A5027132950- Advancements in Photolithography Techniques
- Software Engineering Research
- Low-power high-performance VLSI design
- Parallel Computing and Optimization Techniques
- Integrated Circuits and Semiconductor Failure Analysis
- Electrostatic Discharge in Electronics
- VLSI and Analog Circuit Testing
- Ferroelectric and Negative Capacitance Devices
- VLSI and FPGA Design Techniques
- Machine Learning in Materials Science
- Time Series Analysis and Forecasting
- Music and Audio Processing
- Physical Unclonable Functions (PUFs) and Hardware Security
- Network Traffic and Congestion Control
- Image Processing Techniques and Applications
- Advanced Optical Network Technologies
- Advanced Image Processing Techniques
- Optimal Experimental Design Methods
- Multimedia Communication and Technology
- Machine Learning and Algorithms
- Advanced Radiotherapy Techniques
- Handwritten Text Recognition Techniques
- Multimodal Machine Learning Applications
- Domain Adaptation and Few-Shot Learning
- Medical Imaging Techniques and Applications
Chinese University of Hong Kong
2021-2024
University of Hong Kong
2022
Pursuing accurate and robust recognizers has been a long-lasting goal for scene text recognition (STR) researchers. Recently, attention-based methods have demonstrated their effectiveness achieved impressive results on public benchmarks. The attention mechanism enables models to recognize with severe visual distortions by leveraging contextual information. However, recent studies revealed that the implicit over-reliance of context leads catastrophic out-of-vocabulary performance. On contrary...
With the rapid development of semiconductors, size transistors is continuously scaling down. The shrinking circuit poses great challenges to optical proximity correction (OPC) and hotspot detection (HSD). Recent advancements in OPC HSD commonly employ deep neural networks, achieving impressive performance within a limited runtime. Based on these achievements, we observe that learning-based models both require knowledge layout structure information. Furthermore, two tasks are closely related...
Physical design flow through associated electronic automation (EDA) tools plays an imperative role in the advanced integrated circuit design. Mostly, parameters fed into physical are mainly manually picked based on domain knowledge of experts. Nevertheless, owing to ever-shrinking scaling down technology nodes and complexity space spanned by combinations parameters, even coupled with time-consuming simulation process, such manual explorations for parameter configurations have become...
Power efficiency has become a nonneglected issue of modern CPUs. Therefore, accurate and robust power models are highly demanded in academia industry. However, it is hard for existing to balance modeling speed, generality, accuracy well. This article introduces McPAT-Calib, microarchitecture framework, which combines McPAT with machine learning (ML) calibration active (AL) sampling. McPAT-Calib can quickly accurately estimate the different benchmarks executed on CPU configurations, provide...
Energy efficiency has become the core issue of modern CPUs, and it is difficult for existing power models to balance speed, generality, accuracy. This paper introduces McPAT-Calib, a microarchitecture modeling framework, which combines McPAT with machine learning (ML) calibration methods. McPAT-Calib can quickly accurately estimate different benchmarks running on CPU configurations, provide an effective evaluation tool design CPUs. First, McPAT-7nm introduced support analytical 7nm...
With the rapid development of semiconductors and continuous scaling-down circuit feature size, hotspot detection has become much more challenging crucial as a critical step in physical verification flow. In recent years, advanced deep learning techniques have spawned many frameworks for detection. However, most existing detectors can only detect defects arising central region small clips, making whole process time-consuming on large layouts. Some multiple hotspots area but need to propose...
Inverse lithography technique (ILT) is one of the most widely used resolution enhancement techniques (RETs) to compensate for diffraction effect in process. However, ILT suffers from runtime overhead issues with shrinking size technology nodes. In this paper, our proposed L2O-ILT framework unrolls iterative optimization algorithm into a learnable neural network high interpretability, which can generate high-quality initial mask fast refinement. Experimental results demonstrate that method...
Design Rule Checking (DRC) is a critical step in integrated circuit design. DRC requires formatted scripts as the input to design rule checkers. However, these are manually generated foundry, which tedious and error prone for generation of thousands rules advanced technology nodes. To mitigate this issue, we propose first script framework, leveraging deep learning-based key information extractor automatically identify essential arguments from translator organize extracted into executable...
Design rule checking (DRC) is a critical step in integrated circuit design. DRC requires formatted scripts as the input to design checker. However, these are always generated manually foundry, and such generation process extremely inefficient, especially when encountering large number of rules. To mitigate this issue, we first propose deep learning-based key information extractor automatically identify essential arguments from Then, script translator designed organize extracted into...
Design rule checking (DRC) is a critical step in integrated circuit design. DRC requires formatted scripts as the input to design checker. However, these are always generated manually foundry, and such generation process extremely inefficient, especially when encountering large number of rules. To mitigate this issue, we first propose deep learning-based key information extractor automatically identify essential arguments from Then, script translator designed organize extracted into...