Our paper “Agentic AI-Empowered Dynamic Survey Framework” has been accepted to NeurIPS 2026! Read preprint now.
My work mostly focuses on
- anomaly detection
- adversarial machine learning
- robustness
- lately: agentic AI systems
I am particularly interested in how learning systems behave under instability, especially in settings where small changes propagate in unexpected ways. This site contains some of my work, along with a few things adjacent to it.
Lately
All updates ↗Our paper “Language Models in Video Anomaly Detection: A Dynamic Survey” is published in IEEE Access.
Our paper “Leveraging Multimodal LLM Descriptions of Activity for Explainable Semi-Supervised Video Anomaly Detection” has been published in TMLR. Read it now.
Our paper “Universal and Efficient Detection of Adversarial Data through Nonuniform Impact on Network Layers” has been published in TMLR. Read it now.
Notes
All notes ↗Technical
Layer Regression: State-of-the-Art Adversarial Attack Detection Made Simple
A fast, clean-data-trained method for detecting adversarial examples through inconsistencies between intermediate and deeper neural network representations.
Technical
Is video anomaly detection misframed?
Most benchmarks reward recognizing known abnormal actions, rather than understanding what makes an event anomalous within a specific scene.
Technical
Robustness changes when systems become agents
Why robustness in agentic systems cannot be treated as a purely global constraint, and how stability depends on where failures emerge.
Music
Even in Arcadia
On certain songs that stay, and why some patterns do not disappear even when everything else changes.