neale ratzlaff

Research Assistant

Oregon State University•  January 2017 - Present

Deep Learning and Computer Vision research. I have written and published papers in the fields of machine learning, reinforcement learning, anomaly detection, adversarial robustness, and uncertainty aware neural networks.

Research Intern

Horizon Robotics •  June 2019 - December 2019

Research Internship where I was the lead contributor on a project culminating in a conference paper submission to ICLR. We studied model-based reinforcement learning and we achieved state-of-the-art results on pure exploration tasks in a continuous control setting. We leveraged a new intrinsic reward based on the uncertainty of an agent's belief to drive exploration.

Software Engineering Intern

Intel•  June 2016 - July 2017

Worked on a variety of projects spanning deep learning systems for object localization and recognition, graphical UI based neural network training tools, and mixed reality computing infrastructure.

Oregon State University

Computer Science, PhD•  September 2018 - Present

Working on making deep learning systems know when they don't know. This involves research in how deep neural networks make predictions at the lowest level, with proven applications to classification, regression, computer vision, and reinforcement learning. Neural networks can be trained as a distribution, or set of solutions instead of just a single one. This distribution helps us make more informed decisions, and alert humans if a reasoned decision can not be made.