Toward an Improvement of the Analysis of Neural Coding

ALEGRE-CORTÉS, JAVIER; SOTO-SÁNCHEZ, CRISTINA; ALBARRACÍN, ANA L.; FARFÁN, FERNANDO D.; VAL-CALVO, MIKEL; FERRANDEZ, JOSÉ M.; FERNANDEZ, EDUARDO

Frontiers in Neuroinformatics

Huazhong University of Science and Technology

Año: 2018 vol. 11

Machine learning and artificial intelligence have strong roots on principles of neuralcomputation. Some examples are the structure of the first perceptron, inspired in theretina, neuroprosthetics based on ganglion cell recordings or Hopfield networks. Inaddition, machine learning provides a powerful set of tools to analyze neural data,which has already proved its efficacy in so distant fields of research as speechrecognition, behavioral states classification, or LFP recordings. However, despite thehuge technological advances in neural data reduction of dimensionality, pattern selection,and clustering during the last years, there has not been a proportional development ofthe analytical tools used for Time?Frequency (T?F) analysis in neuroscience. Bearing thisin mind, we introduce the convenience of using non-linear, non-stationary tools, EMDalgorithms in particular, for the transformation of the oscillatory neural data (EEG, EMG,spike oscillations?) into the T?F domain prior to its analysis with machine learning tools.We support that to achieve meaningful conclusions, the transformed data we analyzehas to be as faithful as possible to the original recording, so that the transformationsforced into the data due to restrictions in the T?F computation are not extended tothe results of the machine learning analysis. Moreover, bioinspired computation suchas brain?machine interface may be enriched from a more precise definition of neuronalcoding where non-linearities of the neuronal dynamics are considered.