影响力指数

71.84/100

前 2.3%

全站排名 #1,494

发表论文15 篇

平均评分5.5

年均产出5.0 篇/年

Adriana Romero-Soriano

Core Industry Member@Mila·加拿大·OpenReview

研究方向

deep learning · generative models · synthetic data · responsible AI · active sensing · computer vision · graph structured data · biomedical data · semantic segmentation · image classification · convolutional neural networks · model compression · sparsity · unsupervised learning · neural networks

The Intricate Dance of Prompt Complexity, Quality, Diversity and Consistency in T2I Models

ICLR 2026Poster

Grounding Computer Use Agents on Human Demonstrations

ICLR 2026Poster

PairBench: Are Vision-Language Models Reliable at Comparing What They See?

ICLR 2026Rejected

Entropy Rectifying Guidance for Diffusion and Flow Models

NeurIPS 2025Poster

Object-centric binding in Contrastive Language-Image Pretraining

NeurIPS 2025Poster

Increasing the Utility of Synthetic Images through Chamfer Guidance

NeurIPS 2025Poster

Improving the Scaling Laws of Synthetic Data with Deliberate Practice

Boosting Latent Diffusion with Perceptual Objectives

ICLR 2025Poster

OC-CLIP : Object-centric binding in Contrastive Language-Image Pretraining

ICLR 2025Rejected

FairLoRA: Unpacking Bias Mitigation in Vision Models with Fairness-Regularized Low-Rank Adaptation

ICLR 2025Rejected

Augmented Conditioning is Enough for Effective Training Image Generation

ICLR 2025Withdrawn

合作者 (20)

Michal Drozdzal

Reihaneh Rabbany

Karteek Alahari