A5078107730- Advanced Photonic Communication Systems
- Optical Network Technologies
- Wireless Signal Modulation Classification
- Remote-Sensing Image Classification
- Advanced Neural Network Applications
- Domain Adaptation and Few-Shot Learning
- Photonic and Optical Devices
- Advanced Fiber Laser Technologies
- Advanced MIMO Systems Optimization
- Network Security and Intrusion Detection
- Topology Optimization in Engineering
- Antenna Design and Optimization
- Artificial Immune Systems Applications
- Photonic Crystal and Fiber Optics
- Advanced Fiber Optic Sensors
- Anomaly Detection Techniques and Applications
Southwest Jiaotong University
2022-2024
University of the Witwatersrand
2023
A*STAR Graduate Academy
2023
Hohai University
2018
Object detection in remote sensing images relies on a large amount of labeled data for training. However, the increasing number new categories and class imbalance make exhaustive annotation impractical. Few-shot object (FSOD) addresses this issue by leveraging meta-learning seen base classes fine-tuning novel with limited samples. Nonetheless, substantial scale orientation variations objects pose significant challenges to existing few-shot methods. To overcome these challenges, we propose...
Aiming at the problems such as accuracy and time complexity of existing abnormal network traffic detection methods, an method based on improved LOF algorithm is proposed in this paper. By analyzing characteristics parameter traditional algorithm, model combined with DBSCAN to achieve adaptive dynamic adjustment changed data optimized for scenarios. In addition, optimization large amounts reduces consumption model. The experimental results show that great value practical application.
To improve the receiver sensitivity of digitized radio-over-fiber (DRoF) transmission system, a vector quantization scheme based on probabilistic codeword shaping (PQCS) is proposed. The PQCS performs bits (QBs) rematching original codebook to optimize proportion distribution level '0' and '2' in 4-Pulse Amplitude Modulation (PAM-4) for improving system sensitivity. A 16-Quadrature (16-QAM) DRoF using intensity-modulation/direct-detection employed experimentally verify proposed scheme....
We designed a multi-channel comb filter generation scheme with nearly consistent channel numbers and free spectral range (FSR) in standard 220-nm-thick silicon-on-insulator (SOI) technology. This relies on the formation of optical microcavities using Bragg grating structures, which serve as reflectors. By precisely designing path length microcavity, we can generate filters specific number channels. Using this scheme, developed tested three devices to implement two-channel, three-channel,...
A photonic-assisted approach is proposed for modulation format identification (MFI) on radio frequency (RF) signals under low sampling rate. In this approach, a interferometer (PAI) designed computation-free data augmentation by transforming the signal's phase and variations into format-sensitive amplitude features. fully connected neural network (FCNN) used to implement end-to-end MFI. An experiment conducted of amplitude-shift keying (ASK), binary phase-shift (BPSK), quadrature (QPSK),...
Object detection in remote sensing images relies on a large amount of labeled data for training. However, the increasing number new categories and class imbalance make exhaustive annotation impractical. Few-shot object (FSOD) addresses this issue by leveraging meta-learning seen base classes fine-tuning novel with limited samples. Nonetheless, substantial scale orientation variations objects pose significant challenges to existing few-shot methods. To overcome these challenges, we propose...
The on-chip pulse shaper is a key device that enables ultra-high-speed, ultra-wideband linear signal processing in compact size with low power consumption. We propose pulsed time-domain waveform shaping method based on chirped period Bragg gratings and contradirectional couplers, allowing flexible control of the grating's to achieve frequency-specific reflections reflection strengths. demonstrate two types sawtooth generation analyze relationship between waveforms chirp period. Compared...
A photonic-assisted architecture is used to implement self-interference on radio-frequency (RF) signals create modulation-biased amplitude patterns. The interfered are identified from their modulation formats using a convolutional neural network (CNN). In the proof-of-concept experiments, average accuracy reaches 97.06% for with 10 at 1 G Baud.