Expert Load Matters: Operating Networks at High Accuracy and Low Manual Effort
Ender Konukoglu
January 26, 2024, Friday, 2:00 PM - 3:00 PM EST
Abstract
Integration of advanced machine learning algorithms in critical applications require the algorithms to be robust enough. Such systems should not lead to mistakes with severe consequences and advance what humans can achieve without the algorithm. One avenue towards this end is to consider Human-AI collaboration systems, where learning algorithms aim to reduce workload of a human expert. Algorithms do most of the work but cases that are challenging for the algorithm are planned to be sent to the human expert, such that the collaboration leads to both efficient and accurate workflow. In this talk, I will present our recent work which takes a step in this direction. I will discuss how accuracy and human load can both be integrated in a loss function and present results using neural networks trained with the proposed loss on multiple data sets. Across all the experiments, we observed that more accurate networks that require fewer delegations to human experts could be achieved. I will conclude with the current limitations and directions forward.
Bio
Dr. Ender Konukoglu is Associate Professor of Biomedical Image Computing at ETH-Zurich and a member of the Computer Vision Laboratory at the Department of Information Technology and Electrical Engineering. His research focuses on developing computational methods for analyzing medical images. The goals are to extract semantic information, perform quantitative measurements and perform population comparisons to aid diagnosis, treatment and clinical research.

He did his B.S. and M.S. degrees at Bogazici University / Electrical and Electronics Engineering Department in 2003 and 2005. He got his PhD from University of Nice Sophia Antipolis working at INRIA Sophia Antipolis Mediterranean under the supervision of Prof. Nicholas Ayache in 2009. After the PhD he worked as a post-doctoral researcher at Microsoft Research in Cambridge between 2009 and 2012. Between 2012 and 2016 he worked at Athinoula A. Martinos Center for Biomedical Imaging and Harvard Medical School as an Instructor in Radiology and Assistant in Neuroscience. He was a member of the Laboratory for Computational Neuroimaging.

In August 2016 he started as an Assistant Professor (with tenure-track) of Biomedical Image Computing at ETH-Zurich.