A5101719623- Anomaly Detection Techniques and Applications
- Reinforcement Learning in Robotics
- Information and Cyber Security
- Network Security and Intrusion Detection
- Advanced Malware Detection Techniques
- Adversarial Robustness in Machine Learning
- Advanced Bandit Algorithms Research
- Data Quality and Management
- Security and Verification in Computing
- Winter Sports Injuries and Performance
- Evolutionary Algorithms and Applications
- Time Series Analysis and Forecasting
- Transportation and Mobility Innovations
- Advanced Neural Network Applications
- Face and Expression Recognition
- Artificial Intelligence in Games
- Business Process Modeling and Analysis
- Stock Market Forecasting Methods
- Advanced Data Storage Technologies
- Parallel Computing and Optimization Techniques
- Neural Networks and Applications
- AI in cancer detection
- Distributed and Parallel Computing Systems
- Explainable Artificial Intelligence (XAI)
- Martial Arts: Techniques, Psychology, and Education
Pennsylvania State University
2013-2022
Harbin Institute of Technology
2021
National University of Singapore
2021
Baoji University of Arts and Sciences
2018
Tsinghua University
2004
Moving Target Defense techniques have been proposed to increase uncertainty and apparent complexity for attackers. When more than one are effective limit opportunities of an attack, it is required compare these select the best defense choice. In this paper, we propose a three-layer model evaluate effectiveness different Defenses. This designed as attempt fill gap among existing evaluation methods works systematic framework comparison.
As an increasing number of deep-learning-based malware scanners have been proposed, the existing evasion techniques, including code obfuscation and polymorphic malware, are found to be less effective. In this work, we propose a reinforcement learning based semantics-preserving (i.e. functionality-preserving) attack against black-box GNNs (Graph Neural Networks) for detection. The key factor adversarial generation via semantic <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML"...
Deep reinforcement learning (DRL) has achieved remarkable success in various domains, yet its reliance on neural networks results a lack of transparency, which limits practical applications safety-critical and human-agent interaction domains. Decision trees, known for their notable explainability, have emerged as promising alternative to networks. However, decision trees often struggle long-horizon continuous control tasks with high-dimensional observation space due limited expressiveness....
Deep neural networks have proven to be quite effective in a wide variety of machine learning tasks, ranging from improved speech recognition systems advancing the development autonomous vehicles. However, despite their superior performance many applications, these models been recently shown susceptible particular type attack possible through generation synthetic examples referred as adversarial samples. These samples are constructed by manipulating real training data distribution order...
Although deep convolutional neural networks (DCNNs) have achieved significant accuracy in skin lesion classification comparable or even superior to those of dermatologists, practical implementation these models for cancer screening low resource settings is hindered by their limitations computational cost and training dataset. To overcome limitations, we propose a low-cost high-performance data augmentation strategy that includes two consecutive stages search network search. At the stage,...
Artificial intelligence (AI) is the simulation of human thought processes and consciousness on computer systems. Experts have gone so far as to chart future development AI systems that surpass a single civilization-shifting event, known "technological singularity" or "the Singularity." But in more near future, promises concrete applications throughout our everyday lives, including education, challenge China has already taken up. At present, main this so-called "weak" artificial Chinese...
Unmanned Combat Aerial Vehicle (UCAV) dogfight, which refers to a fight between two or more UCAVs usually at close quarters, plays decisive role on the aerial battlefields. With evolution of artificial intelligence, dogfight progressively transits towards intelligent and autonomous modes. However, development policy learning is hindered by challenges such as weak exploration capabilities, low efficiency, unrealistic simulated environments. To overcome these challenges, this paper proposes...
Deep reinforcement learning (DRL) has achieved remarkable success in various research domains. However, its reliance on neural networks results a lack of transparency, which limits practical applications. To achieve explainability, decision trees have emerged as popular and promising alternative to networks. Nonetheless, due their limited expressiveness, traditional struggle with high-dimensional long-horizon continuous control tasks. In this paper, we proposes SkillTree, novel framework...
Achieving complete and accurate cyber situation awareness (SA) is crucial for security analysts to make right decisions. To facilitate SA, existing tools, algorithms, techniques like attack graph, should be integrated together extract the most critical information synthesize knowledge from different areas. Based on theories of awareness, a SA model an SKRM (Situation Knowledge Reference Model) are constructed enhance coupling current enable analysts' effective analysis complex cyber-security...
The limousine service in luxury hotels is an integral component of the whole customer journey hospitality industry. One largest Singapore manages a fleet both in-house and outsourced vehicles around clock, serving 9000 trips per month on average. need for may scale up rapidly, especially during special events festive periods country. excess demand met by having additional standby, incurring millions dollars expenses year hotel. Determining required number limousines hour day challenging...
Developing an excellent quantitative trading strategy to obtain a high Sharpe ratio requires optimizing several parameters at the same time. Example include window length of moving average sequence, choice instruments, and thresholds used generate signals. Simultaneously all these seek is daunting time-consuming task, partly because unknown mechanism determining ratio. This article proposes using Bayesian optimization systematically search for optimal parameter configuration that leads The...
This manuscript documents our approach to addressing the data challenge posted by ICPHM23 conference [1]. The task is a time series classification problem. We see two general toolsets can be used complete and produce promising high accuracy for such large set. One deep neural networks, other gradient boosting. choose During feature preparation, we developed customized C++ parallel computing software extract all desired features. includes thought process final cross validation results.