biological neural systems. In-memory computing with emerging memristors expected solve the memory bottleneck von Neumann architecture commonly used in conventional digital computers and deemed a promising solution this bioinspired paradigm. Nonetheless, are incapable implementing STL neurons due their first-order dynamics. Here, second-order memristor...
Abstract Artificial visual systems that dynamically process spatiotemporal optoelectronic signals under complex real‐life environments bear a wide spectrum of edge applications. Despite significant progress in sensors and neuromorphic computing algorithms, developing can adapt to broad illumination range while retaining high performance, efficiency, low training costs remains challenge. Here, this work reports bioinspired in‐sensor reservoir (RC) for self‐adaptive recognition. By leveraging...
Abstract Time‐delayed reservoir computing with marked strengths of friendly hardware implementation and low training cost is regarded as a promising solution to realize time energy‐efficient series information processing thus receives growing attention. However, achieving sufficient number reproducible states remains significant challenge, which severely limits its performance. Here, an electric‐double‐layer‐coupled oxide‐based electrolyte‐gated transistor shared gate varying channel lengths...
Abstract The reservoir computing (RC) system, known for its ability to seamlessly integrate memory and functions, is considered as a promising solution meet the high demands time energy-efficient in current big data landscape, compared with traditional silicon-based systems that have noticeable disadvantage of separate storage computation. This review focuses on in-materio RC based nanowire networks (NWs) from perspective materials, extending devices applications. common methods used...
Time-delayed reservoir computing (RC) equipped with prominent superiorities such as easy training and friendly hardware implementation is identified a high-efficient answer to complex temporal tasks, thereby draws increasing attention. Oxygen ion-based oxide electrolyte-gated transistor (Ox-EGT) rich ion dynamic characteristics deemed promising candidate for RC. However, it still challenge produce the required RC implementation. Herein, we develop an Ox-EGT oxygen vacancy-electron-coupled...
Abstract The hardware implementation of artificial neural networks requires synaptic devices with linear and high‐speed weight modulation. Memristors as a candidate suffer from excessive write variation asymmetric resistance modulation that inherently rooted in their stochastic mechanisms. Thanks to controllable ion intercalation/deintercalation mechanism, electrolyte‐gated transistors (EGTs) hold prominent switching linearity low variation, thus have been the promising alternative for...
Hyperdimensional computing (HDC) is a brain-inspired computational framework that exploits hypervectors as an alternative to with numbers. In-memory implementation of HDC (IM-HDC) provides robust and energy-efficient approach process spatio-temporal (ST) signals since it significantly reduces data transfer overhead. However, previous IM-HDC suffers from the large peripheral circuit overheads assist component-wise hypervector operations. To address these issues, we propose voltage-mode...
Learning graph structured data from limited examples on-the-fly is a key challenge to smart edge devices. Here, we present the first chip-level demonstration of few-shot learning which homogeneously implements both controller and associative memory memory-augmented neural network using 1T1R resistive random-access (RRAM). Leveraging in-memory computing paradigm, validated high end-to-end accuracy 78% (GPU baseline 80%) robustness on node classification CORA dataset, while achieved 70-fold...
Li-ion-based electrolyte-gated transistors (Li-EGTs) have been extensively studied as synaptic devices due to their potential provide good analog switching of channel conductance, which is a desirable property for the emulation weight modulation. However, chemical activity lithium ion electrolytes during device fabrication detrimental stability Li-EGT and limits its application. In this work, we developed silica protective process fabrication. By continuously depositing electrolyte layer,...