Special Guest Seminar
Date:
Lecture / Seminar
Time: 10:00-11:00
Title: Extracting the invisible - visual interpretability of deep learning models in cell imaging
Location: Arthur and Rochelle Belfer Building for Biomedical Research
Lecturer: Dr. Assaf Zaritsky
Abstract: Deep learning (aka “AI”) has emerged as a powerful technique to identify hid ... Read more Deep learning (aka “AI”) has emerged as a powerful technique to identify hidden patterns that exceed human intuition in biomedical imaging data. However, this success comes at the cost of interpretability making deep learning a “black box” lacking human meaningful explanations for the models’ decision. Interpretability is especially critical in biomedical domains, because understanding the “cause” for a machine’s prediction is key for the generation of new biological insight and testable hypotheses. In this seminar I will present computational methods that we developed to "reverse-engineer" the model’s decision in an intuitive biologically-meaningful manner and their applications to multiple bioimaging domains. The seminar will be designed for life scientists assuming no prior computational background. Close abstract
