Sparse Coding Modifications for Machine Learning
- Thursday, March 1, 2018 from 11:00am to 11:50am
- Strand Union Building, 233 - view map
Please join us for ECE Computer Faculty Candidate Mr. Bradley Whitaker’s presentation.
Abstract: Cross-disciplinary work involving machine learning has led to recent advances in many areas, but there are still situations where traditional machine learning methods struggle. My research focuses on overcoming two important obstacles that appear in the context of many classification tasks, from healthcare to farming. First, how can we develop more effective machine learning tools when there is a lack of labeled data? Second, how do we model datasets where some types of interesting events occur very infrequently? In this talk, I will present an application-based overview of machine learning, discuss sparse coding as a feature extraction tool that can be used in machine learning classification, and introduce modifications I have made to sparse coding in order to deal with the unfortunate realities of data collection in some applications.