A5102926008- Neural Networks and Reservoir Computing
- Advanced Memory and Neural Computing
- Optical Network Technologies
- Photonic and Optical Devices
- Neural dynamics and brain function
- Adaptive Control of Nonlinear Systems
- Nonlinear Dynamics and Pattern Formation
- Control and Dynamics of Mobile Robots
- Underwater Vehicles and Communication Systems
- Guidance and Control Systems
- Photoreceptor and optogenetics research
- Semiconductor Lasers and Optical Devices
- Digital Marketing and Social Media
- CCD and CMOS Imaging Sensors
- Anesthesia and Sedative Agents
- Slime Mold and Myxomycetes Research
- Advanced Optical Sensing Technologies
- Power Systems and Renewable Energy
- Neural Networks and Applications
- stochastic dynamics and bifurcation
- Robot Manipulation and Learning
- Dynamics and Control of Mechanical Systems
- Chaos-based Image/Signal Encryption
- Advanced Decision-Making Techniques
- Recommender Systems and Techniques
Xidian University
2018-2025
Baoding People's Hospital
2024
Shanghai University of Finance and Economics
2023
Dalian Maritime University
2019-2020
Southwest Jiaotong University
2019
Abstract The explosive growth of data and information has motivated various emerging non-von Neumann computational approaches in the More-than-Moore era. Photonics neuromorphic computing attracted lots attention due to fascinating advantages such as high speed, wide bandwidth, massive parallelism. Here, we offer a review on optical neural our research groups at device system levels. photonics neuron synapse plasticity are presented. In addition, introduce several architectures algorithms...
Photonic neuromorphic computing has emerged as a promising approach to building low-latency and energy-efficient non-von Neuman system. A photonic spiking neural network (PSNN) exploits brain-like spatiotemporal processing realize high-performance computing. However, the nonlinear computation of PSNN remains significant challenge. Here, we propose fabricate neuron chip based on an integrated Fabry–Perot laser with saturable absorber (FP-SA). The neuron-like dynamics including temporal...
Neuromorphic photonic computing has emerged as a competitive paradigm to overcome the bottlenecks of von-Neumann architecture. Linear weighting and nonlinear spike activation are two fundamental functions spiking neural network (PSNN). However, they separately implemented with different materials devices, hindering large-scale integration PSNN. Here, we propose, fabricate experimentally demonstrate neuro-synaptic chip enabling simultaneous implementation linear based on distributed feedback...
Multi-channel physical random bit generator (RBG) based on a chaotic semiconductor lasers (SLs) network is proposed and experimentally demonstrated. In experiments, small ring consisting of three SLs mutually coupled with heterogeneous delays constructed to obtain temporal waveforms. The results show that the time-delay signature intensities in all can be well concealed over wide range parameters. Through linear combination outputs, seven channels entropy sources obtained. Furthermore, by...
Chaos, occurring in a deterministic system, has permeated various fields such as mathematics, physics, and life science. Consequently, the prediction of chaotic time series received widespread attention made significant progress. However, many problems, high computational complexity difficulty hardware implementation, could not be solved by existing schemes. To overcome we employ system vertical-cavity surface-emitting laser (VCSEL) mutual coupling network to generate through optical...
Photonic reservoir computing (RC) is a simple and efficient neuromorphic framework for human cortical circuits, which featured with fast training speed low cost. time delay RC, as hardware implementation method of has attracted widespread attention. In this paper, we present experimentally demonstrate RC system based on Fabry Perot (FP) laser multiple tasks processing. Here, the various are attempted to perform in parallel longitudinal modes FP laser. It found that can successfully handle...
In the field of information processing, traditional computing methods encounter limitations in handling increasingly complex tasks and meeting growing performance requirements. Reservoir computing, as a new paradigm, has demonstrated excellent time series prediction tasks. However, photonic reservoir still needs improvement certain aspects, such high computational complexity relatively high-power consumption for processing. our work, spiking system based on single distributed feedback laser...
We propose to realize photonic spike timing dependent plasticity (STDP) by using a vertical-cavity semiconductor optical amplifier (VCSOA) subject dual pulse injections. The computational model of the STDP is presented for first time based on well-known Fabry-Pérot approach. Through numerical simulations, dependences bias current VCSOA and input powers are analyzed carefully. Besides, effect initial wavelength detuning also explored. It found that, scheme requires much lower power obtain...
