Some types of clustering with relational constraints

For the clustering problem with general (not necessarily symmetric) relational constraints, different sets of feasible clusterings, also called clustering types, determined by the same relation, can be defined. In this paper some clustering types are discussed and adaptations of the hierarchical clustering method compatible with these clustering types are proposed.     

Differential relational encoding of categorical information in memory for action events

Memory for action phrases is better if the actions are enacted in subject-performed tasks (SPTs) than if they are only listened to in verbal tasks (VTs). This effect is ascribed to better item-specific encoding of SPTs than of VTs. The role of interitem relational information is controversial, and the findings of clustering with categorically structured lists are inconsistent (see Engelkamp, 1998). The present study contributes to clarifying these effects by demonstrating that intentional relational encoding can be used more efficiently in VTs than in SPTs and influences the degree of clustering. If the list structure is not obvious, inducing intentional encoding by presenting the category labels prior to list presentation and asking subjects to use this preinformation increases clustering in VTs but not in SPTs. Without preinformation, clustering scores of VTs and SPTs did not differ, with preinformation, clustering of VTs was stronger than that of SPTs. The authors suggest how the inconsistent findings with regard to clustering effects can be explained.     

Memory organization of action events and its relationship to memory performance

Previous research yielded inconsistent results regarding the memory organization of self-performed actions. The authors propose that task performance changes the very basis of memory organization. Enactment during study and test (Experiment 1) yielded stronger enactive clustering (based on motor-movement similarities), whereas verbal encoding yielded stronger conceptual clustering (based on semantic-episodic similarities). Enactment enhanced memory quantity and memory accuracy. Both measures increased with enactive clustering under self-performance instructions but with conceptual clustering under verbal instructions. Enactment only during study (Experiment 2) or only during testing (Experiment 3) also enhanced enactive clustering. It is proposed that different conditions affect the relative salience of different types of memory organization and their relative contribution to recall.     

Cluster Analytic Detection of Biological Sex: A Second Look

Golden and Meehl (1980) conducted a simulation with MMPI items to determine the accuracy of six popular clustering methods in detecting biological sex. They concluded that the results were "poor to mediocre" and that clustering methods have limited potential for contributing to knowledge. Their conclusion is inconsistent with the prior Monte Carlo literature on cluster analysis. A second look at the Golden-Meehl results reveals satisfactory clustering performance when they used the better MMPI item sets. Using Ward's clustering method and vocational interest items, the Golden-Meehl design was replicated with conditions providing a firmer inferential base. The replication results matched the favorable Golden-Meehl trials. In general, the results showed that Ward's method performed well when clustering was done with the more appropriate items.     

Passage organization and imposed learning strategies in comprehension and recall of connected discourse

Ninety-nine Ss were assigned randomly to learn a passage comprised of six paragraphs in which the statements were organized by concept name, or concept attribute, or in which the statements were scrambled. Each complete passage contained the same 36 statements. These treatments were orthogonally crossed with instructions to employ a name clustering strategy, an attribute clustering strategy, or a subjectively determined organizing strategy. Three learning trials were administered, each of which was followed by free recall. Passages organized by concept name were found to result in greater recall than passages organized by attributes. The name clustering strategy was more dominant than the attribute clustering strategy, lncongruence between passage organization and advocated clustering strategy resulted in greater recall than did congruency. Implications of these results for cognitive processing of information are discussed     

Analysis of meteorological conditions in Spain by means of clustering techniques

A comprehensive analysis of clustering techniques is presented in this paper through their application to data on meteorological conditions. Six partitional and hierarchical clustering techniques (k-means, k-medoids, SOM k-means, Agglomerative Hierarchical Clustering, and Clustering based on Gaussian Mixture Models) with different distance criteria, together with some clustering evaluation measures (Calinski–Harabasz, Davies–Bouldin, Gap and Silhouette criterion clustering evaluation object), present various analyses of the main climatic zones in Spain. Real-life data sets, recorded by AEMET (Spanish Meteorological Agency) at four of its weather stations, are analyzed in order to characterize the actual weather conditions at each location. The clustering techniques process the data on some of the main daily meteorological variables collected at these stations over six years between 2004 and 2010.     

