UROP Openings

Have a UROP opening you would like to submit?

Please fill out the form.

Submit your UROP opening

Decoding the Brain Using Deep Learning


Term:

IAP

Department:

MAS: Media Arts and Sciences

Faculty Supervisor:

Pattie Maes

Faculty email:

pattie@media.mit.edu

Apply by:

Jan 10 2020

Contact:

+1-857-500-0980

Project Description

We record brain waves from subjects watching videos generated by a GAN (generative adversarial network) and process and model the obtained signals using modern deep neural networks. The goal is to decode and extract images from brain activity and translate thoughts into images. UROPs will be working on the following tasks: 1- EEG recording: You will learn to set up an EEG recording environment and the recording headset. You will also learn to use an EEG acquisition software tool. 2- EEG data processing: You will use various signal processing algorithms and filters (in numpy and scipy) to prepare the signals for downstream machine learning and visualization tasks. 3- Deep Learning with sequences: You will use convolutional, recurrent and transformer neural networks to build decoding models for EEG signals. In particular, you will help with testing and visualizing various models. * We prefer applicants who are available for the Summer term as well.

Pre-requisites

1- Interest in experimental work in particular recording brain waves 2- Solid background in linear algebra, probability, Markov models, python, numpy, scipy and scikit-learn 3- Background in machine learning (regression, clustering, neural networks, dimensionality reduction (PCA, ICA, SVD)) 4- experience with deep learning (Keras/PyTorch/TensorFlow)