IEEE Systems, Man and Cybernetics Magazine - April 2020 - 38

Analysis Tools
NERFcM,
CLODD,
and CCV
Microarray
Gene Expression
Data

Dice's
Similarity
Coefficient

Linear Order
Statistic
Operator

Square
Dissimilarity
Matrix
Fuzzy
Aggregation
Operator

TranscriptionRelated Arabidopsis
Genes

GO Annotations

Augmented
Fuzzy Measure
Similarity

Square
Dissimilarity
Matrix

Figure 17. A fuzzy cluster analysis framework for GO data proposed by Havens et al. [119].

followed by estimation of the number of clusters applying
CCV to the sampled data.
To cluster real-life benchmark gene-expression data, a
novel interactive genetic-algorithm-based, multiobjective
approach was proposed by Mukhopadhyay et al. [122] that
simultaneously finds the clustering solution and evolves
the set of validity measures that are to be optimized. The
proposed method interactively takes the input from a
human decision maker based on the VAT-based visualization tool and adaptively learns from that input to obtain
the final set of validity measures along with the final clustering result.
Monitoring System for Older Individuals
With the significant increase in the population of older
individuals in developed countries and the limited number
of care centers, the concept of aging in place (AIP) has
gained significant attention. AIP revolves around the
notion of independent or partially assisted living and the
ability to continuously receive any necessary support for
growing needs. For successful implementation of AIP projects, personalizing the care of older individuals through
environmental monitoring is essential.
Sledge et al. [123] established a framework for recognizing temporal trends in feature data extracted from passive
sensors (e.g., infrared motion and pneumatic bed sensors;
bed restlessness, pulse, and respiration sensors) used to
monitor individuals. The GNGC algorithm was used for
temporal clustering to assign different activity names to
different sensor measurement profiles, and the VAT image
was used to visualize the changing cluster structure of the
temporal data stream.
As a part of passive fall-risk-assessment research in
home environments, Banerjee et al. [124] presented a
method to identify older residents at risk by using features extracted from their gait information from a singledepth camera. The VAT algorithm was used to determine
38

IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE Apri l 2020

the number of clusters, which was then used as an input
to the PCM clustering technique [125]. The analysis helps
detect changes in gait patterns, which can be used to
analyze fall risk for older residents by passively observing them in their home environments. Li et al. [126]
proposed an acoustic fall-detection system called acoustic-FADE that employs an eight-microphone circular
array to automatically detect falls. The iVAT algorithm
was used to analyze the relationship between fall and
nonfall acoustic signatures in conjunction with the nearest-neighbor-based distance to find and remove the most
challenging false alarms based on an efficient featureselection technique.
Natural-Language Processing
Document and word clustering are well studied problems
in the natural-language-processing community. Most algorithms cluster documents and words separately but not
simultaneously. Dhillon [92] proposed a novel algorithm to
cluster documents and words simultaneously as a bipartite
graph-partitioning problem solved using spectral techniques. However, since most spectral clustering techniques require the number of clusters to seek as an input,
Liu and Lu [127] explored a VAT-based method for determining the number of clusters present in the given data set
for coclustering documents and words. It includes three
main steps. First, generate a VAT image of the input
matrix, which is produced by spectral coclustering documents and words. Next, use some common image-processing techniques, such as a grayscale morphological
operation to filter the VAT image. Finally, the cluster number is estimated by computing the eigengap of the grayscale matrix of the filtered image.
In the area of automated support for argument reconstruction from natural-language texts, Winkels et al. [128],
[129] investigated several possibilities to support a manual
process of extracting arguments, which is a nontrivial task

IEEE Systems, Man and Cybernetics Magazine - April 2020

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - April 2020