A5038232685- VLSI and Analog Circuit Testing
- Integrated Circuits and Semiconductor Failure Analysis
- Physical Unclonable Functions (PUFs) and Hardware Security
- Cell Image Analysis Techniques
- Advanced Graph Neural Networks
- Molecular Biology Techniques and Applications
- Human Pose and Action Recognition
- Software Testing and Debugging Techniques
- Generative Adversarial Networks and Image Synthesis
- Advanced Neural Network Applications
- Machine Learning in Materials Science
- Cryptographic Implementations and Security
- Quantum-Dot Cellular Automata
- Adversarial Robustness in Machine Learning
- Handwritten Text Recognition Techniques
- Anomaly Detection Techniques and Applications
- Fault Detection and Control Systems
Huazhong University of Science and Technology
2022-2025
Graph neural networks (GNNs) have shown considerable promise for graph-structured data. However, they are also known to be unstable and vulnerable perturbations attacks. Recently, the Lipschitz constant has been adopted as a control on stability of Euclidean networks, but calculating exact is difficult even very shallow networks. In this paper, we extend analysis graphs by providing systematic scheme estimating upper bounds constants GNNs. We derive concrete widely used GNN architectures...
The physical unclonable function (PUF) is a widely used hardware security primitive. Before hacking into PUF-protected system, intruders typically initiate attacks on the PUF as first step. Many strong designs have been proposed to thwart non-invasive that exploit acquired CRPs. In this work, we propose general framework for efficient PUFs by investigating from two perspectives, namely, statistical covariances in challenge space and design dependency among compositions. consists of novel...
Gradient boosting machine (GBM) is a powerful and widely used type of ensemble learning methods, among which the most famous one XGBoost (XGB). However, cost running large GBMs on hardware could become prohibitive given stringent resources. Ensemble reduction intrinsically hard because member models are constructed in sequential order, where training targets for latter ones depend performance former ones. In this work, GBM framework proposed first time to tackle problem. For time, supports...
The pursuit of accurate diagnosis with good resolution is driven by yield learning during both early bring-up and production excursions. Unfortunately, fault callouts from tools often render poor that hinders the follow-up failure analysis. In this work, we propose a method significantly improves diagnosis. By modeling logic circuits under test as graphs, employs graph neural networks to determine each candidate callout either true or false candidate. This novel deep mainly makes full use...
Failure diagnosis is a software-based, data-driven procedure. Collecting an excessive amount of fail data not only increases the overall test cost but can also potentially reduce diagnostic resolution. Thus, test-termination prediction proposed to dynamically determine appropriate failing pattern terminate testing, producing that sufficient for accurate analysis. In this work, we describe set novel methods utilizing advanced machine learning techniques efficient prediction. To implement...
Oftentimes fault candidates produced by logic diagnosis are too many to effectively guide the follow-on failure analysis. In this work, we propose a novel two-stage fast screening method sift through large amount of in callout outputted commercial tool. Candidates that unlikely be true ones re-assigned lower ranks or discarded as noise. Experimental results based on benchmark designs from various sources show within new categorical ranking list, number remaining is only 53.78% original tool,...