Selected Five Papers (Full List: Google Scholar)

This paper investigates vulnerabilities in localized crowd counting, an important computer vision application, by introducing a novel adversarial attack framework.

Keywords: Crowded counting, density map, point regression, adversarial attack, transferability

This paper introduces a multi-agent architecture with a routing model that dispatches commands to external tools or LLM agents to support interactive commands simulation.

Keywords: LLM multi-agent, terminal simulation, shell honeypot, LLM fine-tuning

WavePurifier is an audio purification framework to defend against audio adversarial attacks.

Keywords: AI security, speech recognition, diffusion model, mobile security

Glint is the first graph learning-based system for interactive threat detection in smart homes.

Keywords: graph neural network, contrastive learning, smart home, security

Facer is the first active acoustic facial expression recognition system on smartphone devices.

Keywords: acoustic sensing, expression recognition, contrastive learning, domain adaptation