IEEE Systems, Man and Cybernetics Magazine - October 2021 - 25

T
he automatic feature-extraction capability of
deep neural networks (DNNs) endows them
with the potential for analyzing complicated
electroencephalogram (EEG) data captured
from brain functionality research. This article
investigates the potential coherent correspondence
between the region of interest (ROI) for DNNs to explore,
and the ROI for conventional neurophysiological-oriented
methods to work with, as exemplified in the case of a
working memory study. The attention mechanism induced
by global average pooling (GAP) is applied to a public EEG
data set of a working memory test to unveil these coherent
ROIs via a classification problem. The results show the
potential alignment of the ROIs from different discipline
methods, and consequently asserts
the confidence and promise of utilizing
DNNs for EEG data analysis.
DNNs for EEG
The success of DNNs in various
fields has drawn the attention of
brain researchers to apply these
models for EEG data analysis,
either to promote deeper neuroscientific
understandings or to
faci l itate wider brain-computer
interface applications [1]-[4].
Although it is not as strict as a clinical
requirement, the black-box
operations of DNNs still arouse
plenty of concerns. For example, it is difficult to reach
intuitive interpretations to the model behavior without
knowing the underlying mechanism. Despite there being
ways to interpret the neural network dynamics that foster
intuition [5]-[7], approaches to combine techniques in
other disciplines to improve the performance [8]-[11], and
methods to quantify the uncertainty of DNNs to increase
the trustworthiness [12], it is still necessary to study the
characteristics and to assert the feasibility of network
models by linking and comparing the achievements from
other methodologies.
Properties such as high temporal resolution, mobility,
economy, and so on [13] speak to the EEG's indispensable
role in brain research. To blindly employ DNNs to analyze
EEG data might provide satisfying results catered to the
application itself; however, a loss of intuition can hinder
the theoretical depth of the achievements. Compared with
other DNN-affiliated applications, certain brain research
experiments are conducted only in restricted environments
or idealized conditions due to practical constraints,
and the results are then demonstrated to the
community without actual deployments of the models.
This can lead to potential unintended consequences being
buried in the learned models when using neural networks,
thus turning the overall research into hallucinations,
which can also produce confusing results at a later stage
[14]-[16]. Hence, cross validating the results of DNN models
in EEG data analysis by referencing knowledge from
other disciplines is more critical in this field than in other
utilizations of DNNs.
This article aims to address some of these concerns by
Properties such
as high temporal
resolution, mobility,
economy, and so on
speak to the EEG's
indispensable role in
brain research.
considering the implicit attention mechanism induced by
class-activation mapping via GAP [17]. Actually, for certain
fundamental brain research topics, such as working memory
[18], there are already common recognitions of the
neuronal basis underpinning this mechanism. It is regarded
that the prefrontal cortex (PFC) and hippocampus are
actively involved in the functioning of working memory
[19]-[21]. Therefore, it is expected that for the working
memory load test, when a DNN is applied for harvesting
EEG features automatically [2], the
model should not switch absurdly
among different areas over the
scalp. Although there is still some
dispute over the characteristics of
neuronal activities, such as,
in
essence, these are discrete dynamics
or sustained activities [22], the
dist inct ion among workload
extents should incur only the network
to focus on approximately
common areas.
The work in [17] demonstrates
that from the adoption of GAP,
even if trained via class-level
labels, the network can still exhibit
some localization ability, which enables it to identify the
discriminative image regions in a picture. This apparent
simplicity provides the means to verify some conjectures,
for example, in the case of working memory, and whether
the network tends to look at similar areas with different
activation strengths or not. It is known that due to the stilllimited
understanding of working memory, enforcing the
network to explicitly explore specific regions might not be
a good idea. Instead, the network's intrinsic dynamics and
behaviors, if they are in accordance with certain assumptions,
are strong evidence of the feasibility of utilizing
DNNs to analyze EEG data. And this is the purpose and
potential contribution of our work in this article.
Methodology
As mentioned previously, working memory is an indispensable
component in many important cognitive
0.5
3.5
Test New Trial
Figure 1. The experiment paradigm for conducting
the research.
October 2021 IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE 25

IEEE Systems, Man and Cybernetics Magazine - October 2021

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - October 2021

Contents
IEEE Systems, Man and Cybernetics Magazine - October 2021 - Cover1
IEEE Systems, Man and Cybernetics Magazine - October 2021 - Cover2
IEEE Systems, Man and Cybernetics Magazine - October 2021 - Contents
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 2
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 3
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 4
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 5
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 6
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 7
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 8
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 9
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 10
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 11
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 12
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 13
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 14
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 15
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 16
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 17
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 18
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 19
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 20
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 21
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 22
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 23
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 24
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 25
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 26
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 27
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 28
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 29
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 30
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 31
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 32
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 33
IEEE Systems, Man and Cybernetics Magazine - October 2021 - 34
IEEE Systems, Man and Cybernetics Magazine - October 2021 - Cover3
IEEE Systems, Man and Cybernetics Magazine - October 2021 - Cover4
https://www.nxtbook.com/nxtbooks/ieee/smc_202110
https://www.nxtbook.com/nxtbooks/ieee/smc_202107
https://www.nxtbook.com/nxtbooks/ieee/smc_202104
https://www.nxtbook.com/nxtbooks/ieee/smc_202101
https://www.nxtbook.com/nxtbooks/ieee/smc_202010
https://www.nxtbook.com/nxtbooks/ieee/smc_202007
https://www.nxtbook.com/nxtbooks/ieee/smc_202004
https://www.nxtbook.com/nxtbooks/ieee/smc_202001
https://www.nxtbook.com/nxtbooks/ieee/smc_201910
https://www.nxtbook.com/nxtbooks/ieee/smc_201907
https://www.nxtbook.com/nxtbooks/ieee/smc_201904
https://www.nxtbook.com/nxtbooks/ieee/smc_201901
https://www.nxtbook.com/nxtbooks/ieee/smc_201810
https://www.nxtbook.com/nxtbooks/ieee/smc_201807
https://www.nxtbook.com/nxtbooks/ieee/smc_201804
https://www.nxtbook.com/nxtbooks/ieee/smc_201801
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_1017
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0717
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0417
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0117
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_1016
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0716
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0416
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0116
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_1015
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0715
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0415
https://www.nxtbook.com/nxtbooks/ieee/systems_man_cybernetics_0115
https://www.nxtbookmedia.com