Create a Project by Logging In or Creating an Account

Wavelet correlations

Tue, 09/25/2007 - 03:06 pm

This module performs correlations on a set of wave pairs. The waves represent instances of stimulus/response pairs, and the result is an estimate of the linear transformation ("kernel") between stimulus and response. Two methods are provided: the conventional way of deriving the kernel, by transforming to the frequency domain, multiplying, normalizing, and inverse transforming; and a novel method that uses the wavelet transform to provide time-frequency representations of stimulus and response, then divides response by stimulus, optionally performs median filtering on the amplitude samples across time at each frequency, and averages over time to derive a kernel estimate.

A demonstration of these correlations is provided, along with a help file and a notebook with instructions.

Comments would be appreciated: Alan Saul .

Project Details

Current Project Release