Interpretable functional specialization emerges in deep convolutional networks trained on brain signals

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11130%2F22%3A10443089" target="_blank" >RIV/00216208:11130/22:10443089 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/00064203:_____/22:10443089

Výsledek na webu

<a href="https://verso.is.cuni.cz/pub/verso.fpl?fname=obd_publikace_handle&handle=de24omQx6e" target="_blank" >https://verso.is.cuni.cz/pub/verso.fpl?fname=obd_publikace_handle&handle=de24omQx6e</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1088/1741-2552/ac6770" target="_blank" >10.1088/1741-2552/ac6770</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Interpretable functional specialization emerges in deep convolutional networks trained on brain signals

Popis výsledku v původním jazyce

OBJECTIVE: Functional specialization is fundamental to neural information processing. Here, we study whether and how functional specialization emerges in artificial deep convolutional neural networks (CNNs) during a brain-computer interfacing (BCI) task. APPROACH: We trained CNNs to predict hand movement speed from intracranial EEG (iEEG) and delineated how units across the different CNN hidden layers learned to represent the iEEG signal. MAIN RESULTS: We show that distinct, functionally interpretable neural populations emerged as a result of the training process. While some units became sensitive to either iEEG amplitude or phase, others showed bimodal behavior with significant sensitivity to both features. Pruning of highly-sensitive units resulted in a steep drop of decoding accuracy not observed for pruning of less sensitive units, highlighting the functional relevance of the amplitude- and phase-specialized populations. SIGNIFICANCE: We anticipate that emergent functional specialization as uncovered here will become a key concept in research towards interpretable deep learning for neuroscience and BCI applications.

Název v anglickém jazyce

Interpretable functional specialization emerges in deep convolutional networks trained on brain signals

Popis výsledku anglicky

OBJECTIVE: Functional specialization is fundamental to neural information processing. Here, we study whether and how functional specialization emerges in artificial deep convolutional neural networks (CNNs) during a brain-computer interfacing (BCI) task. APPROACH: We trained CNNs to predict hand movement speed from intracranial EEG (iEEG) and delineated how units across the different CNN hidden layers learned to represent the iEEG signal. MAIN RESULTS: We show that distinct, functionally interpretable neural populations emerged as a result of the training process. While some units became sensitive to either iEEG amplitude or phase, others showed bimodal behavior with significant sensitivity to both features. Pruning of highly-sensitive units resulted in a steep drop of decoding accuracy not observed for pruning of less sensitive units, highlighting the functional relevance of the amplitude- and phase-specialized populations. SIGNIFICANCE: We anticipate that emergent functional specialization as uncovered here will become a key concept in research towards interpretable deep learning for neuroscience and BCI applications.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

30103 - Neurosciences (including psychophysiology)

Projekt

<a href="/cs/project/GA20-21339S" target="_blank" >GA20-21339S: Dynamika neuronálních sítí v průběhu řešení interně a externě zaměřených kognitivních úloh</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Journal of Neural Engineering [online]

ISSN

1741-2552

e-ISSN

—

Svazek periodika

19

Číslo periodika v rámci svazku

3

Stát vydavatele periodika

GB - Spojené království Velké Británie a Severního Irska

Počet stran výsledku

15

Strana od-do

036006

Kód UT WoS článku

000792469300001

EID výsledku v databázi Scopus

2-s2.0-85130000495