A5026765811- Distributed systems and fault tolerance
- Advanced Data Storage Technologies
- Radiation Effects in Electronics
- Parallel Computing and Optimization Techniques
- Cloud Computing and Resource Management
- Distributed and Parallel Computing Systems
- Scientific Computing and Data Management
- Security and Verification in Computing
- Algorithms and Data Compression
- Software System Performance and Reliability
- Tensor decomposition and applications
- Remote Sensing and LiDAR Applications
- Advanced Vision and Imaging
- 3D Surveying and Cultural Heritage
- Robotics and Sensor-Based Localization
- Graph Theory and Algorithms
- Advancements in Photolithography Techniques
- Advanced Neural Network Applications
- Data Quality and Management
- Electron and X-Ray Spectroscopy Techniques
- Anomaly Detection Techniques and Applications
- Data Visualization and Analytics
- Advanced Graph Neural Networks
- Machine Learning and ELM
- Advanced Electron Microscopy Techniques and Applications
Pacific Northwest National Laboratory
2018-2025
University of California, Merced
2018-2023
Lawrence Livermore National Laboratory
2019
Nanchang Hangkong University
2013
Neural Architecture Search (NAS) is a powerful approach of automating the design efficient neural architectures. In contrast to traditional NAS methods, recently proposed one-shot methods prove be more in performing NAS. One-shot works by generating singular weight-sharing supernetwork that acts as search space (container) subnetworks. Despite its achievements, designing remains major challenge. this work we propose strategy for Vision Transformer (ViT)-based particular, convert Segment...
Sparse tensor algebra is widely used in many applications, including scientific computing, machine learning, and data analytics. The performance of sparse kernels strongly depends on the intrinsic characteristics input tensors, hence storage formats are designed for tensors to achieve optimal particular applications/architectures, which makes it challenging implement optimize every operation interest a given architecture. We propose domain-specific language (DSL) compiler framework...
As high-performance computing systems scale in size and computational power, the danger of silent errors, i.e., errors that can bypass hardware detection mechanisms impact application state, grows dramatically. Consequently, applications running on HPC need to exhibit resilience such errors. Previous work has found that, for certain codes, this come free, some are naturally resilient, but few studies have shown code patterns-combinations or sequences computations-that make an resilient. In...
Understanding application resilience (or error tolerance) in the presence of hardware transient faults on data objects is critical to ensure computing integrity and enable efficient application-level fault tolerance mechanisms. However, we lack a method tool quantify objects. The traditional method, random injection, cannot help, because losing semantics insufficient information how where errors are tolerated. In this paper, introduce (called "MOARD") model Our based systematically...
Automating the theory-experiment cycle requires effective distributed workflows that utilize a computing continuum spanning lab instruments, edge sensors, resources at multiple facilities, data sets across information sources, and potentially cloud. Unfortunately, obvious methods for constructing platforms, orchestrating workflow tasks, curating datasets over time fail to achieve scientific requirements performance, energy, security, reliability. Furthermore, achieving best use of depends...
Tensor algebra is widely used in many applications, such as scientific computing, machine learning, and data analytics. The tensors represented real-world are usually large sparse. There tens of storage formats designed for sparse matrices and/or the performance tensor operations depends on a particular architecture selected format, which makes it challenging to implement optimize every operation interest transfer code from one another. We propose domain-specific language (DSL) compiler...
A critical performance challenge in distributed scientific workflows is coordinating tasks and data flows on resources. To guide these decisions, this paper introduces flow lifecycle analysis. Workflows are commonly represented using directed acyclic graphs (DAGs). Data lifecycles (DFL) enrich task DAGs with objects properties that describe how interact flow. Lifecycles enable analysis from several important perspectives: task, data, We representation, measurement, analysis, visualization,...
As high-performance computing systems scale in size and computational power, the danger of silent errors, i.e., errors that can bypass hardware detection mechanisms impact application state, grows dramatically. Consequently, applications running on HPC need to exhibit resilience such errors. Previous work has found that, for certain codes, this come free, some are naturally resilient, but few studies have shown code patterns---combinations or sequences computations---that make an resilient....
Frame structure estimation from line segments is an important yet challenging problem in understanding indoor scenes. In practice, segment extraction can be affected by occlusions, illumination variations, and weak object boundaries. To address this problem, approach for frame recovery based on refinement voting proposed. We refined the revising, connecting, adding operations. then propose iterative mechanism selecting segments, where a cross ratio constraint enforced to build crab-like...
Tensor Algebra is a powerful tool for carrying out high-order data analytics in scientific applications, such as finite element analysis, N-body simulation, and quantum chemistry. Many of these applications are critical terms correctness safety. Since often run on High Performance Computing (HPC) systems, which susceptible to soft errors caused by cosmic rays, unstable voltage, etc., we must ensure that the execution reliable resilient, outcome highly trustworthy. However, traditional fault...
MPI has been ubiquitously deployed in flagship HPC systems aiming to accelerate distributed scientific applications running on tens of hundreds processes and compute nodes. Maintaining the correctness integrity application execution is critical, especially for safety-critical applications. Therefore, a collection effective fault tolerance techniques have proposed enable efficiently resume from system failures. However, there no structured way study compare different designs, so guide...
High-performance computing (HPC) systems play a critical role in facilitating scientific discoveries. Their scale and complexity (e.g., the number of computational units software stack) continue to grow as new are expected process increasingly more data reduce time. However, with processing elements, probability that these will experience random bit-flip error corrupts program's output also increases, which is often recognized silent corruption. Analyzing resiliency HPC applications...
Image segmentation is a critical enabler for tasks ranging from medical diagnostics to autonomous driving. However, the correct semantics - where are boundaries located? what segments logically similar? change depending on domain, such that state-of-the-art foundation models can generate meaningless and incorrect results. Moreover, in certain domains, fine-tuning retraining techniques infeasible: obtaining labels costly time-consuming; domain images (micrographs) be exponentially diverse;...
One of the most common operations in multimodal scientific data management is searching for $k$ similar items (or, $k$-nearest neighbors, KNN) from database after being provided a new item. Although recent advances machine learning models offer \textit{semantic} index, so-called \textit{embedding vectors} mapped original data, dimension resulting embedding vectors are usually on order hundreds or thousand, which impractically high time-sensitive applications. This work proposes to reduce...
Checkpoint/Restart (C/R) has been widely deployed in numerous HPC systems, Clouds, and industrial data centers, which are typically operated by system engineers. Nevertheless, there is no existing approach that helps engineers without domain expertise, scientists fault tolerance knowledge identify those critical variables accounted for correct application execution restoration a failure C/R. To address this problem, we propose an analytical model tool (AutoCheck) can automatically to...