A5033259177- Radiation Effects in Electronics
- Advanced Neural Network Applications
- Parallel Computing and Optimization Techniques
- Adversarial Robustness in Machine Learning
- Privacy-Preserving Technologies in Data
- Advanced Graph Neural Networks
- Distributed systems and fault tolerance
- Domain Adaptation and Few-Shot Learning
- Graph Theory and Algorithms
- Energy Efficient Wireless Sensor Networks
- Low-power high-performance VLSI design
- Big Data and Digital Economy
- Advanced Data Storage Technologies
- Brain Tumor Detection and Classification
- Security and Verification in Computing
- Recommender Systems and Techniques
- Semiconductor materials and devices
- Data Management and Algorithms
- Age of Information Optimization
- Indoor and Outdoor Localization Technologies
- Physical Unclonable Functions (PUFs) and Hardware Security
- Smart Grid Security and Resilience
- COVID-19 diagnosis using AI
- Software Testing and Debugging Techniques
- Energy Load and Power Forecasting
Jilin University
2018-2025
Jilin Medical University
2024-2025
Jilin Province Science and Technology Department
2018-2024
Real-time underwater monitoring has been widely applied in many applications of wireless sensor networks (UWSNs). Due to the long acoustic communication delays, real-time data collection UWSNs is challenging. Moreover, transmission faces problem high loss rate, which causes a longer delay time due need for packet retransmissions. To address these problems, we propose recurrent neural network (RNN)-based framework with consideration delay, energy, and quality. We drop automatic retransmission...
Sparse mobile crowdsensing (Sparse MCS), a new paradigm for large-scale fine-grained urban monitoring applications, collects sensing data from relatively few areas and infers uncovered areas. In MCS, the task allocation problem is simplified to area selection since it typically assumed that there were enough participants across target area. However, in many real scenarios, no guarantee platform can find execute tasks vital this case, additional moving costs are incurred, which not beneficial...
Nowadays, selective instruction duplication (SelDup) is the typical approach to detect silent data corruption (SDC) in GPGPU. However, owing up-to-billions fault sites of parallel GPGPU kernel functions, it usually introduces tremendous overhead perform injections (FIs) for obtaining duplication-candidate set (although can be conducted parallel). Moreover, current SelDup typically considers all SDCs severe and tends duplicate more instructions. The nontrivial seriously restricts deployment...
Today, various domain-specific convolutional neural network (CNN) accelerators are deployed in large-scale systems to satisfy the massive computational demands of current deep CNNs. Although bringing significant performance improvements, highly integrated CNN more susceptible faults caused by radiation, aging, and process variation. CNNs have been increasingly security-critical areas, requiring attention reliable execution. classical fault-tolerant approaches error-effective,...
To satisfy prohibitively massive computational requirements of current deep Convolutional Neural Networks (CNNs), CNN-specific accelerators are widely deployed in large-scale systems. Caused by high-energy neutrons and α-particle strikes, soft error may lead to catastrophic failures when CNN is on high integration density accelerators. As CNNs become ubiquitous mission-critical domains, ensuring the reliable execution presence errors increasingly essential. In this article, we propose Re...
As GPUs become ubiquitous in large-scale general purpose HPC systems (GPGPUs), ensuring the reliable execution of such presence soft errors is increasingly essential. To provide insights into how resilient GPU programs are toward errors, researchers typically rely on random Fault Injection (FI) to evaluate tolerance programs. However, it expensive obtain a statistically significant resilience profile and not suitable identify all error-critical fault sites
Graphics Processing Units (GPUs) are widely used in general-purpose high-performance computing applications (i.e., GPGPUs), which require reliable execution the presence of soft errors. To support massive thread level parallelism, a sizeable register file is adopted GPUs, highly vulnerable to Although modern commercial GPUs provide single-error-correction double-error-detection (SEC-DED) ECC for file, it consumes considerable amount energy due frequent accesses and leakage power storage.In...
Graphics processing units (GPUs) are widely used in general-purpose high-performance computing applications (i.e., GPGPUs), which require reliable execution the presence of soft errors. To support massive thread-level parallelism, a sizeable register file is adopted GPUs, highly vulnerable to Although modern commercial GPUs provide single-error-correction double-error-detection (SEC-DED) error correction code (ECC) for file, it consumes considerable amount energy due frequent accesses and...
Graphics Processing Units (GPUs) are widely used in a range of High Performance Computing fields because high parallelism. As the technology scaling down, GPUs more susceptible to soft errors which dramatically impact applications output qualities. Silent Data Corruption (SDC) is one most concerned reliability issues, require efficient protection mechanisms eliminate it. Software-directed instruction replication has been flexible technique solve SDCs. However, this method requires trade-off...
General-Purpose Graphics Processing Units (GPGPUs) are widely utilized for graph processing thanks to their high throughput, massive parallelism and powerful computing capacity. However, due the increasing integration, GPGPUs susceptible soft errors, which can undermine reliability of applications accelerated using GPGPUs. Typically, fault tolerance strategies such as thread replication checkpoint mechanism applied ensure program execution. these techniques require appropriate trade-off...