Modern Machine Learning as a Benchmark for Fitting Neural Responses
Neuroscience has long focused on finding encoding models that effectively ask “what predicts neural spiking?” and generalized linear models (GLMs) are a typical approach. It is often unknown how much of explainable neural activity is captured, or missed, when fitting a model. Here we compared the pr...
Main Authors: | , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Frontiers Media S.A.
2018-07-01
|
Series: | Frontiers in Computational Neuroscience |
Subjects: | |
Online Access: | https://www.frontiersin.org/article/10.3389/fncom.2018.00056/full |
_version_ | 1811274884391632896 |
---|---|
author | Ari S. Benjamin Hugo L. Fernandes Tucker Tomlinson Pavan Ramkumar Pavan Ramkumar Chris VerSteeg Raeed H. Chowdhury Raeed H. Chowdhury Lee E. Miller Lee E. Miller Lee E. Miller Konrad P. Kording Konrad P. Kording |
author_facet | Ari S. Benjamin Hugo L. Fernandes Tucker Tomlinson Pavan Ramkumar Pavan Ramkumar Chris VerSteeg Raeed H. Chowdhury Raeed H. Chowdhury Lee E. Miller Lee E. Miller Lee E. Miller Konrad P. Kording Konrad P. Kording |
author_sort | Ari S. Benjamin |
collection | DOAJ |
description | Neuroscience has long focused on finding encoding models that effectively ask “what predicts neural spiking?” and generalized linear models (GLMs) are a typical approach. It is often unknown how much of explainable neural activity is captured, or missed, when fitting a model. Here we compared the predictive performance of simple models to three leading machine learning methods: feedforward neural networks, gradient boosted trees (using XGBoost), and stacked ensembles that combine the predictions of several methods. We predicted spike counts in macaque motor (M1) and somatosensory (S1) cortices from standard representations of reaching kinematics, and in rat hippocampal cells from open field location and orientation. Of these methods, XGBoost and the ensemble consistently produced more accurate spike rate predictions and were less sensitive to the preprocessing of features. These methods can thus be applied quickly to detect if feature sets relate to neural activity in a manner not captured by simpler methods. Encoding models built with a machine learning approach accurately predict spike rates and can offer meaningful benchmarks for simpler models. |
first_indexed | 2024-04-12T23:27:03Z |
format | Article |
id | doaj.art-78920c26a9a5463a97819d6ed0139a63 |
institution | Directory Open Access Journal |
issn | 1662-5188 |
language | English |
last_indexed | 2024-04-12T23:27:03Z |
publishDate | 2018-07-01 |
publisher | Frontiers Media S.A. |
record_format | Article |
series | Frontiers in Computational Neuroscience |
spelling | doaj.art-78920c26a9a5463a97819d6ed0139a632022-12-22T03:12:22ZengFrontiers Media S.A.Frontiers in Computational Neuroscience1662-51882018-07-011210.3389/fncom.2018.00056317343Modern Machine Learning as a Benchmark for Fitting Neural ResponsesAri S. Benjamin0Hugo L. Fernandes1Tucker Tomlinson2Pavan Ramkumar3Pavan Ramkumar4Chris VerSteeg5Raeed H. Chowdhury6Raeed H. Chowdhury7Lee E. Miller8Lee E. Miller9Lee E. Miller10Konrad P. Kording11Konrad P. Kording12Department of Bioengineering, University of Pennsylvania, Philadelphia, PA, United StatesDepartment of Physical Medicine and Rehabilitation, Rehabilitation Institute of Chicago, Northwestern University, Chicago, IL, United StatesDepartment of Physiology, Northwestern University, Chicago, IL, United StatesDepartment of Physical Medicine and Rehabilitation, Rehabilitation Institute of Chicago, Northwestern University, Chicago, IL, United StatesDepartment of Neurobiology, Northwestern University, Evanston, IL, United StatesDepartment of Biomedical Engineering, Northwestern University, Evanston, IL, United StatesDepartment of Physiology, Northwestern University, Chicago, IL, United StatesDepartment of Biomedical Engineering, Northwestern