Home → News and Events → Colloquia → Colloquium
3/21/2019 Faculty Candidate: Daniel Haehn
When: 10:00AM - 11:30AM , March 21, 2019
Speaker: Daniel Haehn
With advances in computer science, my goal is to reduce the gap between natural and artificial intelligence: First, with bottom-up investigations for how visual computing methods can aid brain connectivity analysis, and second, exploring top-down how perceptual studies can increase our understanding of machine learning models. Connectomics aims to study neurons and their connections using modern electron microscopes. Resulting image volumes can be petabytes in size and we need automatic segmentation methods to process the data. However, no automatic method is perfect and the output requires proofreading by humans. In this talk, I present our work on interactive and semi-automatic methods for correcting cell membrane segmentations. Then, I will switch from the micro scale to the human scale and consider our recent studies of machine perception, where we replicate Cleveland and McGill’s 1984 human graphical perception experiments using Convolutional Neural Networks.
Daniel Haehn is a biomedical imaging and visualization researcher at Harvard University. Prior to his doctoral studies, Daniel worked as a researcher at Boston Children’s Hospital where he was the lead developer of XTK (goXTK.org), a pioneering web-based computer graphics framework for medical images. Daniel also worked at Brigham and Women’s Hospital and the University of Pennsylvania as one of the top contributors to 3D Slicer, a popular medical image analysis software. Daniel is a strong supporter of open science and all of his research is freely available.