The Deep Brainhack was held on August 24-25 at Notman House, Montreal. This two-day hacking sprint was designed to familiarize the deep learning community with the challenges of working with brain scans and neuroimaging data, as well as introducing researchers to opportunities and applications in neuroscience presented by cutting-edge machine learning technology.
This event saw more than 100 participants representing a diversity of backgrounds that ranged from high school students to industry professionals, coming from as far away as California and Vermont.
Local researchers and graduate students presented scientific challenges in neuroimaging and data resources, as well as machine learning and deep learning techniques for problem-solving.
- Alain Dagher, MNI-BIC, McGill U. Neuroimaging research topics and Human Connectome Project data
- Christophe Grova, PERFORM, Concordia U. – EEG/MEG data and current research topics
- Olexa Bilaniuk, MILA, U. of Montreal – Deep Learning
- Paul Lemaitre, R2 (Montreal) – Machine Learning tutorial
- Greg Kiar, MCIN, McGill U. – “Brainhack 101” tools
- Ian Watt, CIM, McGill U – Reinforcement Learning
- Automated Quality Control on Autism brain scans (ABIDE)
- Automated classification of monkey action potentials
- Healthy tissue grey matter/white matter segmentation
- Brain tissue identification for PET images
- Automated 4D segmentation of “In vivo 2-photon images of axon growth”
- Generating FLAIR images from T1 anatomical brain scans
- Subject-age regression using a multi-channel neural network incorporating raw images and extracted features
- EEG signal analysis projects facilitated by NeurotechX
By the end of second day, participants gained hands-on experience with cutting edge techniques and high-performance computing, and a better understanding of the scientific applications of deep learning in neuroscience. Visit Brainhack 2017 on Github.
Day 1 Gallery
Day 2 Gallery
With special thanks to: