A5061995397- Computational Drug Discovery Methods
- Power Systems and Technologies
- Ferroelectric and Negative Capacitance Devices
- Pharmacogenetics and Drug Metabolism
- Biosimilars and Bioanalytical Methods
- Generative Adversarial Networks and Image Synthesis
- Image Enhancement Techniques
- Power Systems Fault Detection
- Icing and De-icing Technologies
- High-Voltage Power Transmission Systems
- Transportation and Mobility Innovations
- Smart Grid Security and Resilience
- Topic Modeling
- Advanced Neural Network Applications
- Adversarial Robustness in Machine Learning
- Advanced Sensor and Control Systems
- Advanced Algorithms and Applications
- Advanced Image Fusion Techniques
- Speech Recognition and Synthesis
- Traffic Prediction and Management Techniques
- Human Pose and Action Recognition
- Thermal Analysis in Power Transmission
- Smart Grid and Power Systems
- Face recognition and analysis
- Advanced Memory and Neural Computing
Jiaxing Hengchuang Electric Design and Research Institute
2024
University of California, Santa Barbara
2020-2024
Guangxi University of Science and Technology
2011-2014
Spiking Neural Networks (SNNs) are emerging as a brain-inspired alternative to traditional Artificial (ANNs), prized for their potential energy efficiency on neuromorphic hardware. Despite this, SNNs often suffer from accuracy degradation compared ANNs and face deployment challenges due fixed inference timesteps, which require retraining adjustments, limiting operational flexibility. To address these issues, our work considers the spatio-temporal property inherent in SNNs, proposes novel...
Zero knowledge Neural Networks draw increasing attention for guaranteeing computation integrity and privacy of neural networks (NNs) based on zero-knowledge Succinct Non-interactive ARgument Knowledge (zkSNARK) security scheme. However, the performance zkSNARK NNs is far from optimal due to million-scale circuit with heavy scalar-level dependency. In this paper, we propose a type-based optimizing framework efficient NN inference, namely ZENO (ZEro network Optimizer). We first introduce...
With the rapid development of high-throughput technologies, parallel acquisition large-scale drug-informatics data provides significant opportunities to improve pharmaceutical research and development. One important application is purpose prediction small-molecule compounds with objective specifying therapeutic properties extensive purpose-unknown repurposing novel FDA-approved drugs. Such a problem extremely challenging because compound attributes include heterogeneous various feature...
Considering the influence that cycles of signal lamp have on waiting time, a bus scheduling model is presented in this paper based trade-off between cost operator and benefits passengers. In order to handle with low efficiency brought about by refused strategy, new fitness function designed according penalty then traditional genetic algorithm replaced quantum accelerate search optimal parameters further. The results experiment show method effective.
With the rapid growth in scale and complexity of large language models (LLMs), costs training inference have risen substantially. Model compression has emerged as a mainstream solution to reduce memory usage computational overhead. This paper presents Group Quantization Sparse Acceleration (\textbf{GQSA}), novel technique tailored for LLMs. Traditional methods typically focus exclusively on either quantization or sparsification, but relying single strategy often results significant...
Training generative adversarial networks (GAN) in a distributed fashion is promising technology since it contributed to training GAN on massive of data efficiently real-world applications. However, known be difficult train by SGD-type methods (may fail converge) and the may also suffer from amount communication cost. In this paper, we propose {distributed GANs algorithm with quantized gradient, dubbed DQGAN,} which first method gradient for GANs. The new trains based specific single machine...
Systematic approach for the transmission line positive sequence parameters, temperature, and sag based on wavelet analysis to detect error is developed in this work. Unbiased (random/Gaussian) such as, transient meter failures, malfunction, measurements captured during system transients, are inherently form of large abrupt change short duration a measurement-sequence. These should be detected before data used because their presence will lead insecure unstable power grid. The test results...
At first, this paper analyzed the key technologies of 4G communication system and main advantages in comparison with 3G rate, time-delay, spectral efficiency, etc. And then, it smart power grids’ demand for technology limitations traditional electric communication. By Comparing communication, demonstrated that is suitable grids. last, prospected application fields grids, also raised issues challenges enterprises will face.
The temperature of transmission line is an important parameter in the condition monitoring system on and its measured value affected by state sensors. General regression neural network (GRNN) was used to construct auto-detection for sensors line. Optimizing design network, error controlling effect testing were studied, also a method threshold sensor detection advanced. verified practical data from 220kv Sichuan Province proved be with good engineering application unit.
Infrared thermograph has been applied in electric equipment inspection widely, but the visual effects of infrared images are always undesirable. Considering limitation low luminance，low contrast images，an enhancement method based on fuzzy Renyi entropy and quantum genetic algorithm is presented this paper．Firstly，the contrast-sketching function [1] improved idea segmentation. Then, order to segment image, extend domain considering nature employed threshold image following maximal principle....
With the rapid development of high-throughput technologies, parallel acquisition large-scale drug-informatics data provides huge opportunities to improve pharmaceutical research and development. One significant application is purpose prediction small molecule compounds, aiming specify therapeutic properties extensive purpose-unknown compounds repurpose novel FDA-approved drugs. Such problem very challenging since compound attributes contain heterogeneous with various feature patterns such as...