A5007149590- Low-power high-performance VLSI design
- Electromagnetic Compatibility and Noise Suppression
- Cell Image Analysis Techniques
- AI in cancer detection
- VLSI and FPGA Design Techniques
- Power Line Communications and Noise
- Anatomy and Medical Technology
- Terahertz technology and applications
- Electrostatic Discharge in Electronics
- Optimal Power Flow Distribution
- Advanced Memory and Neural Computing
- Photonic and Optical Devices
- Molecular Biology Techniques and Applications
- VLSI and Analog Circuit Testing
- Ferroelectric and Negative Capacitance Devices
- Advancements in Semiconductor Devices and Circuit Design
- Digital Filter Design and Implementation
- Generative Adversarial Networks and Image Synthesis
- Image Processing Techniques and Applications
- Optical and Acousto-Optic Technologies
- Advancements in PLL and VCO Technologies
- Semiconductor materials and devices
- Ferroelectric and Piezoelectric Materials
- Radio Frequency Integrated Circuit Design
- Smart Grid Security and Resilience
Zhejiang University
2022-2025
University of California, Los Angeles
2024
Samueli Institute
2023
Abstract Traditional histochemical staining of post-mortem samples often confronts inferior quality due to autolysis caused by delayed fixation cadaver tissue, and such chemical procedures covering large tissue areas demand substantial labor, cost time. Here, we demonstrate virtual autopsy using a trained neural network rapidly transform autofluorescence images label-free sections into brightfield equivalent images, matching hematoxylin eosin (H&E) stained versions the same samples. The...
Due to the mismatch between package scaling and relentless silicon technology scaling, limited power supply bumps have bear more stresses on bump reliability. A too high through-bump (TB) current may induce increased thermal mechanical issues, thereby damaging integrity of solder joint microstructure. Thus, it is critical analyze TB under different test scenarios at sign-off ensure integrity. Since full chip delivery verification (PDV) needs solve a linear system with billions nodes, then...
We present deep learning-based virtual staining of label-free autopsy tissue sections, eliminating severe autolysis-induced artifacts caused by delayed fixation inherent in traditional histochemical H&E staining.
High-speed serial links are fundamental to energy-efficient and high-performance computing systems such as artificial intelligence, 5G mobile automotive, enabling low-latency high-bandwidth communication. Transmitters (TXs) within these key signal quality, while their modeling presents challenges due nonlinear behavior dynamic interactions with links. In this paper, we propose LiTformer: a Transformer-based model for high-speed link TXs, non-sequential encoder Transformer decoder incorporate...
To meet the huge computing demands of data-intensive applications, resource constrained IoT devices have to address data migration issue, i.e., memory wall, in conventional Von Neumann architecture achieve desired energy efficiency. Recently, Computing-in-Memory (CiM) has been considered as a promising alternative overcome wall and extensively investigated for its deployment using SRAM, DRAM Non-Volatile Memory (NVM). Among various devices, FeFET (Ferroelectric FET) is an non-volatile device...
Histological examination is a crucial step in an autopsy; however, the traditional histochemical staining of post-mortem samples faces multiple challenges, including inferior quality due to autolysis caused by delayed fixation cadaver tissue, as well resource-intensive nature chemical procedures covering large tissue areas, which demand substantial labor, cost, and time. These challenges can become more pronounced during global health crises when availability histopathology services limited,...
Sign-off is a crucial step in the chip design flow to guarantee performance and reliability of chips prior tape-out. However, ever-growing integration density voltage scaling post-Moore era have made conventional sign-off inaccurate expensive, presenting challenges such as dynamic parasitics effects, large-scale circuits, multi-objective coupling. To overcome these challenges, this paper introduces concept agile sign-off, which has significant potential for validation, combining modeling,...
We present the first demonstration of unidirectional imaging that permits image formation along only one direction, from an input field-of-view to output field-of-view, while eliminating optical transmission in reverse direction. This imager is formed by diffractive layers composed isotropic linear materials spatially-coded with thousands phase features optimized using deep learning. experimentally tested our design a terahertz setup and 3D-printed layers, which revealed good agreement...
Fast Fourier Transform (FFT) is playing an important role in signal processing. This paper proposes a FFT -specific approximate multiplier design to improve the energy efficiency. The based on approximated twiddle factor and compressor reduce during partial product accumulation. Instead of complex arithmetic operations, can achieve reduced power consumption hardware cost with limited error.
A unidirectional imager would only permit image formation along one direction, from an input field-of-view (FOV) to output FOV B, and in the reverse path, be blocked. Here, we report first demonstration of imagers, presenting polarization-insensitive broadband imaging based on successive diffractive layers that are linear isotropic. These optimized using deep learning consist hundreds thousands phase features, which collectively modulate incoming fields project intensity onto FOV, while...