Dimitris N. Metaxas

Board of Governors Professor@Rutgers University·美国·OpenReview

研究方向

computer vision · medical image analysis · machine learning · computer graphics · Foundation Models · Multimodal Learning · Explainability · Interpretability · Dynamical Systems Learning · Generative AI · Diffusion Methods · Generative methods · Explainable AI · Agentic AI · LLMs · VLSs · Physics-based modeling · deformable models · deep neural networks · 3D object and shape estimation · motion estimation from video · human behavior analytics · healthcare applications · AI

Dimitris N. Metaxas

K-Prism: A Knowledge-Guided and Prompt Integrated Universal Medical Image Segmentation Model

PrefGen: Multimodal Preference Learning for Preference-Conditioned Image Generation

TokUR: Token-Level Uncertainty Estimation for Large Language Model Reasoning

Learning User Preferences for Image Generation Models

Visual Sparse Steering (VS2): Unsupervised Adaptation for Image Classification via Sparsity-Guided Steering Vectors

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

Re-Meanflow: Efficient One-Step Generative Modeling via Meanflow on Rectified Trajectories

Exploring the Entropy Mechanism in LLM Agents On-policy Optimization

INSES: Intelligent Navigation and Similarity Enhanced Search for Knowledge Graph Reasoning

Test-Time Spectrum-Aware Latent Steering for Zero-Shot Generalization in Vision-Language Models

Improved Training Technique for Latent Consistency Models

Implicit In-context Learning

LoR-VP: Low-Rank Visual Prompting for Efficient Vision Model Adaptation

AutoEdit: Automatic Hyperparameter Tuning for Image Editing

Discrete Inversion: A Controllable Latent Space for Multinomial Diffusion and Masked Generative Models

The Hidden Life of Tokens: Reducing Hallucination of Large Vision-Language Models Via Visual Information Steering