Imaging Science Ph.D. Defense - Ryne Roady

Deep Learning with Open Set Classification in Large-scale and Continual Learning Models

Ryne P. Roady
Imaging Science Ph.D. Candidate
Chester F. Carlson Center for Imaging Science, RIT

Join Zoom Meeting

Abstract
:

Supervised classification methods often assume the train and test data distributions are the same and that all classes in the test set are present in the training set. However, deployed classifiers require the ability to recognize inputs from outside the training set as unknowns and update representations in near real-time to account for novel concepts unknown during offline training. This problem has been studied under multiple paradigms including out-of-distribution detection and open set recognition; however, for convolutional neural networks, there have been two major approaches: 1) inference methods to separate knowns from unknowns and 2) feature space regularization strategies to improve model robustness to novel inputs. In this dissertation, we explore how these two strategies perform in large-scale computer vision datasets and establish baselines for the first time for open set learning in continual and online data streams including a novel dataset created for evaluating image open set classification capabilities of streaming learning algorithms. We draw conclusions as to what the most computationally efficient means of detecting novelty in pre-trained models and what properties of an efficient open world learning algorithm should possess. We finally propose a new feature regularization scheme that achieves state-of-the-art robustness to novel inputs and a novel open world learning model based on a modular approach. This open world learner uses self-supervised contrastive learning for feature representation and an incrementally learned classification stage utilizing replay and confidence loss to enable open set classification capabilities in streaming learning settings.

Intended Audience:
Undergraduates, graduates, and experts. Those with interest in the topic.


Contact
Beth Lockwood
Event Snapshot
When and Where
August 17, 2020
2:00 pm - 3:00 pm
Room/Location: Zoom
Who

Open to the Public

Interpreter Requested?

No

Topics
research