A5069830698- Synthetic Aperture Radar (SAR) Applications and Techniques
- Advanced SAR Imaging Techniques
- Domain Adaptation and Few-Shot Learning
- CCD and CMOS Imaging Sensors
- Analog and Mixed-Signal Circuit Design
- Remote-Sensing Image Classification
- Advanced Memory and Neural Computing
- Viral Infections and Outbreaks Research
- Underwater Acoustics Research
- Digital Filter Design and Implementation
- Soil Moisture and Remote Sensing
- Ferroelectric and Negative Capacitance Devices
- Geological and Geophysical Studies
- Advancements in Semiconductor Devices and Circuit Design
- Viral gastroenteritis research and epidemiology
- Neural Networks and Reservoir Computing
Xi’an University of Posts and Telecommunications
2023-2024
The University of Texas at Austin
2022-2023
Tsinghua University
2022-2023
National Yang Ming Chiao Tung University
2022
National University of Singapore
2022
University of Utah
2022
Carnegie Mellon University
2022
University of Twente
2022
Nagoya University
2022
Texas Instruments (United States)
2022
The article presents a charge-domain computing ternary neural network (TNN) classifier with complete four-layer (NN) on chip. proposed provides 1.5-b resolution (0/+1/−1) for weights and activations, leading to 3.9× fewer operations (OPs) per inference than binary (BNN) the same Modified National Institute of Standards Technology (MNIST) accuracy. multiply-and-accumulate (MAC) is implemented by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML"...
Although deep learning-based methods have made remarkable achievements in polarimetric synthetic aperture radar (PolSAR) image classification, these require a large number of labeled samples. However, for PolSAR it is difficult to obtain samples, which requires extensive human labor and material resources. Therefore, new classification method based on multi-scale contrastive learning proposed, can achieve good results with only small During the pre-training process, we propose network model...
With the development of science and technology, although more polarimetric synthetic aperture radar (PolSAR) data are collected, marking PolSAR still requires a lot costs. Moreover, datasets between different domains have class distribution shift problem, which reduces reusability labeled samples cross-domain images. To address this issue, article proposed an unsupervised domain adaptation (UDA) network based on coordinate attention (CA) weighted clustering. Firstly, adversarial UDA with...
This paper presents an end-to-end successive-approximation-register (SAR) analog-to-digital converter (ADC) compiler that generates design solutions from top-level specification to GDSII layout with a short turnaround time of 5 hours. Two prototype SAR ADCs operating at 1MS/s and 80MS/s are compiled in 40nm CMOS. Measurement results demonstrate wide conversion range, presenting both analog technology-limited performances. The requires minimum manual involvement can significantly boost...
CTDSMs with high resolution and bandwidth greater than 200kHz are needed in industrial, medical, automotive applications. Such performance demands very low noise distortion. The distortion have to be suppressed even further advanced technologies due the voltage headroom. A major challenge of design is large area cost DAC loop filters. main feedback RDAC occupies a [1]. 1st-order data weighted average (DWA) used but has limited mismatch error suppression. There also kink SNDR plot [1] at...