Rene Vidal

Full Professor@University of Pennsylvania·美国·OpenReview

研究方向

interpretable learning · vision and language · adversarial learning · scene graphs · nonconvex optimization · deep learning theory · sparse and low-rank representation · subspace clustering · action recognition · semantic segmentation

Rene Vidal

Shuffling the Data, Extrapolating the Step: Sharper Bias In Constant Step-Size SGD

A Spectral Characterization of Generalization in GCN: Escaping the Curse of Dimensionality

SSD: A Sparse Semantic Defense Against Semantic Adversarial Attacks to Image Classifiers

Certifiably Robust Classifiers: Bridging the Gap Between Theory and Practice

LoRA Provably Reduces Forgetting and Enables Adapter Merging in Multiclass Linear Classification

Softmax-Induced Ill-Conditioning in Transformer Models

Neural Collapse under Gradient Flow on Shallow ReLU Networks for Orthogonally Separable Data

Conformal Information Pursuit for Interactively Guiding Large Language Models

MotionBind: Multi-Modal Human Motion Alignment for Retrieval, Recognition, and Generation

Convergence Rates for Gradient Descent on the Edge of Stability for Overparametrised Least Squares

InCoDe: Interpretable Compressed Descriptions For Image Generation

SECA: Semantically Equivalent and Coherent Attacks for Eliciting LLM Hallucinations

Gradient Flow Provably Learns Robust Classifiers for Data from Orthonormal Clusters

Gradient Flow Provably Learns Robust Classifiers for Orthonormal GMMs

Guarantees of a Preconditioned Subgradient Algorithm for Overparameterized Asymmetric Low-rank Matrix Recovery

LoRanPAC: Low-rank Random Features and Pre-trained Models for Bridging Theory and Practice in Continual Learning

Gradient Descent and Attention Models: Challenges Posed by the Softmax Function

KDA: A Knowledge-Distilled Attacker for Scalable LLM Red Teaming

Duality Based Generalization Bounds for Parallel Positively Homogeneous Networks