IEEE Computational Intelligence Magazine - February 2022 - 29

0.2
0.4
0.6
0.2
0.4
0.2
-0.4
-0.2
0.2
-0.2
-0.2
0 123 4 01234
0.1
-0.3
-0.2
-0.1
0 123 4 01234
-0.2
-0.1
0 123 4 01234
(a) Six fNIRS Signals.
1
No R1 R2 R3 R4
1
2
3
4
5
6
7
8
9
10
10
(d) Conceptual Label (CoL) MVM.
20
30
∆ Conc. Hb × 10-6 µM
(c) Interval Type 2 Membership Functions.
Inactive
Visual
Cortical Networks:
Prefrontal Cortex
CCh8h8
Extended System
Interval Type-2 Fuzzy
Logic System (IT2-FLS)
Core System
Auditory
Cortical Networks:
Prefrontal Cortex
Ch8
Evolutionary Algorithm
Genetic Algorithm (GA)
(e) xMVPA Inference Mechanism
P3
P6
Extended System
P5
Ch9
Core System
(f) The Patterns delineated by xMVPA.
Occipital Cortex
Ch1
Temporal Cortex
P4
Ch4 Ch5
P2
P2
Temporal Cortex
Ch4
Ch6
Ch4
P1
P2
Ch6
P1
P1
Ch2
Ch2
Occipital Cortex
Ch1
Ch1
Active
Very
Active
40
50
R5
R6 Stimulus
0.8
0.6
0.4
0.2
(b) Numerical Multivariate Matrix (MVM).
Statistical
Feature
Trial No R1
1
2
3
4
5
6
7
8
9
10
-4.10 ∗ 10-6
-1.01 ∗ 10-5
2.13 ∗ 10-5
2.49 ∗ 10-5
4.77 ∗ 10-5
-6.68 ∗ 10-5
-6.22 ∗ 10-5
3.04 ∗ 10-5
3.46 ∗ 10-5
4.40 ∗ 10-6
R2
-1.74 ∗ 10-5
6.56 ∗ 10-5
2.79 ∗ 10-5
-3.05 ∗ 10-5
1.27 ∗ 10-5
-3.14 ∗ 10-5
5.89 ∗ 10-5
2.07 ∗ 10-5
-5.83 ∗ 10-5
5.27 ∗ 10-5
R3
-15.05 ∗ 10-5
3.48 ∗ 10-5
-1.50 ∗ 10-5
-16.97 ∗ 10-5
-3.32 ∗ 10-5
-2.92 ∗ 10-5
-15.35 ∗ 10-5
-4.33 ∗ 10-5
-9.39 ∗ 10-5
6.90 ∗ 10-6
R4
1.19 ∗ 10-5
-3.50 ∗ 10-5
-3.00 ∗ 10-5
-4.93 ∗ 10-5
-1.52 ∗ 10-5
-8.90 ∗ 10-5
2.04 ∗ 10-5
1.77 ∗ 10-5
1.05 ∗ 10-5
1.00 ∗ 10-6
R5
-4.90 ∗ 10-6
8.00 ∗ 10-7
-1.72 ∗ 10-5
1.30 ∗ 10-6
-5.84 ∗ 10-5
-4.00 ∗ 10-7
-6.60 ∗ 10-6
5.00 ∗ 10-6
-8.80 ∗ 10-6
-6.00 ∗ 10-7
R6
-4.91 ∗ 10-6
-4.65 ∗ 10-6
0.90 ∗ 10-6
-1.90 ∗ 10-6
6.90 ∗ 10-6
-0.90 ∗ 10-5
-5.50 ∗ 10-6
-7.97 ∗ 10-6
-2.90 ∗ 10-4
-4.61 ∗ 10-6
Stimulus
FIGURE 7 (a) Six hypothetical fNIRS signals (b) An illustrative multivariate matrix (MVM) of brain activity from 6 different regions (R) for a total of
10 trials. The rows are trials, whereas the columns are the different regions from where brain activity is measured. (c) An illustrative plot to exemplify
how conceptual labels (CoLs) can be used to represent brain activity. Brain regions can be expressed with the CoLs Inactive, Active and Very
Active with approximate degree of membership values, μ. The derived ambiguity in the degree of membership ensures that uncertainty in the
numeric data (or neuroimaging reading from fNIRS) is well retained upon transformation into a conceptual label (CoL). (d) The corresponding
CoL MVM. Here, the colors denote the level of activity in different brain regions in response to the presented stimuli. For example, violet representing
high activity, cyan low activity, and white no activity. (e) The CoL MVM is given as an input to xMVPA which computes the classification
accuracy of a given set of patterns on the given data by computing with words (CWW) using interval type-2 fuzzy logic system (IT2-FLS). The optimal
set of patterns is found using an evolutionary algorithm (a genetic algorithm is used in the study). (f) A model for visual and auditory processing
in six-month-old infants, based on the six patterns (P1-P6) revealed by the xMVPA inference mechanism [65]. The color of the channel's
(Ch) octagon is based on its activity level: Inactive (white), Active (cyan), and Very Active (violet).
FEBRUARY 2022 | IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE 29
Degree of Membership, µ
IEEE Computational Intelligence Magazine - February 2022

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - February 2022
