During 2020’s fall a team from the Machine Learning Journal Club participated in the ‘ProjectX Climate Change’ competition held by the University of Toronto (www.projectx2020.com).

The event, which required the participant to be undergraduated students, had the objective of allowing the participants to find new and innovative way of exploiting ML techniques to address problems related to Climate Change. Teams had the opportunity to talk with domain experts, scientists and businessmen in order to better gain knowledge and insights about the chosen field of operation.
Of the three possible categories, our team was appointed to the “extreme events” one: after a deep review of existing literature on the subjet, the team decided to focus on wildfire spread prediction.
A brand new technique, NeuralPDE, was used, resulting in a completly new approach to the subject with a lot of potential for future development and expansion. The work culminated in an open-access paper as well as in a lot of knowledge for the team and the MLJC as a whole.

You can find the link to the paper here: https://arxiv.org/abs/2012.06825 and the link to the Github material here: https://github.com/MachineLearningJournalClub/MLJC-UniTo-ProjectX-2020-public

Open source code
Open publication

Between the 10th and 12th of October, two teams from the Machine Learning Journal Club took part to the Virtual Br41n.io hackathon, held in Toronto (BR41N.IO).

Br41n.io is a two days event designed to be a learning experience for developers, students, scientists and engineers who cram and build Brain Computer Interface (BCI) applications.
Our members approached the data analysis projects, facing the hard challenge of decoding the EEG signal and optimizing pre-processing, feature extraction, and classification algorithms.
One Team dealt with a motor Imagery dataset from a chronic stroke patient.
The other team analyzed a vibrotactile P300 BCI data-set from a patient with disorders of consciousness.

Open source code