Methods & Data Analysis
Exploring Neural Data (Online, Linden & Sheinberg):
Exploring Neural Data is an opportunity to learn about neuroscience research and explore questions related to how brains work. It is an introductory level course designed to help you understand the real-life challenges faced by neuroscientists as they work with the large amount of data they collect from the brain. Leading neuroscientists will give tours of their labs, describe their research, and explain their data analytic techniques. You will have the chance to explore actual data collected in these researchers’ labs.
Throughout the course, you will gain knowledge in three main areas: basic principles of neuroscience and questions driving research in this field; programming with the open-source language Python; and essential techniques for data analysis. We will begin by exploring single neurons, then turn our attention to multiple neurons, and finally consider tools that sample from tens of thousands of neurons. At the end of the class, you will have the opportunity to investigate in detail a data set of your choice provided by one of the researchers whose work you’ve learned about.
Link to coursera site.
Statistical Neuroscience (NEUR2110-S01CRN, Spring, Truccolo):
A lecture and computing lab course for senior undergraduate and graduate students with background in either systems neuroscience or applied math/biomedical engineering on the statistical analysis and modeling of neural data, with hands-on Matlab(Octave)/Python-based applications to real and simulated data. Topics will include signal processing, hypothesis testing and statistical inference, modeling of multivariate time series and stochastic processes in neuroscience and neuroengineering, neural point processes, time and spectral domain analyses, and state-space models. Example datasets include neuronal spike trains, local field potentials, ECoG/EEG, and fMRI. Sign-up sheet in Sidney Frank Hall, Room 315 beginning on the first day of registration. Instructor permission required.
Machine Learning Methods for Brain and Cog Sciences (TBD, Serre):
Covers statistical methods as well as machine learning methods (regularization, Granger causality, SVM, etc) with hands on application to neuroscience data (fMRI, spiking data, LFP, etc).
Quantitative Model Fitting (TBD, Frank):
Model fitting to behavioral data (e.g., accuracy, response time distributions) used to assess latent computations to be regressed against neural data. Covers model selection, maximum likelihood estimates, parameter stability and identifiabilty, hierarchical bayesian parameter estimation across groups and individuals (MCMC, EM), frequentist and bayesian statistics on inferred parameters, and applications to neural data. Prominent examples in reinforcement learning, decision making, inhibition and memory models, but tools extended to any domain. Programming in matlab and/or python.