William Berrios

BI-LAVA: Biocuration with Hierarchical Image Labelling through Active Learning and Visual Analytics

Juan Trelles Trabucco, Andrew Wentzel, William Berrios, Hagit Shatkay, G. Elisabeta Marai

Computer Graphics Forum

A visual analytics and active learning system for hierarchical biomedical image classification supporting biocuration workflows.

LMUnit: Fine-grained Evaluation with Natural Language Unit Tests

Jon Saad-Falcon, Rajan Vivek, William Berrios, Nandita Shankar Naik, Matija Franklin, Bertie Vidgen, Amanpreet Singh, Douwe Kiela, Shikib Mehri

A natural language unit testing paradigm for fine-grained LLM evaluation, achieving state-of-the-art on FLASK and BigGenBench benchmarks.

Leveraging Diffusion Perturbations for Measuring Fairness in Computer Vision

Nicholas Lui, Bryan Chia, William Berrios, Candace Ross, Douwe Kiela

Proceedings of the AAAI Conference on Artificial Intelligence

Using diffusion model perturbations to create demographically balanced datasets for measuring racial bias in vision-language models.

Towards Language Models That Can See: Computer Vision Through the LENS of Natural Language

William Berrios, Gautam Mittal, Tristan Thrush, Douwe Kiela, Amanpreet Singh

A modular approach enabling LLMs to solve computer vision tasks by reasoning over descriptive vision module outputs, achieving competitive performance without multimodal training.

Joint Rotational Invariance and Adversarial Training of a Dual-Stream Transformer Yields State of the Art Brain-Score for Area V4

William Berrios, Arturo Deza

A dual-stream Vision Transformer achieving state-of-the-art Brain-Score for area V4 through joint rotational invariance and adversarial training.