Classification is an integral part of any pattern recognition system. Depending on
whether a set of labeled training samples is available or not, classification can be
either supervised or unsupervised. Clustering is an important unsupervised classification
technique where a number of data points are grouped into clusters such that
points belonging to the same cluster are similar in some sense and points belonging
to different clusters are dissimilar in the same sense. Cluster analysis is a complex
problem as a variety of similarity and/or dissimilarity measures exist in the literature
without any universal definition. In a crisp clustering technique, each pattern is
assigned to exactly one cluster, whereas in the case of fuzzy clustering, each pattern
is given a membership degree to each class. Fuzzy clustering is inherently more
suitable for handling imprecise and noisy data with overlapping clusters.
For partitioning a data set, one has to define a measure of similarity or proximity
based on which cluster assignments are done. The measure of similarity is usually
data dependent. It may be noted that, in general, one of the fundamental features of
shapes and objects is symmetry, which is considered to be important for enhancing
their recognition. Examples of symmetry abound in real life, such as the human
face, human body, flowers, leaves, and jellyfish. As symmetry is so common, it
may be interesting to exploit this property while clustering a data set. Based on this
observation, in recent years a large number of symmetry-based similarity measures
have been proposed. This book is focused on different issues related to clustering,
with particular emphasis on symmetry-based and metaheuristic approaches.
The aim of a clustering technique is to find a suitable grouping of the input data
set so that some criteria are optimized. Hence, the problem of clustering can be
posed as an optimization problem. The objective to be optimized may represent
different characteristics of the clusters, such as compactness, symmetrical compactness,
separation between clusters, connectivity within a cluster, etc. A straightforward
way to pose clustering as an optimization problem is to optimize some cluster
validity index that reflects the goodness of the clustering solutions. All possible values
of the chosen optimization criterion (validity index) define the complete search
space. Traditional partitional clustering techniques, such as K-means and fuzzy Cmeans,
employ a greedy search technique over the search space in order to optimize
Classification is an integral part of any pattern recognition system. Depending onwhether a set of labeled training samples is available or not, classification can beeither supervised or unsupervised. Clustering is an important unsupervised classificationtechnique where a number of data points are grouped into clusters such thatpoints belonging to the same cluster are similar in some sense and points belongingto different clusters are dissimilar in the same sense. Cluster analysis is a complexproblem as a variety of similarity and/or dissimilarity measures exist in the literaturewithout any universal definition. In a crisp clustering technique, each pattern isassigned to exactly one cluster, whereas in the case of fuzzy clustering, each patternis given a membership degree to each class. Fuzzy clustering is inherently moresuitable for handling imprecise and noisy data with overlapping clusters.For partitioning a data set, one has to define a measure of similarity or proximitybased on which cluster assignments are done. The measure of similarity is usuallydata dependent. It may be noted that, in general, one of the fundamental features ofshapes and objects is symmetry, which is considered to be important for enhancingtheir recognition. Examples of symmetry abound in real life, such as the humanface, human body, flowers, leaves, and jellyfish. As symmetry is so common, itmay be interesting to exploit this property while clustering a data set. Based on thisobservation, in recent years a large number of symmetry-based similarity measureshave been proposed. This book is focused on different issues related to clustering,with particular emphasis on symmetry-based and metaheuristic approaches.The aim of a clustering technique is to find a suitable grouping of the input dataset so that some criteria are optimized. Hence, the problem of clustering can beposed as an optimization problem. The objective to be optimized may representdifferent characteristics of the clusters, such as compactness, symmetrical compactness,separation between clusters, connectivity within a cluster, etc. A straightforwardway to pose clustering as an optimization problem is to optimize some clustervalidity index that reflects the goodness of the clustering solutions. All possible valuesof the chosen optimization criterion (validity index) define the complete searchspace. Traditional partitional clustering techniques, such as K-means and fuzzy Cmeans,employ a greedy search technique over the search space in order to optimize
การแปล กรุณารอสักครู่..
Classification is an integral part of any pattern recognition system. Depending on
whether a set of labeled training samples is available or not, classification can be
either supervised or unsupervised. Clustering is an important unsupervised classification
technique where a number of data points are grouped into clusters such that
points belonging to the same cluster are similar in some sense and points belonging
to different clusters are dissimilar in the same sense. Cluster analysis is a complex
problem as a variety of similarity and/or dissimilarity measures exist in the literature
without any universal definition. In a crisp clustering technique, each pattern is
assigned to exactly one cluster, whereas in the case of fuzzy clustering, each pattern
is given a membership degree to each class. Fuzzy clustering is inherently more
suitable for handling imprecise and noisy data with overlapping clusters.
For partitioning a data set, one has to define a measure of similarity or proximity
based on which cluster assignments are done. The measure of similarity is usually
data dependent. It may be noted that, in general, one of the fundamental features of
shapes and objects is symmetry, which is considered to be important for enhancing
their recognition. Examples of symmetry abound in real life, such as the human
face, human body, flowers, leaves, and jellyfish. As symmetry is so common, it
may be interesting to exploit this property while clustering a data set. Based on this
observation, in recent years a large number of symmetry-based similarity measures
have been proposed. This book is focused on different issues related to clustering,
with particular emphasis on symmetry-based and metaheuristic approaches.
The aim of a clustering technique is to find a suitable grouping of the input data
set so that some criteria are optimized. Hence, the problem of clustering can be
posed as an optimization problem. The objective to be optimized may represent
different characteristics of the clusters, such as compactness, symmetrical compactness,
separation between clusters, connectivity within a cluster, etc. A straightforward
way to pose clustering as an optimization problem is to optimize some cluster
validity index that reflects the goodness of the clustering solutions. All possible values
of the chosen optimization criterion (validity index) define the complete search
space. Traditional partitional clustering techniques, such as K-means and fuzzy Cmeans,
employ a greedy search technique over the search space in order to optimize
การแปล กรุณารอสักครู่..