Shungeng Zhang

Scholarship

Selected Recent Publications

• DARD: Dice Adversarial Robustness Distillation against Adversarial Attacks, 2025
  Journal Article, Academic Journal
• Adversarial attacks on large language models, 2024
  Conference Proceeding
• Different attack and defense types for AI cybersecurity, 2024
  Conference Proceeding
• Sora: A Latency Sensi- tive Approach for Microservice Soft Resource Adaptation, 2023
  Conference Proceeding
• $\mu$ConAdapter: Reinforcement Learning-based Fast Concurrency Adaptation for Microservices in Cloud, 2023
  Conference Proceeding

Research Interests

My research interests lie in the field of distributed systems, cloud computing, and cybersecurity, which focuses on improving the performance and scalability of large-scale web applications and IoT stream processing systems to simultaneously achieve both good performance and high resource utilization efficiency in the cloud.
In my research, I actively adopt sophisticated timeline analysis methods with fine-grained monitoring data to find and study the transient bottlenecks which could lead to the long tail latency problem of the web-facing applications in the cloud. The transient bottlenecks (even with a very short lifespan such as 50ms) could lead to significant performance loss caused by the propagation and amplification effect, which is due to the complex dependency chains among components in web-facing applications. By correlating the transient bottlenecks with observed system performance metrics (e.g., throughput and load), the system can identify and remove the transient events which cause transient bottlenecks and bring better scalability and elasticity to web applications in the cloud.