This tutorial describes how to apply linear network theory to the analysis and interpretation of neural data. It introduces the concept of “sloppy models” that capture a common problem in model-fitting, in which individual model parameters are poorly constrained by available data (i.e. have “poorly/sloppily constrained parameter values”). Simple methods are illustrated for describing which combination of parameters most affect a particular model fit. This material is relevant to problems in neuroscience involving the interpretation of multi-dimensional data from recurrently connected systems.

Taught by: Mark Goldman, UC Davis

Slides:

• Linear network theory and sloppy models - Mark Goldman’s lecture slides
• Introduction to linear algebra - Mark Goldman and Emily Mackevicius slides

Additional Resources:

• Sloppy model problems
• Linear network theory problems
• MATLAB code, integration, for recurrently connected network of two neurons
• MATLAB code, amplification, for recurrently connected network of two neurons