A New Development to Aid Interpretation of Hierarchical Cluster Analysis of Repertory Grid Data

Hierarchical clustering is a common way to analyze repertory grid data, as it allows visual identification of homogeneous groups of elements and constructs. This article explores a bootstrap approach to clustering that additionally allows significance testing of dendrogram structures. Given that the choice of the specific clustering method and distance measure can be justified, the method yields indications of closely related element or construct structures. Closely integrated groups of constructs link to the concept of cognitive complexity. Several new complexity indexes are derived from the bootstrap solution. The behavior of the indexes is compared (a) across the most common clustering methods and distance measures and (b) to standard complexity measures. The derived measures reflect a different aspect of complexity than standard measures.     

Evaluating mixture modeling for clustering: recommendations and cautions

This article provides a large-scale investigation into several of the properties of mixture-model clustering techniques (also referred to as latent class cluster analysis, latent profile analysis, model-based clustering, probabilistic clustering, Bayesian classification, unsupervised learning, and finite mixture models; see Vermunt & Magdison, 2002). Focus is given to the multivariate normal distribution, and 9 separate decompositions (i.e., class structures) of the covariance matrix are investigated. To provide a link to the current literature, comparisons are made with K-means clustering in 3 detailed Monte Carlo studies. The findings have implications for applied researchers in that mixture-model clustering techniques performed best when the covariance structure and number of clusters were known. However, as the information about the shape and number of clusters became unknown, degraded performance was observed for both K-means clustering and mixture-model clustering.     

The role of encoding strategies in the verbal memory performance in patients with schizophrenia

Background. Verbal learning and memory is often compromised in patients with schizophrenia who prefer encoding words in order of their presentation (serial clustering) rather than using semantic categories (semantic clustering). Method. One hundred and four in‐patients with schizophrenia were assessed twice with the California Verbal Learning Test. Results. Patients showed significantly less semantic than serial clustering at both assessment times. Usage of encoding strategies were not stable over time. An increase in semantic clustering improved recall and recognition performance. Conclusions. Patients with schizophrenia should be taught to use the more effective encoding strategy of semantic clustering in order to improve their memory performance.     

Differential Development of Learning Strategies on a Pictorial Verbal Learning Test (PVLT) in Primary-School Children

The aim of this study was to test the hypothesis that developmental differences exist in the use of learning strategies in primary school children. Serial and subjective clustering in a multitrial Pictorial Verbal Learning Test (PVLT) were compared in 79 children aged 6–12. Correlation analyses indicated that serial clustering yielded better performance when information was presented on the initial trials of the test. Subjective clustering was superior when information was presented repeatedly, i.e., after three or more trials. Analyses of variance indicated that subjective clustering was used more often in older children with repeated presentations. On the other hand, there was no increase in the use of serial clustering with age and with repeated presentations. The findings imply that training in the use of proper strategies could have benefit for children who use an inefficient learning strategy and/or have a learning problem. In addition, they point to the importance of the factor “age” in relation to the way information is presented to children.     

Qualitative analysis of verbal fluency output: review and comparison of several scoring methods

We compared three approaches to scoring qualitative aspects of verbal fluency performance in 88 healthy young adults. Phonemic and semantic fluency output was scored for word clustering and switching between clusters. Convergent validity analyses using other tests presumed to tap into strategy use (California Verbal Learning Test, Ruff Figural Fluency Test) support scoring of phonemic and semantic clusters on both fluency tasks. Task-discrepant clustering (e.g., semantic clustering on phonemic fluency) may index intentional strategy use on both fluency tasks, whereas task consistent clustering (e.g., phonemic clustering on phonemic fluency) appears strategic only on semantic fluency. Switching can be decomposed into subtypes that appear to reflect different cognitive processes on phonemic versus semantic fluency. Principal components analyses suggest that earlier scoring methods do not fully capture the "process" aspects of verbal fluency performance.     

