Nhat Ho

Associate Professor@University of Texas at Austin·美国·OpenReview

研究方向

Bayesian Nonparametrics · Mixture and Hierarchical Models · Graphical Models · Statistical Learning Theory · Optimal Transport · (Non)Convex Optimization · Deep Learning · (Approximate) Bayesian Inference

Nhat Ho

HoRA: Cross-Head Low-Rank Adaptation with Joint Hypernetworks

One-Prompt Strikes Back: Sparse Mixture of Experts for Prompt-based Continual Learning

CompeteSMoE - Statistically Guaranteed Mixture of Experts Training via Competition

DoRAN: Stabilizing Weight-Decomposed Low-Rank Adaptation via Noise Injection and Auxiliary Networks

FACET: A Fragment-Aware Conformer Ensemble Transformer

Revisit Visual Prompt Tuning: The Expressiveness of Prompt Experts

Fast Estimation of Wasserstein Distances via Regression on Sliced Wasserstein Distances

Reflecting and Linking knowledge: Dynamic Label Structures for Prompt-based Continual Learning

S-Chain: Structured Visual Chain-of-Thought for Medicine

Improving Generalization with Flat Hilbert Bayesian Inference

Towards Marginal Fairness Sliced Wasserstein Barycenter

On Minimax Estimation of Parameters in Softmax-Contaminated Mixture of Experts

Lightspeed Geometric Dataset Distance via Sliced Optimal Transport

RepLoRA: Reparameterizing Low-rank Adaptation via the Perspective of Mixture of Experts

Revisiting Prefix-tuning: Statistical Benefits of Reparameterization among Prompts

ExGra-Med: Extended Context Graph Alignment for Medical Vision-Language Models

Improving Generalization with Flat Hilbert Bayesian Inference

Statistical Advantages of Perturbing Cosine Router in Mixture of Experts

X-Drive: Cross-modality Consistent Multi-Sensor Data Synthesis for Driving Scenarios

On Zero-Initialized Attention: Optimal Prompt and Gating Factor Estimation

Exploiting Hierarchical Taxonomies in Pretrained Continual Learning

LoGra-Med: Long-Context Multi-Graph Alignment for Medical Visual-Language Models

ACUS: Audio Captioning with Unbiased Sliced Wasserstein Kernel

On Expert Estimation in Hierarchical Mixture of Experts: Beyond Softmax Gating Functions