Steven Lamp - McNichols Family Foundation Scholar
1st Year Scholar, PhD Candidate, Computer Science
University of Virginia
To evaluate novel machine-learning (ML) techniques to detect autonomous bots online, one needs to accurately mimic the digital artifacts created by human behavior. This research creates human-like responses by analyzing a real human's digital behavior, takes those behavioral traits, and augments an AI control unit to utilize a computer in a realistic way.
Describe the expected benefit of your research to society:
Global infrastructures, from dams to bank vaults, are linked to the internet and are bombarded by autonomous bots attempting to compromise critical areas vital to normal operations. Creating defensive measures against these bots is imperative to ensuring an infrastructure's security. To determine which techniques are effective, we must evaluate them. Evaluating a defense when human behavior is involved is an open problem due to the fact that human behavior is extremely difficult to simulate realistically. Therefore, this research will deliver a stronger method for evaluating defensive techniques by providing realistic human behaviors to an isolated testbed environment, ensuring our global infrastructure remains resilient to cyber-attacks. DOLOS (named for the Greek spirit of trickery) focuses on providing realistic human-like computer operation mimics for an accurate testbed of a real-world cyber defensive situation. DOLOS mimics behavior similar to a real human sitting at a computer desk. Hence, a single DOLOS control unit is called a mimic. The mimic’s AI-enabled behaviors include realistic mouse movements, web browsing, email and document editing, and operating system interactions such as moving or copying files. The mimics can be tailored to fit behaviors indicative of different online environments, such as a corporate office or an army base. DOLOS is the first of its kind to provide realistic human-like behavior of a computer's operation in an isolated testbed environment, enabling faster development and evaluation of novel cyber defensive techniques and ensuring our global infrastructure is resilient to cyber-attacks.
How will an ARCS award might benefit your research:
The ARCS Fellowship will help support my research in several ways. It will provide me with the ability to travel and visit academic scholars, industry and government research laboratories, and attend relevant scientific conferences to disseminate the results of my research. The fellowship will also help provide support to mentor and involve undergraduate students interested in participating in the DOLOS cybersecurity research project.