Logan Peters

Biography

Many machine learning models are black boxes, with internal functions that are difficult to interpret. This lack of interpretability stymies machine learning adoption in applications where transparency is critical, such as healthcare and scientific research, and limits our ability to understand the patterns identified by machine learning models.

My research focuses on developing interpretable machine learning models for use in diagnosing and treating neurological diseases and understanding how neural population dynamics encode cognitive activity. Specifically, I work with intracranial electroencephalogram signals, whose complex, non-linear, composite and noisy but informative signals I work to interpret with principled application of machine learning models. My current work focuses on the effects of neuromodulation and what that can tell us about the brain.