Jonas Guan

Hi. It’s Jonas.

I work on AI systems, especially the places where they fail or become unsafe.

I’m a PhD candidate in Computer Science at the University of Toronto, advised by Nicolas Papernot and William Cunningham. I expect to graduate in December 2026.

I’m currently a Student Researcher at Google DeepMind, developing epistemic reasoning and memory systems for LLM agents.

Email: <firstname>@cs.toronto.edu

Publications

The Erasure Illusion: Stress-Testing the Generalization of LLM Forgetting Evaluation

Hengrui Jia, Taoran Li, Jonas Guan, Varun Chandrasekaran.

Under review.
Temporal-Difference Learning Using Distributed Error Signals

Jonas Guan, Shon Verch, Claas Voelcker, Ethan Jackson, Nicolas Papernot, William Cunningham.

NeurIPS, 2024.
Dataset Inference for Self-Supervised Models

Adam Dziedzic, Haonan Duan, Muhammad Ahmad Kaleem, Nikita Dhawan, Jonas Guan, Yannis Cattan, Franziska Boenisch, Nicolas Papernot.

NeurIPS, 2022.
On the Difficulty of Defending Self-Supervised Learning Against Model Extraction

Adam Dziedzic, Nikita Dhawan, Muhammad Ahmad Kaleem, Jonas Guan, Nicolas Papernot.

ICML, 2022.
A Zest of LIME: Towards Architecture-Independent Model Distances

Hengrui Jia, Hongyu Chen, Jonas Guan, Ali Shahin Shamsabadi, Nicolas Papernot.

ICLR, 2022.
StateFormer: Fine-Grained Type Recovery From Binaries Using Generative State Modeling

Kexin Pei, Jonas Guan, Matthew Broughton, Zhongtian Chen, Songchen Yao, David Williams-King, Vikas Ummadisetty, Junfeng Yang, Baishakhi Ray, Suman Jana.

FSE, 2021.
XDA: Accurate, Robust Disassembly With Transfer Learning

Kexin Pei, Jonas Guan, David Williams-King, Junfeng Yang, Suman Jana.

NDSS, 2021.

Publications

The Erasure Illusion: Stress-Testing the Generalization of LLM Forgetting Evaluation

Temporal-Difference Learning Using Distributed Error Signals

Dataset Inference for Self-Supervised Models

On the Difficulty of Defending Self-Supervised Learning Against Model Extraction

A Zest of LIME: Towards Architecture-Independent Model Distances

StateFormer: Fine-Grained Type Recovery From Binaries Using Generative State Modeling

XDA: Accurate, Robust Disassembly With Transfer Learning