1 / 5

Evaluating Classifier Performance in Neural Data Decoding

This project explores the efficacy of different classifiers for decoding neural data, specifically examining the superior performance of Support Vector Machines (SVMs) over Probabilistic Neural Bayes (PNB) classifiers, as reported by Graf et al. in Nature Neuroscience (2011). We aim to replicate their findings, investigate the reasons behind SVMs' better performance, and analyze their effectiveness on various data types, including computer vision features. The goal is to deepen the understanding of correlated variability in neural responses and its impact on decoding accuracy.

thom
Télécharger la présentation

Evaluating Classifier Performance in Neural Data Decoding

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Evaluating which classifiers work best for decoding neural data

  2. Background: Neural decoding neuron 1 neuron 2 neuron 3 Learning association between neural activity an image neuron n Pattern Classifier

  3. Background • A recent paper by Graf et al. (Nature Neuroscience 2011) showed that SVMs worked better than PNB classifiers for decoding information from simultaneously recorded populations of neural data from V1. • They claimed that SVMs performed better because they took into account correlated variability in neural responses • However, a more detailed examination of what led to the higher decoding accuracy was not done.

  4. Purpose of this project • Goals of this project are: • Try to replicate the finding that SVMs work better than PNB (and other) classifiers on simultaneously recorded neural data • Understand why SVMs are working better (is it really due to correlated variability? Can we say something more precise?). 3. Examine other types of data (e.g., computer vision features). Do SVMs work better, and why?

  5. Resources • Paper: • Graf et al., Decoding the activity of neuronal populations in macaque primary visual cortex Nature Neuroscience, 2011 http://www.nature.com/neuro/journal/vaop/ncurrent/abs/nn.2733.html • I can supply additional code and data that can be used to test different population decoding algorithms

More Related