Unsupervised Community Detection Research Articles

K-plex-based community detection with graph neural networks

Community detection is an effective way to determine the structure and characteristics of a complex network. With the expansion of the network scale, traditional community detection approaches such as modularity-based optimization models face new challenges related to representing and learning the topological structure and node attributes from a large scale complex network. Moreover, in many practical applications, there is little knowledge about label information or the number of communities, which greatly limits the performance of existing supervised or semi-supervised community detection approaches. To solve these problems, in this paper, we propose a graph neural network-based unsupervised community detection approach, which first applies the k-plex to generate the community seeds, then uses a node sampling algorithm to reduce the network complexity, and finally constructs a graph neural network model to learn the relationships of the network nodes and assign the nodes to different communities. Extensive empirical studies on various scale networks demonstrate both the effectiveness and efficiency of the proposed approach. Our codes are available at https://github.com/lol12854/KPGN.

Self-supervised graph autoencoder with redundancy reduction for community detection

Community detection is a significant research topic in network science, which has been revisited with graph neural networks. As a powerful graph representation learning model, graph autoencoder (GAE) is commonly used for unsupervised community detection. However, most GAE-based methods ignore multi-scale features of encoding layers, which inherently provide useful information for community detection. Moreover, these methods fail to simultaneously improve the representation learning process and clustering performance through a unified objective function. To address these issues, we propose a self-supervised graph autoencoder model with redundancy reduction for community detection. Firstly, we introduce a multi-scale module based on GAE to obtain discriminative node representations from different encoding layers. In particular, a redundancy reduction strategy is employed to eliminate redundancy information in the latent embedding space. Then, a node clustering module is used to obtain initial community labels. To fully utilize the multi-scale features to further refine clustering performance, a self-supervised module is designed to utilize current clustering labels to supervise the node representation learning process, thus constructing an end-to-end model for community detection. Finally, we validate the effectiveness of the proposed method on real-world networks. Experimental results demonstrate that our method outperforms several state-of-the-art methods in community detection.

Uncertainty in GNN Learning Evaluations: A Comparison between Measures for Quantifying Randomness in GNN Community Detection.

(1) The enhanced capability of graph neural networks (GNNs) in unsupervised community detection of clustered nodes is attributed to their capacity to encode both the connectivity and feature information spaces of graphs. The identification of latent communities holds practical significance in various domains, from social networks to genomics. Current real-world performance benchmarks are perplexing due to the multitude of decisions influencing GNN evaluations for this task. (2) Three metrics are compared to assess the consistency of algorithm rankings in the presence of randomness. The consistency and quality of performance between the results under a hyperparameter optimisation with the default hyperparameters is evaluated. (3) The results compare hyperparameter optimisation with default hyperparameters, revealing a significant performance loss when neglecting hyperparameter investigation. A comparison of metrics indicates that ties in ranks can substantially alter the quantification of randomness. (4) Ensuring adherence to the same evaluation criteria may result in notable differences in the reported performance of methods for this task. The W randomness coefficient, based on the Wasserstein distance, is identified as providing the most robust assessment of randomness.

GCN-Based Encoder-Decoder Networks With Markov Random Field for Unsupervised Community Detection

GCN-Based Encoder-Decoder Networks With Markov Random Field for Unsupervised Community Detection

A community detection model using node embedding approach and graph convolutional network with clustering technique

The communities in a network have distinct characteristics and interrelationships. Community discovery methods based on node embedding and deep learning have surpassed spectral clustering and statistical inference as the preferred methods for handling high-dimensional network data. Community detection in graph networks using graph neural networks (GNNs) is a rapidly expanding field of study that has attracted much attention recently. Inspired by the graph convolutional network (GCN) approach, promising results in evaluating graph structure data, a different method for discovering communities using node embedding using Node2Vec via graph convolutional networks (CD2NE-GCN) model has been proposed. It has been evaluated against the most recent standard methods of graph attention network (GAT), GCN-based approach for Unsupervised Community Detection (GUCD), and Graph Adversarial Networks (GAN). In this paper, a technique for community detection that combines the ideas of graph neural network design with node embedding via message forwarding has been presented. The experimentation has made use of five benchmark datasets: Karate, Cora, CiteSeer, Facebook, and DBLP. In the first set of trials, the node2vec uses the skip-gram with negative sampling (SGNS) approach via biased random walks of the second order to produce node embeddings and feed them to the graph convolutional network. The weighted average matrix and scaled normalized matrices of low-dimensional vertices have been calculated. In the next set of experiments, the average weighted matrix was sent to the 3-layer GCN with the activation function for generating feature vectors and applied clustering technique for predicting communities. The CD2NE-GCN approach outperforms previous literature by detecting graph network communities with accuracy scores of 98.45%, 99.04%, 96.39%, 98.79%, and 97.67% for Karate, Cora, CiteSeer, Facebook, and DBLP datasets, respectively. Thus, graph convolutional network models with a node embedding approach enhance the accuracy.

