Talks and presentations

Multipath Face Recognition

June 01, 2021

Presentation, Drexel AI, Virtual

In this talk, I present some of our more recent work in multipath sparse coding with applications in mitigating bias in face recognition.

Fighting Bias in AI

April 01, 2021

Panel Presentation, PhillyTechWeek, Virtual

Sponsored by Drexel University’s College of Computing & Informatics(CCI) and CCI’s Diversity, Equity & Inclusion Council, join us for a conversation about fighting bias in artificial intelligence (AI). Mathematical models are often viewed as fair and objective. One might think that algorithms do not “see” race and therefore cannot be prejudiced; they base their decisions upon big data patterns and correlations that arise from statistics. However, in an experiment conducted by the American Civil Liberties Union (ACLU), Amazon’s face recognition system falsely matched 28 members of U.S. Congress with mugshots where false matches were disproportionately of people of color. In the past few years, studies have shown that algorithms can exhibit racial and gender bias, discriminate within a computer-vision facial recognition systems, and encode gendered bias in natural language processing. As AI becomes more pervasive in consumer-based technology, it is important that considerations be taken to prevent bias in the algorithmic decision making process. Panelists will share their knowledge of this developing topic and discuss current projects.

AI/ML Reading group at Drexel University

September 01, 2020

Reading group, Drexel University AIML Weekly Reading Group, Philadelphia, PA

The goal of the group is to keep up with the literature/state of the art, learn about others research interests, form possible collaborations, and provide a venue for those to practice upcoming talks or presentations. Youtube Channel

CVPR 2020

July 16, 2020

Conference Presentation, CVPR2020, CVPR

Our experiments show that by leveraging sparsity and modeling the biological mechanisms at a cellular level, we are able to mitigate the effect of adversarial alterations to the signal that have no perceptible meaning. Furthermore, we present and illustrate the effects of top-down functional processes that contribute to the inherent immunity in human perception in the context of exploiting these properties to make a more robust machine vision system.

NICE 2019

December 01, 2019

Conference Presentation, Neuro inspired Computational Elements, NY

Spatiotemporal Sequence Memory for Prediction using Deep Sparse Coding. For our project, we sought to create a predictive vision model using spatiotemporal sequence memories learned from deep sparse coding. This model is implemented using a biologically inspired architecture: one that utilizes sequence memories, lateral inhibition, and top-down feedback in a generative framework.

Digital Pathology Annotation Data 2017

February 01, 2017

Conference Presentation, SPIE Medical Imaging 2017, San Diego, CA

In the field of digital pathology, there is an explosive amount of imaging data being generated. Thus, there is an ever growing need to create assistive or automatic methods to analyze collections of images for screening and classification. Machine learning, specifically deep learning algorithms, developed for digital pathology have the potential to assist in this way. Deep learning architectures have demonstrated great success over existing classification models but require massive amounts of labeled training data that either doesn’t exist or are cost and time prohibitive to obtain. In this project, we present a framework for representing, collecting, validating, and utilizing cytopathology features for improved neural network classification.

wHealth - A Window to your future health

January 01, 2013

Invited Presentation, Health 2.0 Fall Conference, CA

wHealth is an interactive storytelling application that can provide insight into a user’s willingness to pay for health care, provide insight into how quality information, and compare aggregate rates or perform subgroup analyses, i.e. gender/age/income differences in what factors are most important. We use gamification techniques such as storytelling, personalization, and immediate feedback to drive user engagement. We are excited to announce that we are the first place winner of the Robert Wood Johnson Foundation Games to Generate Data Challenge and the recipient of $100,000!