A5049646773- AI in cancer detection
- Security and Verification in Computing
- Parallel Computing and Optimization Techniques
- Medical Image Segmentation Techniques
- Cloud Data Security Solutions
- Adversarial Robustness in Machine Learning
- Advanced Memory and Neural Computing
- Image Retrieval and Classification Techniques
- Advanced Data Storage Technologies
- Infrared Thermography in Medicine
- Hydrogen Storage and Materials
- Quantum Computing Algorithms and Architecture
- Quantum Information and Cryptography
- Radiomics and Machine Learning in Medical Imaging
- Quantum and electron transport phenomena
- Advanced Malware Detection Techniques
- Ammonia Synthesis and Nitrogen Reduction
- MXene and MAX Phase Materials
- Ferroelectric and Negative Capacitance Devices
- Advanced Image Fusion Techniques
- COVID-19 diagnosis using AI
- Brain Tumor Detection and Classification
- Magnesium Alloys: Properties and Applications
- High-Energy Particle Collisions Research
- Diamond and Carbon-based Materials Research
University of Pittsburgh
2021-2024
Yanshan University
2023-2024
State Key Laboratory of Metastable Materials Science and Technology
2024
Zhengzhou Business University
2024
Zhengzhou University
2024
Shandong University of Science and Technology
2020
Lanzhou University
2015-2019
University of Augsburg
2019
One of the frontier issues that severely hamper development automatic snore sound classification (ASSC) associates to lack sufficient supervised training data. To cope with this problem, we propose a novel data augmentation approach based on semi-supervised conditional generative adversarial networks (scGANs), which aims automatically learn mapping strategy from random noise space original distribution. The proposed has capability well synthesizing "realistic" high-dimensional data, while...
Rowhammer Attack, a new DRAM-based attack, was developed exploiting weak cells to alter their content. Such attacks can be launched at the user level without requiring access permission victim memory cells. Leveraging such attacks, bit-flip-based adversarial weights attack (BFA) targeting deep neural network models. When BFA attackers acquire DNN model, they manipulate existing into locating vulnerable bits in target model. By flipping subset of them using Rowhammer, crash that model within...
Modern x86 processors have many prefetch instructions that can be used by programmers to boost performance. However, these may also cause security problems. In particular, we found on Intel processors, there are two flaws in the implementation of PREFETCHW, an instruction for accelerating future writes. First, this execute data with read-only permission. Second, execution time leaks current coherence state target data. Based design issues, build cross-core private cache attacks work both...
Abstract Nonlinear science is a fundamental area of physics research that investigates complex dynamical systems, which are often characterized by high sensitivity and nonlinear behaviors. Numerical simulation plays pivotal role in science, serving as critical tool for revealing the underlying principles governing these systems. In addition, it crucial accelerating progress across various fields, such climate modeling, weather forecasting, fluid dynamics. However, their computational cost...
The Super $τ$-Charm facility (STCF) is an electron-positron collider proposed by the Chinese particle physics community. It designed to operate in a center-of-mass energy range from 2 7 GeV with peak luminosity of $0.5\times 10^{35}{\rm cm}^{-2}{\rm s}^{-1}$ or higher. STCF will produce data sample about factor 100 larger than that present factory -- BEPCII, providing unique platform for exploring asymmetry matter-antimatter (charge-parity violation), in-depth studies internal structure...
Modern $\times$86 processors feature many prefetch instructions that developers can use to enhance performance. However, with some instructions, users more directly manipulate cache states which may result in powerful covert channel and side attacks. In this work, we reverse-engineer the detailed behavior of PREFETCHNTA on various Intel processors. Based results, first propose a new conflict-based named NTP+NTP. Prior channels often require priming set order cause conflicts. contrast,...
La–Mg–Ni-based alloys with a novel AB 4 -type superlattice structure is supposed as potential anode materials for nickel metal hydride (Ni/MH) batteries due to the excellent discharge ability at high rates and long cycling life. However, it still challenging achieve phase content of during annealing complex peritectic reaction virous structures. Herein, we study transformation upon elucidate its effect on electrochemical characteristics based La 0.60 Sm 0.20 Mg Ni 3.50 Al alloy. It found...