We propose and demonstrate experimentally numerically a network of three globally coupled semiconductor lasers (SLs) that generate triple-channel chaotic signals with time delayed signature (TDS) concealment. The effects the coupling strength bias current on concealment TDS are investigated. generated further applied to reinforcement learning, parallel scheme is proposed solve multiarmed bandit (MAB) problem. influences mutual correlation between from different channels, sampling interval...
We propose a modified supervised learning algorithm for optical spiking neural networks, which introduces synaptic time-delay plasticity on the basis of traditional weight training. Delay is combined with remote method that incorporated photonic spike-timing-dependent plasticity. A spike sequence task implemented via proposed found to have better performance than weight-based method. Moreover, also applied two benchmark data sets classification. In simple network structure only few neurons,...
Spiking neural network (SNN) have attracted lots of attention due to its event-driven nature and powerful computation capability. However, it is still limited simple task the training difficulty. In this work, we propose a hybrid architecture photonic convolutional spiking (PCSNN) realize speech recognition task. PCSNN, feature extraction realized by convolution SNN with unsupervised learning algorithm, classification modified time-based supervised algorithm. The TIDIGITS dataset used test...
Hardware implementation of reservoir computing (RC), which could reduce the power consumption machine learning and significantly enhance data processing speed, holds potential to develop next generation hardware devices chips. Due existing solution only implementing layers, information speed photonics RC system are limited. In this paper, a photonic VMM-RC based on single Vertical Cavity Surface Emitting Laser (VCSEL) with two Mach Zehnder modulators (MZMs) has been proposed. Unlike previous...
Abstract Visual obstacle avoidance is widely applied to unmanned aerial vehicles (UAVs) and mobile robot fields. A simple system architecture, low power consumption, optimized processing, real‐time performance are extremely needed due the limited payload of some mini UAVs. To address these issues, an harnessing rate encoding features a photonic spiking neuron based on Fabry–Pérot (FP) laser proposed, which simulates monocular vision. Here, time collision used describe distance obstacles. The...
This paper studies the problem of spatial path-following control underactuated autonomous underwater vehicles (AUVs) with multiple uncertainties and input saturation taken into account. Initially, reduced-order extended state observes (ESOs) are introduced to estimate compensate all lumped due model parameters perturbations, unmodeled dynamics, environmental disturbances, nonlinear hydrodynamic damping terms. Furthermore, strategy is established by combining backstepping, integral sliding...
Photonic spiking neural networks (SNN) have the advantages of high power efficiency, bandwidth and low delay, but limitations are encountered in large-scale integration. The silicon photonics platform is a promising candidate for realizing photonic SNN because it compatible with current mature CMOS platforms. Here, we present an architecture which consists neuron, spike timing dependent plasticity (STDP) weight configuration that all based on micro-ring resonators (MRRs), via taking...
We present a simple experimental approach based on photonic time delay reservoir computing (RC) system for modulation format recognition. Here an optically injected vertical cavity surface emitting laser with single feedback is trained the cross sequence of instantaneous characteristics signals. Three widely used formats including on–off keying, binary phase shift and frequency where optical signal-to-noise ratio varies from 4 dB to 36 are considered. Besides, we propose post-processing...
Dendrites, branches of neurons that transmit signals between synapses and soma, play a vital role in spiking information processing, such as nonlinear integration excitatory inhibitory stimuli. However, the investigation dendrites photonic fabrication including dendritic neural networks (SNNs) remain open problems. Here, we fabricate integrate two one soma single Fabry–Perot laser with an embedded saturable absorber (FP-SA) neuron to achieve Note intrinsic electrodes gain section (SA) FP-SA...
Spiking neural networks (SNNs) utilize brain-like spatiotemporal spike encoding for simulating brain functions. Photonic SNN offers an ultrahigh speed and power efficiency platform implementing high-performance neuromorphic computing. Here, we proposed a multi-synaptic photonic SNN, combining the modified remote supervised learning with delay-weight co-training to achieve pattern classification. The impact of connections robustness network were investigated through numerical simulations. In...
Reservoir computing (RC), especially photonic RC based on a single semiconductor laser with feedback loop, as method of machine learning, shows excellent performance in time series prediction and classification tasks. The faster the processing speed, shorter loop should be employed. However, system is not ideal enough due to limited reservoir nodes caused by short loop. To overcome this drawback, it necessary challenging develop integrated RC. In paper, we propose an neuromorphic...