• Softmax is 1/2-Lipschitz: A tight bound across all \ell_p norms
  Pravin Nair. [arXiv]
• Plug-and-Play Linear Attention for Pre-trained Image and Video Restoration Models
  Srinivasan Kidambi, and Pravin Nair. [arXiv]
• Detecting Localized Deepfake Manipulations Using Action Unit-Guided Video Representations.
  Tharun Anand, Siva Sankar, and Pravin Nair.
  Proceedings of the Computer Vision and Pattern Recognition Conference Workshops. 2025. [arXiv]