A5088135049- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Neuroscience and Neural Engineering
- Neural Networks and Reservoir Computing
- CCD and CMOS Imaging Sensors
- Photoreceptor and optogenetics research
- Semiconductor materials and devices
- Natural Language Processing Techniques
- Topic Modeling
- Full-Duplex Wireless Communications
- Neural dynamics and brain function
- Cooperative Communication and Network Coding
- Transition Metal Oxide Nanomaterials
- Advanced Wireless Communication Technologies
- Blind Source Separation Techniques
- Hand Gesture Recognition Systems
- Human Pose and Action Recognition
- Robotics and Sensor-Based Localization
- Plant and Fungal Species Descriptions
- Plant-derived Lignans Synthesis and Bioactivity
- TiO2 Photocatalysis and Solar Cells
- Pigment Synthesis and Properties
- Electronic and Structural Properties of Oxides
- Text Readability and Simplification
- Software Reliability and Analysis Research
Peking University
2017-2024
Quality and Reliability (Greece)
2022
Institute of Microelectronics
2017-2021
Beijing Institute of Technology
2011
In this letter, we experimentally demonstrated a novel resistive device with hybrid switching mode that can be alternated between volatile threshold and non-volatile switching. The consists of dual-functional layers VO2 /HfO2 sandwiched by symmetrical TiN electrodes. A >20 unified ratio for selectivity memory window is obtained. Owing to the stable behavior HfO2 insulator-metal transition VO2, shows excellent uniform parameters in both modes high on-state current density (1E4 A/cm <sup...
With the rapid development of Large Language Models (LLMs), increasing attention has been paid to their safety concerns. Consequently, evaluating LLMs become an essential task for facilitating broad applications LLMs. Nevertheless, absence comprehensive evaluation benchmarks poses a significant impediment effectively assess and enhance In this work, we present SafetyBench, benchmark LLMs, which comprises 11,435 diverse multiple choice questions spanning across 7 distinct categories Notably,...
For the first time, this work investigated time-dependent variability (TDV) in RRAMs and its interaction with RRAM-based analog neuromorphic circuits for pattern recognition. It is found that even are well trained, TDV effect can introduce non-negligible recognition accuracy drop during operating condition. The impact of on increases when higher resistances used circuit implementation, challenging future low power operation. In addition, cannot be suppressed by either scaling up more...
Abstract 3D integration of vertical resistive random access memory (VRRAM) with organic materials is promising for ultra‐high density flexible data storage. However, it extremely challenging to heterogeneously fabricate an VRRAM due complicated issues such as chemical/thermal robustness, compatibility organic/inorganic materials, and processes stacking/patterning multi‐layer organic/metal films. Herein, based on parylene‐C experimentally demonstrated a standard complementary metal oxide...
Point clouds have important applications, such as computer vision, autonomous driving, and robotics. However, point cloud recognition task encounters challenges related to data input equivalence computational overhead in conventional hardware. To address these issues, we propose a Hierarchical Synchronous Parallel Architecture (HSPA) for resistive-random-access-memory (RRAM) based computing-in-memory (CIM), which significantly enhances the parallelism of with DeepSets network. The processing...
This work demonstrates the emulation of biphasic plasticity in electrical synapses by integrating Ag-based memristor with a photosensitive element to form an optical pre-processing unit (OPU).
The device non-idealities and array thermal cross-talk will inevitably impair the performance of RRAM-based storage neuromorphic systems. In this paper, we present modeling non-idealities, including parameter variations time-dependent fluctuation (TDV), simulation crosstalk. We also investigate impact issues on computing from device, array, algorithm perspectives, providing guidelines for technology-array-algorithm co-optimization.
Neurons are basic elements of the human brain to transmit and process various kinds information. Besides synaptic plasticity, neuronal intrinsic plasticity (NIP) plays a vital role in improving stabilizing neural circuit for learning adaptation environment. In this article, we emulate NIP behavior by combining hybrid VO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> <inline-formula xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math...
Compressed sensing with a tunable random sampling strategy has great potential in reducing the energy/time consumption during constant acquisition of external information, thereby it is considered one most promising strategies edge-terminal. In this letter, novel stochastic oscillator (TSO) based on VO2/TaOx structure was proposed and employed as core controlling device for compressed sensing. The TSO jointly governed by metal-insulator-transition (IMT) threshold switching VO2 layer...
Speech command interaction has drawn much attention in smart application market. Many of previous chips achieve an ultra-low power consumption at the cost a certain accuracy loss, and they are designed only for fixed speech recognition tasks, which is inflexible restrains further development. Here, we demonstrate configurable ASIC with ultra-high fabricated by TSMC commercial 180-nm CMOS technology. In this chip, Mel-Frequency Cepstrum Coefficients (MFCCs) used as features One-Dimension...
This brief presents a novel spatio-temporal domain adaptation (DA) method that is unsupervised and event-driven for dynamic vision sensor (DVS) gesture recognition. realizes the transfer between source target data in both spatial temporal dimension without need of labels. Specifically, it consists deep spiking neural network (SNN)-based feature extractor, label predictor discriminator. A time-space gradient reversal layer responsible building bridge discriminator which essential to alignment...
Abstract Recent studies indicate that synaptic scaling is a vital mechanism to solve instability risks brought by the positive feedback of weight change related with standalone Hebbian plasticity. There are two kinds in neural network, including local and global scaling, both important for stabilizing function. In this paper, first time, emulated based on MoS 2 neuristor. The first‐principle calculation reveals achieved neuristor associated an internal residual Li + ‐related weak dynamical...
Random telegraph noise (RTN) in resistive random access memory (RRAM) introduces variation resistance which might cause errors RRAM based logic circuits. In this paper, we find multiple RTN patterns devices and build a simulation model. Using model, immunity of IMP circuit is evaluated method to suppress RTN's effect proposed.
Neuronal oscillation, as the fundamental component of information processing and transmission in brain, plays a pivotal role human cognition, learning, memory. In this article, we introduce novel silicon neuron device (SND) featuring light pulse modulation, aiming to emulate oscillatory dynamics observed biological neurons. The SND fabrication utilizes 0.18 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math...
In this paper, we consider a multi-relay Demodulate-and-Forward (DMF) cooperation system, in which the high order superposition modulation (SM), 16-QAM (16-QAM SM), is employed. For given new relay selection scheme with constellation rearrangement (CR) technique proposed, based on optimal total mutual information (MI) criterion. Firstly find bit-symbol mappings for all relays. And then source chooses best relay, maximizes MI, to cooperate with. We assume that channel state (CSI) available. A...
The RRAM-based array is one of the most promising core functional primitives to accelerate inference process neural networks. However, stress-induced disturbance can cause a significant accuracy drop during where input vectors with different voltage levels are fed device. This kind disturb hardly be avoided by optimizing fabrication process. Here, we investigate this phenomenon based on TaOx-based devices electrodes. results indicate that mainly appears in intermediate resistance states when...
Memristor has attracted extensive attention in recent years because of its capability to act as artificial synapse neuromorphic network. In this paper, by combining a designed memristive and resistive synapse, novel neural network circuit with three neurons one operational module been constructed. With the memristor device whose resistance can be purposely changed according time interval between pre- post-spikes, weight that connected two modified required. What's more, transmission signal...
Beyond the emerging device based on a single physical mechanism, we have proposed new type of with hybrid mechanisms—the unified insulator-metal transition (IMT) and resistive switching (RS) device. Several information processing paradigms are established this novel In paper, reviewed applications IMT-RS devices, highlighting their potential for systems.