Nick Pawlowski

Needles in Haystacks: On Classifying Tiny Objects in Large Images

2019 · arXiv (Cornell University)

In some important computer vision domains, such as medical or hyperspectral imaging, we care about the classification of tiny objects in large images. However, most Convolutional Neural Networks (CNNs) for image classification were developed using …

Simultaneous Missing Value Imputation and Structure Learning with Groups

2021 · arXiv (Cornell University)

Learning structures between groups of variables from data with missing values is an important task in the real world, yet difficult to solve. One typical scenario is discovering the structure among topics in the education …

Understanding Causality with Large Language Models: Feasibility and Opportunities

2023 · arXiv (Cornell University)

We assess the ability of large language models (LLMs) to answer causal questions by analyzing their strengths and weaknesses against three types of causal question. We believe that current LLMs can answer causal questions with …