University, Evanston, IL, United StatesDepartment of Physical Medicine and Rehabilitation, Rehabilitation Institute of Chicago, Northwestern University, Chicago, IL, United StatesDepartment of Physiology, Northwestern University, Chicago, IL, United StatesDepartment of Biomedical Engineering, Northwestern University, Evanston, IL, United StatesDepartment of Bioengineering, University of Pennsylvania, Philadelphia, PA, United StatesDepartment of Neuroscience, University of Pennsylvania, Philadelphia, PA, United StatesNeuroscience has long focused on finding encoding models that effectively ask “what predicts neural spiking?” and generalized linear models (GLMs) are a typical approach. It is often unknown how much of explainable neural activity is captured, or missed, when fitting a model. Here we compared the predictive performance of simple models to three leading machine learning methods: feedforward neural networks, gradient boosted trees (using XGBoost), and stacked ensembles that combine the predictions of several methods. We predicted spike counts in macaque motor (M1) and somatosensory (S1) cortices from standard representations of reaching kinematics, and in rat hippocampal cells from open field location and orientation. Of these methods, XGBoost and the ensemble consistently produced more accurate spike rate predictions and were less sensitive to the preprocessing of features. These methods can thus be applied quickly to detect if feature sets relate to neural activity in a manner not captured by simpler methods. Encoding models built with a machine learning approach accurately predict spike rates and can offer meaningful benchmarks for simpler models.https://www.frontiersin.org/article/10.3389/fncom.2018.00056/fullencoding modelsneural codingtuning curvesmachine learninggeneralized linear modelGLM |
spellingShingle | Ari S. Benjamin Hugo L. Fernandes Tucker Tomlinson Pavan Ramkumar Pavan Ramkumar Chris VerSteeg Raeed H. Chowdhury Raeed H. Chowdhury Lee E. Miller Lee E. Miller Lee E. Miller Konrad P. Kording Konrad P. Kording Modern Machine Learning as a Benchmark for Fitting Neural Responses Frontiers in Computational Neuroscience encoding models neural coding tuning curves machine learning generalized linear model GLM |
title | Modern Machine Learning as a Benchmark for Fitting Neural Responses |
title_full | Modern Machine Learning as a Benchmark for Fitting Neural Responses |
title_fullStr | Modern Machine Learning as a Benchmark for Fitting Neural Responses |
title_full_unstemmed | Modern Machine Learning as a Benchmark for Fitting Neural Responses |
title_short | Modern Machine Learning as a Benchmark for Fitting Neural Responses |
title_sort | modern machine learning as a benchmark for fitting neural responses |
topic | encoding models neural coding tuning curves machine learning generalized linear model GLM |
url | https://www.frontiersin.org/article/10.3389/fncom.2018.00056/full |
work_keys_str_mv | AT arisbenjamin modernmachinelearningasabenchmarkforfittingneuralresponses AT hugolfernandes modernmachinelearningasabenchmarkforfittingneuralresponses AT tuckertomlinson modernmachinelearningasabenchmarkforfittingneuralresponses AT pavanramkumar modernmachinelearningasabenchmarkforfittingneuralresponses AT pavanramkumar modernmachinelearningasabenchmarkforfittingneuralresponses AT chrisversteeg modernmachinelearningasabenchmarkforfittingneuralresponses AT raeedhchowdhury modernmachinelearningasabenchmarkforfittingneuralresponses AT raeedhchowdhury modernmachinelearningasabenchmarkforfittingneuralresponses AT leeemiller modernmachinelearningasabenchmarkforfittingneuralresponses AT leeemiller modernmachinelearningasabenchmarkforfittingneuralresponses AT leeemiller modernmachinelearningasabenchmarkforfittingneuralresponses AT konradpkording modernmachinelearningasabenchmarkforfittingneuralresponses AT konradpkording modernmachinelearningasabenchmarkforfittingneuralresponses |