Puneet K. Dokania

University of Oxford — United Kingdom

Computer Science · Physical Sciences

20h-index3.2kcitations93works0.72yr avg
OpenAlex↗Google Scholar↗Lab page↗

Accepting Students?

No reports yet. Know if this professor is taking students?

Research Topics

Advanced Neural Network Applications(25), Domain Adaptation and Few-Shot Learning(24), Adversarial Robustness in Machine Learning(17), Multimodal Machine Learning Applications(15), Anomaly Detection Techniques and Applications(13)

Publications93 total

Foundation Models for Trajectory Planning in Autonomous Driving: A Review of Progress and Open Challenges
arXiv (Cornell University)·2025Open Access
MObi: Multimodal Object Inpainting Using Diffusion Models
2025
Mixture of Experts Made Intrinsically Interpretable
arXiv (Cornell University)·2025Open Access
AnnoCaseLaw: A Richly-Annotated Dataset For Benchmarking Explainable Legal Judgment Prediction
arXiv (Cornell University)·2025Open Access
MObI: Multimodal Object Inpainting Using Diffusion Models
arXiv (Cornell University)·2025Open Access
Segment, Select, Correct: A Framework for Weakly-Supervised Referring Segmentation
Lecture notes in computer science·2025
Placing Objects in Context via Inpainting for Out-of-Distribution Segmentation
Lecture notes in computer science·2024· 5 cited
On Calibration of Object Detectors: Pitfalls, Evaluation and Baselines
Lecture notes in computer science·2024· 2 cited
What Makes and Breaks Safety Fine-tuning? A Mechanistic Study
arXiv (Cornell University)·2024· 3 citedOpen Access
On Calibration of Object Detectors: Pitfalls, Evaluation and Baselines
arXiv (Cornell University)·2024Open Access
Placing Objects in Context via Inpainting for Out-of-distribution Segmentation
arXiv (Cornell University)·2024Open Access
Random Representations Outperform Online Continually Learned Representations
arXiv (Cornell University)·2024Open Access
RanDumb: Random Representations Outperform Online Continually Learned Representations
2024
What Makes and Breaks Safety Fine-tuning? A Mechanistic Study
2024
Segment, Select, Correct: A Framework for Weakly-Supervised Referring Segmentation
arXiv (Cornell University)·2023Open Access
Raising the Bar on the Evaluation of Out-of-Distribution Detection
2023· 3 cited
MoCaE: Mixture of Calibrated Experts Significantly Improves Object Detection
arXiv (Cornell University)·2023· 3 citedOpen Access
Fine-tuning can cripple your foundation model; preserving features may be the solution
arXiv (Cornell University)·2023· 1 citedOpen Access
Towards Building Self-Aware Object Detectors via Reliable Uncertainty Quantification and Calibration
arXiv (Cornell University)·2023Open Access
Sample-Dependent Adaptive Temperature Scaling for Improved Calibration
Proceedings of the AAAI Conference on Artificial Intelligence·2023· 21 citedOpen Access
page 1 of 5Next →

Frequent Co-authors

Philip H. S. Torr(34), Arslan Chaudhry(8), Thalaiyasingam Ajanthan(6), Daniela Massiceti(6), Ameya Prabhu(5), Amartya Sanyal(5), Ser-Nam Lim(5), Philip H. S. Torr(5), Jishnu Mukhoti(4), Pau de Jorge(4), Grégory Rogez(4), Francesco Pinto(4), Viveka Kulharia(3), Amartya Sanyal(3), Arnab Ghosh(3), N. Siddharth(3), Yunchao Wei(3), Ming‐Ming Cheng(3), Tom Joy(3), Richard Hartley(3)