A5068191980- Ferroelectric and Negative Capacitance Devices
- Advanced Memory and Neural Computing
- Neural Networks and Reservoir Computing
- Neural dynamics and brain function
- Neuroscience and Neural Engineering
- Functional Brain Connectivity Studies
- Biomedical Text Mining and Ontologies
- Machine Learning and ELM
- Energy Load and Power Forecasting
- Electric Power System Optimization
- Integrated Energy Systems Optimization
- Energy, Environment, and Transportation Policies
- Climate Change Policy and Economics
- Smart Grid and Power Systems
- Optimal Power Flow Distribution
- Power Systems and Technologies
- Natural Language Processing Techniques
- Power Systems and Renewable Energy
- Microgrid Control and Optimization
- Smart Grid Energy Management
- Energy, Environment, Economic Growth
- Topic Modeling
University of California, San Diego
2019-2021
Capital Medical University
2019
Phase change memory (PCM) has been considered as one of the most promising emerging non-volatile memories for in-memory computing neural networks. In this letter, we investigate impact resistance drift and its statistical variations on two widely-used artificial network (ANN) models, multi-layer perceptron (MLP), convolutional (CNN). We employ experimentally measured characteristics into ANN models to accurately model weight updates represented by PCM synaptic devices. Our results suggest...
In recent years, Spin-Transfer-Torque Magnetic Random Access Memory (STT-MRAM) has been considered as one of the most promising non-volatile memory candidates for in-memory computing. However, system-level performance gains using STT-MRAM computing at deeply scaled nodes have not assessed with respect to more mature technologies. this letter, we present perpendicular magnetic tunnel junction (pMTJ) devices 28nm and 7nm. We evaluate convolutional neural network (CNN) inference arrays in...
Off-chip memory access is the primary bottleneck toward accelerating neural network operations and reducing energy consumption. In-memory training computation using emerging nonvolatile memories (eNVMs) have been proposed to address this problem. However, a small number of conductance states limit in-memory online learning performance. Here, we introduce device-algorithm co-design approach its application phase change (PCM) for improving accuracy. We present an adaptive quantization method,...
Objective. Electrical recordings of neural activity from brain surface have been widely employed in basic neuroscience research and clinical practice for investigations circuit functions, brain-computer interfaces, treatments neurological disorders. Traditionally, these potentials believed to mainly reflect local activity. It is not known how informative the locally recorded are activities across multiple cortical regions.Approach. To investigate that, we perform simultaneous electrical...
In recent years, China's electricity market reform has continued to deepen, and the construction of been accelerating. According national power pilot policy, southern regional will start with Guangdong spot market, then gradually develop market. this paper, realization method integration is analysed, several paths for provincial into are put forward according characteristics Southern system.
Abstract In order to improve the adaptability of power market transaction uncertainty renewable energy, multi-scenario forecast model energy is studied and an optimal decision-making method for based on proposed. Multi-scenario a result composed multiple curves, which uncertain in nature. The takes lowest expected cost loss as optimization objectives comprehensively considers constraints including balance, network transmission capacity so on. Finally, case study IEEE-30 buses system shows...
Abstract In order to solve the transaction clearing problem caused by insufficient reactive power support ability of renewable energy, regulation characteristics energy are studied and a distribution network method considering is proposed. Wind power, photovoltaic other sources generally connected grid through electronic equipment. Within capacity control range, there two adopted modes including constant voltage power. According operation needs network, model aiming at lowest purchase cost A...
Abstract Objective Electrical recordings of neural activity from brain surface have been widely employed in basic neuroscience research and clinical practice for investigations circuit functions, brain-computer interfaces, treatments neurological disorders. Traditionally, these potentials believed to mainly reflect local activity. It is not known how informative the locally recorded are activities across multiple cortical regions. Approach To investigate that, we perform simultaneous...