Second order dimensionality reduction using minimum and maximum mutual information models.
Conventional methods used to characterize multidimensional neural feature selectivity, such as spike-triggered covariance (STC) or maximally informative dimensions (MID), are limited to Gaussian stimuli or are only able to identify a small number of features due to the curse of dimensionality. To ov...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Public Library of Science (PLoS)
2011-10-01
|
| Series: | PLoS Computational Biology |
| Online Access: | https://www.ncbi.nlm.nih.gov/pmc/articles/pmid/22046122/?tool=EBI |
| _version_ | 1818435982613020672 |
|---|---|
| author | Jeffrey D Fitzgerald Ryan J Rowekamp Lawrence C Sincich Tatyana O Sharpee |
| author_facet | Jeffrey D Fitzgerald Ryan J Rowekamp Lawrence C Sincich Tatyana O Sharpee |
| author_sort | Jeffrey D Fitzgerald |
| collection | DOAJ |
| description | Conventional methods used to characterize multidimensional neural feature selectivity, such as spike-triggered covariance (STC) or maximally informative dimensions (MID), are limited to Gaussian stimuli or are only able to identify a small number of features due to the curse of dimensionality. To overcome these issues, we propose two new dimensionality reduction methods that use minimum and maximum information models. These methods are information theoretic extensions of STC that can be used with non-Gaussian stimulus distributions to find relevant linear subspaces of arbitrary dimensionality. We compare these new methods to the conventional methods in two ways: with biologically-inspired simulated neurons responding to natural images and with recordings from macaque retinal and thalamic cells responding to naturalistic time-varying stimuli. With non-Gaussian stimuli, the minimum and maximum information methods significantly outperform STC in all cases, whereas MID performs best in the regime of low dimensional feature spaces. |
| first_indexed | 2024-12-14T17:01:32Z |
| format | Article |
| id | doaj.art-3d31e147728e467f9687e6bc3f6a6e5c |
| institution | Directory Open Access Journal |
| issn | 1553-734X 1553-7358 |
| language | English |
| last_indexed | 2024-12-14T17:01:32Z |
| publishDate | 2011-10-01 |
| publisher | Public Library of Science (PLoS) |
| record_format | Article |
| series | PLoS Computational Biology |
| spelling | doaj.art-3d31e147728e467f9687e6bc3f6a6e5c2022-12-21T22:53:50ZengPublic Library of Science (PLoS)PLoS Computational Biology1553-734X1553-73582011-10-01710e100224910.1371/journal.pcbi.1002249Second order dimensionality reduction using minimum and maximum mutual information models.Jeffrey D FitzgeraldRyan J RowekampLawrence C SincichTatyana O SharpeeConventional methods used to characterize multidimensional neural feature selectivity, such as spike-triggered covariance (STC) or maximally informative dimensions (MID), are limited to Gaussian stimuli or are only able to identify a small number of features due to the curse of dimensionality. To overcome these issues, we propose two new dimensionality reduction methods that use minimum and maximum information models. These methods are information theoretic extensions of STC that can be used with non-Gaussian stimulus distributions to find relevant linear subspaces of arbitrary dimensionality. We compare these new methods to the conventional methods in two ways: with biologically-inspired simulated neurons responding to natural images and with recordings from macaque retinal and thalamic cells responding to naturalistic time-varying stimuli. With non-Gaussian stimuli, the minimum and maximum information methods significantly outperform STC in all cases, whereas MID performs best in the regime of low dimensional feature spaces.https://www.ncbi.nlm.nih.gov/pmc/articles/pmid/22046122/?tool=EBI |
| spellingShingle | Jeffrey D Fitzgerald Ryan J Rowekamp Lawrence C Sincich Tatyana O Sharpee Second order dimensionality reduction using minimum and maximum mutual information models. PLoS Computational Biology |
| title | Second order dimensionality reduction using minimum and maximum mutual information models. |
| title_full | Second order dimensionality reduction using minimum and maximum mutual information models. |
| title_fullStr | Second order dimensionality reduction using minimum and maximum mutual information models. |
| title_full_unstemmed | Second order dimensionality reduction using minimum and maximum mutual information models. |
| title_short | Second order dimensionality reduction using minimum and maximum mutual information models. |
| title_sort | second order dimensionality reduction using minimum and maximum mutual information models |
| url | https://www.ncbi.nlm.nih.gov/pmc/articles/pmid/22046122/?tool=EBI |
| work_keys_str_mv | AT jeffreydfitzgerald secondorderdimensionalityreductionusingminimumandmaximummutualinformationmodels AT ryanjrowekamp secondorderdimensionalityreductionusingminimumandmaximummutualinformationmodels AT lawrencecsincich secondorderdimensionalityreductionusingminimumandmaximummutualinformationmodels AT tatyanaosharpee secondorderdimensionalityreductionusingminimumandmaximummutualinformationmodels |