Iterative inverse factor analysis—A rapid method for clustering persons

By interchanging persons and items, iterative inverse factor analysis provides a relatively inexpensive way of clustering persons according to their patterns of response to the items. In addition to permitting the clustering of large numbers of persons, the technique enables one to determine the bases for such clustering. The items of behavior used can be heterogeneous in content and form.     

Contribution of imitative suicide to the suicide rate in prisons

Suicide rates in prisons are high. Our aim was to investigate the contribution of imitative suicide to the prison suicide rate. We used Knox tests for space-time clustering in a case register of natural and self-inflicted deaths in prisons in England and Wales and model simulations to estimate the effect size. We found significant space-time clustering among 657 self-inflicted deaths in 90 prisons over 10 years but no space-time clustering among 430 deaths from natural causes in 87 prisons over this period. Model simulations with an imitation rate of 5.8% (CI 1%-11%) reproduced the observed space-time clustering.     

A discriminated similarity matrix construction based on sparse subspace clustering algorithm for hyperspectral imagery

The clustering of hyperspectral images is a challenging task because of the high dimensionality of the data. Sparse subspace clustering (SSC) algorithm is one of the popularly used clustering algorithm for high dimensionality data. However, SSC has not fully used the spectral and spatial information during similarity matrix construction based on single sparse representation coefficient for hyperspectral Imagery (HSI) clustering. In this paper, two novel similarity matrix construction methods named as Cosine-Euclidean similarity matrix (abbreviated as CE) and Cosine-Euclidean dynamic weighting similarity matrix (abbreviated as CEDW) are proposed for HSI clustering. They can combine the high spectral information and rich spatial information. Firstly, CE utilizes the cosine similarity of spectral information based on overall sparse representation vectors and classical Euclidean distance of spatial information to construct a novel similarity matrix. Secondly, inheriting CE merits, dynamic weighting adjustment method is introduced to CEDW for some external influence factors to the HSI information. Several experiments on HSI demonstrated that the proposed algorithms are effective for HSI clustering.     

The effects of individual differences in test anxiety and type of orienting task on levels of organization in free recall

In Experiment 1, high and low test-anxious subjects recalled a list composed of words that could be organized either by taxonomic categories or first letters. High-anxiety subjects showed less recall and conceptual clustering. There was no anxiety difference in alphabetic clustering, but low-anxiety subjects used alphabetic clusters to bridge transitions between conceptual categories more often than did high-anxiety subjects. An initial orienting task which required semantic processing reduced the recall deficit for high-anxiety subjects but not the clustering deficit. Low-anxiety subjects seemed minimally affected by a nonsemantic-orienting task. Experiment 2 used a list that could be organized by associates or rhymes. High-anxiety subjects recalled less and exhibited less clustering of both types. The nonsemantic-orienting task depressed performance for both anxiety levels, though the effect was somewhat greater for high-anxiety subjects. Overall, these results provide little evidence to support the hypothesis that high-anxiety subjects process differently only for semantic content, but methodological limitations preclude unequivocally rejecting that view.     

Network structure influences speech production

Network science provides a new way to look at old questions in cognitive science by examining the structure of a complex system, and how that structure might influence processing. In the context of psycholinguistics, clustering coefficient-a common measure in network science-refers to the extent to which phonological neighbors of a target word are also neighbors of each other. The influence of the clustering coefficient on spoken word production was examined in a corpus of speech errors and a picture-naming task. Speech errors tended to occur in words with many interconnected neighbors (i.e., higher clustering coefficient). Also, pictures representing words with many interconnected neighbors (i.e., high clustering coefficient) were named more slowly than pictures representing words with few interconnected neighbors (i.e., low clustering coefficient). These findings suggest that the structure of the lexicon influences the process of lexical access during spoken word production.     

