6 papers
Score Shocks: The Burgers Equation Structure of Diffusion Generative Models
Krisanu Sarkar
We analyze the score field of a diffusion generative model through a Burgers-type evolution law. For VE diffusion, the heat-evolved data density implies that the score obeys viscou…
Synthetic Designed Experiments for Diagnosing Vision Model Failure
Krisanu Sarkar
Current synthetic data pipelines for computer vision generate images without diagnosing what the downstream model actually needs. This open-loop paradigm treats synthetic data as c…
On the Spectral Geometry of Cross-Modal Representations: A Functional Map Diagnostic for Multimodal Alignment
Krisanu Sarkar
We study cross-modal alignment between independently pretrained vision (DINOv2) and language (all-MiniLM-L6-v2) encoders using the functional map framework from computational geome…
Adaptive Variance-Penalized Continual Learning with Fisher Regularization
Krisanu Sarkar
The persistent challenge of catastrophic forgetting in neural networks has motivated extensive research in continual learning . This work presents a novel continual learning framew…
Learning Beyond Euclid: Curvature-Adaptive Generalization for Neural Networks on Manifolds
Krisanu Sarkar
In this work, we develop new generalization bounds for neural networks trained on data supported on Riemannian manifolds. Existing generalization theories often rely on complexity…
Hindsight-Guided Momentum (HGM) Optimizer: An Approach to Adaptive Learning Rate
Krisanu Sarkar
We introduce Hindsight-Guided Momentum (HGM), a first-order optimization algorithm that adaptively scales learning rates based on the directional consistency of recent updates. Tra…