A collection of projects, 'sketches', using improv as a framework for streaming analysis and adaptive experimental methods.
improv is our streaming software platform for real-time, adaptive experiments. It is flexible and easily extensible, in part to enable rapid prototyping of new analyses or other steps required for different experimental pipelines. Here, we work on a series of different analyses for streaming calcium fluorescence imaging in larval zebrafish.
This sketch focuses on using a generalized linear model (GLM) to fit the spiking activity of neurons in order to get an estimate of the connections (weights) between pairs of neurons. We adapt common machine learning approaches to the streaming setting, and compare different optimizers.
We welcome contributions to this repository of any 'sketches' that you are willing to share with the neuro community!