A flexible reinforcement learning (RL) optimal collision-avoidance control formulation for unmanned aerial vehicles (UAVs) with discrete-time frameworks is revealed in this work. By utilizing the neural network (NN) estimating capacity and actor-critic scheme of RL technique, an adaptive collision-free controller a minimal parameter (MLP) formulated, which based on novel strategic utility function. The issue, couldn’t be addressed prior literature, can resolved by suggested approaches....
In recent years, quantum computing has undergone significant developments and established its supremacy in many application domains. While hardware is accessible to the public through cloud environment, a robust efficient circuit simulator necessary investigate constraints foster development, such as algorithm development device architecture exploration. this paper, we observe that most of publicly available simulators (e.g., QISKit from IBM, QDK Microsoft, Qsim-Cirq Google) suffer slow...
Multi-GPU systems have emerged as a desirable platform to deliver high computing capabilities and large memory capacity accommodate dataset sizes. However, naively employing multi-GPU incurs non-scalable performance. One major reason is that execution efficiency suffers expensive address translations in systems. The data-sharing nature of GPU applications requires page migration between GPUs mitigate non-uniform access overheads. Unfortunately, frequent substantial table invalidation...
The sharing of last-level cache (LLC) among different CPU cores makes vulnerable to side channel attacks. An attacker can get private information about co-running applications (victims) by monitoring their accesses in LLC. Cache attacks be mitigated partitioning between the victim and attacker. However, previous partition works either make weak assumptions attacker's strength or force security mechanisms thus overhead every user on system, regardless requirement.
DNN-based Adaptive Cruise Control (ACC) systems are very convenient but also safety critical. Although prior work has explored physical adversarial attacks on DNN models, those mostly static and their effects a real-world ACC system not clear. In this work, we propose the first end-to-end attack systems, test indication state-of-the-art products. The experimental results show that our approach can make vehicle driving with accelerate unsafely cause rear-end collision.
Strided memory accesses are an important type of operations for In-Memory Databases (IMDB) applications. often demand data at word granularity with fixed strides. Hence, they tend to produce sub-optimal performance on DRAM (the de facto standard in modern computer systems) that cacheline granularity. Recently proposed optimizations either introduce significant reliability degradation or limited non-volatile crossbar structures.
Bit-flip attack (BFA) has become one of the most serious threats to Deep Neural Network (DNN) security. By utilizing Rowhammer flip bits DNN weights stored in memory, attacker can turn a functional into random output generator. In this work, we propose ModelShield, defense mechanism against BFA, based on protecting integrity using hash verification. ModelShield performs real-time verification weights. Since slow down inference by up 7×, further two optimizations for ModelShield. We implement...
Image segmentation is an important task in the analysis of mammograms and it challenging because masses are low contrast with ambiguous margins. The classical Vector Field Convolution (VFC) method has demonstrated its merits image segmentation; however difficult to capture object boundaries. In this work, a new snake model proposed solve problem, combines morphological filter laplacian operator. A external force calculated by convolving edge map user-defined vector field kernel. algorithm...
Goal: Mammography is the most effective technique for breast cancer screening, and detection of abnormalities plays a vital role in Computer Aided Detection (CAD) system. This paper proposes method abnormal mammograms based on improved saliency detection. Methods: The proposed comprises three main steps: firstly, enhance using dual morphological top-hat operations with non-flat structuring element, secondly, utilize simple efficient detector to obtain value then employ modified Simplified...
Memory safety continues to be a critical challenge in modern computing, with approximately 70% of reported vulnerabilities annually attributed memory-related issues. Among these issues, Use-After-Free (UAF) or bugs, where program accesses memory through dangling pointer, pose significant threats. Existing UAF detection methods, such as Key-And-Lock (KAL) mechanisms, incur notable performance overhead due explicitly propagating the key and lock address (metadata). We identify that 67% KAL's...