Mohammad Hossein Rohban

Associate Professor@Sharif University of Technology·伊朗·OpenReview

研究方向

Out-of-distribution generalization · Out-of-distribution detection · Open-Set Recognition · Reinforcement Learning · Medical Imaging · Adversarial Robustness · Computer Vision · Anomaly Detection · Computational Biology · Topic Modeling · Semi-Supervised Learning

6.0

‌Navigating the MIL Trade-Off: Flexible Pooling for Whole Slide Image Classification

NeurIPS 2025Poster

通讯

6.0

Adversarially Robust Anomaly Detection through Spurious Negative Pair Mitigation

ICLR 2025Poster

通讯

6.0

Trained Models Tell Us How to Make Them Robust to Spurious Correlation without Group Annotation

ICLR 2025Rejected

5.6

A Contrastive Teacher-Student Framework for Novelty Detection under Style Shifts

ICLR 2025Rejected

通讯

-1

Bias in CLIP Encoders: A Study of Encoder Bias and Object Representation in Multi-Object Scenarios

ICLR 2025Desk Rejected

合作者 (20)

Mahdieh Soleymani Baghshah

Mohammad Hossein Rohban

Implicit Regularization of SGD Reduces Shortcut Learning

CELAD: Compositional Evaluation for Logical Anomaly Detection

SORA: Free Second Order Attacks in Fast Adversarial Training

DiffPM: Diffusion-Based Generative Framework for Time Series Synthesis

Spurious-Aware Prototype Refinement for Reliable Out-of-Distribution Detection

FrameShield: Adversarially Robust Video Anomaly Detection

‌Navigating the MIL Trade-Off: Flexible Pooling for Whole Slide Image Classification

Adversarially Robust Anomaly Detection through Spurious Negative Pair Mitigation

Trained Models Tell Us How to Make Them Robust to Spurious Correlation without Group Annotation

A Contrastive Teacher-Student Framework for Novelty Detection under Style Shifts

Bias in CLIP Encoders: A Study of Encoder Bias and Object Representation in Multi-Object Scenarios