On the Complexity of Additive Clustering Models

Additive clustering provides a conceptually simple and potentially powerful approach to modeling the similarity relationships between stimuli. The ability of additive clustering models to accommodate similarity data, however, typically arises through the incorporation of large numbers of parameterized clusters. Accordingly, for the purposes of both model generation and model comparison, it is necessary to develop quantitative evaluative measures of additive clustering models that take into account both data-fit and complexity. Using a previously developed probabilistic formulation of additive clustering, the Bayesian Information Criterion is proposed for this role, and its application demonstrated. Limitations inherent in this approach, including the assumption that model complexity is equivalent to cluster cardinality, are discussed. These limitations are addressed by applying the Laplacian approximation of a marginal probability density, from which a measure of cluster structure complexity is derived. Using this measure, a preliminary investigation is made of the various properties of cluster structures that affect additive clustering model complexity. Among other things, these investigations show that, for a fixed number of clusters, a model with a strictly nested cluster structure is the least complicated, while a model with a partitioning cluster structure is the most complicated. Copyright 2001 Academic Press.     

Aging and semantic cueing during learning and retention of verbal episodic information

The purpose of this study was to determine the effectiveness of semantic cues provided at encoding and during retention for older adults' memory. For the California Verbal Learning Test-II, participants received semantic or nonsemantic cues that were varied across groups at encoding and during the retention interval. Provision of a semantic cue at encoding led to greater semantic clustering at learning, but not increased recall performance. Providing a semantic cue during the retention interval led to better delayed free recall and greater semantic clustering. No group differences in recall or semantic clustering were found at delayed cued recall. The current findings suggest that semantic cues can be beneficial for recalling unstructured information when administered during the retention interval.     

Verbal Learning Strategies of Adolescents and Adults with the Syndrome of Nonverbal Learning Disabilities

Abstract

Although many psychometric studies of individuals with the syndrome of nonverbal learning disabilities (NLD) have been conducted, one relatively neglected area has been the study of their performance on explicit verbal memory measures. We examined the performance of adolescents and adults with NLD on the California Verbal Learning Test, a measure allowing analysis of self-initiated learning strategies, and compared their performance to age- and Full Scale IQ-matched verbal learning-disabled (VLD) controls. Mean performance of the NLD sample on the semantic clustering index fell one standard deviation below the normative mean, whereas their serial clustering score was within normal limits. Additionally, the serial clustering score for our NLD sample was significantly greater than their semantic clustering score, suggesting that these individuals are more likely to spontaneously employ serial verbal learning strategies as opposed to those that are semantically driven. This difference in serial versus semantic clustering scores was not seen in our VLD controls, who performed equally well, and within normal limits, on both indices.     

Genes and family environment in familial clustering of cancer

Familial clustering of a disease is defined as the occurrence of the disease within some families in excess of what would be expected from the occurrence in the population. It has been demonstrated for several cancer types, ranging from rare cancers as the adenomatosis-coli-associated colon cancer or the Li-Fraumeni syndrome to more common cancers as breast cancer and colon cancer. Familial clustering, however, is merely an epidemiological pattern, and it does not tell whether genetic or environmental causes or both in combination are responsible for the familial clustering. Familial clustering may be due to genetic predisposition to the disease, but exposure to environmental factors — shared by members of some families, but not by members of other families — may also cause familial clustering and hence mimic genetic inheritance in the study of nuclear families. Based on assumptions regarding the individual steps in the biological process starting with exposure to carcinogens and ending with death from disseminated cancer we suggest that genetic and environmental factors may both be involved in most of these steps. The present paper focuses on research methodologies necessary to discriminate between the effect of genes and family environment in the development of cancer.