Hesam (Sam) Asadollahzadeh
PhD student, The University of Melbourne · ARC Centre of Excellence for Automated Decision-Making and Society (ADM+S)
I am a PhD student at The University of Melbourne, working on distributional robustness in vision-language models, supervised by Prof. Christopher Leckie and Dr. Sarah Erfani.
My thesis studies the theory and scalable methods for out-of-distribution generalisation in contrastive learning and foundation vision-language models. More broadly, I work on trustworthy machine learning: distributional and adversarial robustness, deep generative models, self-supervised learning and computational biology.
- OOD generalisation
- Vision-language models
- Adversarial robustness
- Spurious correlations
- Generative models
- Computational biology
Publications Full list on Google Scholar
-
NeurIPS 2026
-
ICML 2026
-
Nature Communications 2026 · ICLR 2025
-
ICLR 2026
-
ICLR 2026
-
ICLR 2026
-
bioRxiv 2026
-
arXiv 2026
-
bioRxiv 2026
-
SCSL @ ICLR 2025
-
OPT @ NeurIPS 2025
-
bioRxiv 2025
-
OpenReview 2025
-
arXiv 2023
-
2023
Disentangling Covariates to Predict Counterfactuals for Single-cell Data
Citation counts come from Google Scholar, as of the date shown at the top. Versions of the same paper (preprint and camera-ready) are counted together.
News
- ORCA, hunting compositional failures in text-to-image diffusion, accepted to NeurIPS 2026. arXiv
- Celcomen published in Nature Communications.
- PerturbGen, predicting how perturbations reshape cellular trajectories, is on bioRxiv.
- GHOST, hallucination-inducing image generation for multimodal LLMs, is on arXiv.
Show earlier newsShow less
- Multi-covariate disentanglement with counterfactual analysis on synthetic data is on bioRxiv.
- "Exploiting What Trained Models Learn…" published at the ICLR 2025 Workshop on Spurious Correlation and Shortcut Learning.
- Started my PhD at The University of Melbourne in the ADM+S Centre of Excellence.
- Our proposal for the Workshop on Spurious Correlation and Shortcut Learning: Foundations and Solutions was accepted at ICLR 2025.
- "Trained Models Tell Us How to Make Them Robust to Spurious Correlation without Group Annotation" is on arXiv.
- Two papers accepted to the ICCV 2023 Workshop on OOD Generalization in Computer Vision:
- Joined the Machine Learning Lab at Sharif University of Technology, working with Dr. Mahdieh Soleymani Baghshah and Dr. Mohammad Hossein Rohban, with the Robust and Interpretable Machine Learning Lab (RIML).
Teaching All teaching
Computer Vision
Eight interactive workshop pages, from arrays and filtering to ViTs, diffusion models and segmentation.
Workshop pages →Trustworthy Machine Learning
Seven interactive workshop pages on privacy, anomaly and OOD detection, adversarial robustness and backdoors.
Workshop pages →- Deep Learning · Sharif University of Technology2023
- Intro to Machine Learning; Artificial Intelligence · Sharif University of Technology2023
- Neural Networks; Artificial Intelligence; Algorithms; Probability & Statistics · University of Tehran2022–23