Integrating heterogeneous structures and community semantics for unsupervised community detection in heterogeneous networks

Community detection aims to discover hidden communities or groups in complex networks and is essentially unsupervised clustering behavior. However, most of the existing unsupervised methods are designed for homogeneous networks; therefore, they cannot effectively handle heterogeneous structures and rich semantic information. Under such a situation, it is difficult to accurately detect communities in heterogeneous networks that better reflect the real world. Therefore, this work aims to design an unsupervised framework to fuse heterogeneous structure information and interpret the rich semantics of the network in the form of community semantics. Thus, a heterogeneous network community detection method, called HAESF, is introduced. It includes two modules: the Heterogeneous Auto-Encoder (HAE) and the Semantic Factorization (SF) modules. In more detail, the HAE module adopts a hierarchical attention scheme to represent and aggregate the heterogeneous structure of the network. And it proposes the concept of heterogeneous information combinatorial graphs for structural reconstruction to achieve unsupervised detection. Concerning the SF module, it focuses on learning the semantic information in the network from the community point of view. It uses nonnegative matrix factorization to decompose the network features for obtaining community semantics. Once both modules are implemented, the objective of restricting community segmentation based on these semantics is achieved. The constraint is based on community semantic homogeneity to correct inaccurate node delineation. Furthermore, to improve the algorithm efficiency, a unified framework is designed to optimize the HAE and SF modules jointly. Within this new framework, the SF loss is innovatively used as a judgmental loss for selective segmentation optimizations, helping to obtain more reliable community detection results. As for the results, extensive experiments are performed on three public datasets. The findings show that HAESF outperforms the other popular unsupervised methods, where the composite score of HAESF is 11.73% ahead of the next best, demonstrating the proposed method’s effectiveness.

Unsupervised Community Detection Algorithm with Stochastic Competitive Learning Incorporating Local Node Similarity

Community detection is an important task in the analysis of complex networks, which is significant for mining and analyzing the organization and function of networks. As an unsupervised learning algorithm based on the particle competition mechanism, stochastic competitive learning has been applied in the field of community detection in complex networks, but still has several limitations. In order to improve the stability and accuracy of stochastic competitive learning and solve the problem of community detection, we propose an unsupervised community detection algorithm LNSSCL (Local Node Similarity-Integrated Stochastic Competitive Learning). The algorithm calculates node degree as well as Salton similarity metrics to determine the starting position of particle walk; local node similarity is incorporated into the particle preferential walk rule; the particle is dynamically adjusted to control capability increments according to the control range; particles select the node with the strongest control capability within the node to be resurrected; and the LNSSCL algorithm introduces a node affiliation selection step to adjust the node community labels. Experimental comparisons with 12 representative community detection algorithms on real network datasets and synthetic networks show that the LNSSCL algorithm is overall better than other compared algorithms in terms of standardized mutual information (NMI) and modularity (Q). The improvement effect for the stochastic competition learning algorithm is evident, and it can effectively accomplish the community detection task in complex networks.

CommunityGCN: community detection using node classification with graph convolution network

PurposeA community demonstrates the unique qualities and relationships between its members that distinguish it from other communities within a network. Network analysis relies heavily on community detection. Despite the traditional spectral clustering and statistical inference methods, deep learning techniques for community detection have grown in popularity due to their ease of processing high-dimensional network data. Graph convolutional neural networks (GCNNs) have received much attention recently and have developed into a potential and ubiquitous method for directly detecting communities on graphs. Inspired by the promising results of graph convolutional networks (GCNs) in analyzing graph structure data, a novel community graph convolutional network (CommunityGCN) as a semi-supervised node classification model has been proposed and compared with recent baseline methods graph attention network (GAT), GCN-based technique for unsupervised community detection and Markov random fields combined with graph convolutional network (MRFasGCN).Design/methodology/approachThis work presents the method for identifying communities that combines the notion of node classification via message passing with the architecture of a semi-supervised graph neural network. Six benchmark datasets, namely, Cora, CiteSeer, ACM, Karate, IMDB and Facebook, have been used in the experimentation.FindingsIn the first set of experiments, the scaled normalized average matrix of all neighbor's features including the node itself was obtained, followed by obtaining the weighted average matrix of low-dimensional nodes. In the second set of experiments, the average weighted matrix was forwarded to the GCN with two layers and the activation function for predicting the node class was applied. The results demonstrate that node classification with GCN can improve the performance of identifying communities on graph datasets.Originality/valueThe experiment reveals that the CommunityGCN approach has given better results with accuracy, normalized mutual information, F1 and modularity scores of 91.26, 79.9, 92.58 and 70.5 per cent, respectively, for detecting communities in the graph network, which is much greater than the range of 55.7–87.07 per cent reported in previous literature. Thus, it has been concluded that the GCN with node classification models has improved the accuracy.

Functional integrity of visual coding following advanced photoreceptor degeneration

Photoreceptor degeneration sufficient to produce severe visual loss often spares the inner retina. This raises hope for vision restoration treatments using optogenetics or electrical stimulation, which generate a replacement light input signal in surviving neurons. The success of these approaches is dependent on the capacity of surviving circuits of the visual system to generate and propagate an appropriate visual code in the face of neuroanatomical remodeling. To determine whether retinally degenerate animals possess this capacity, we generated a transgenic mouse model expressing the optogenetic actuator ReaChR in ON bipolar cells (second-order neurons in the visual projection). After crossing this with the rd1 model of photoreceptor degeneration, we compared ReaChR-derived responses with photoreceptor-driven responses in wild-type (WT) mice at the level of retinal ganglion cells and the visual thalamus. The ReaChR-driven responses in rd1 animals showed low photosensitivity, but in other respects generated a visual code that was very similar to the WT. ReaChR rd1 responses had high trial-to-trial reproducibility and showed sensitivity normalization to code contrast across background intensities. At the single unit level, ReaChR-derived responses exhibited broadly similar variations in response polarity, contrast sensitivity, and temporal frequency tuning as the WT. Units from the WT and ReaChR rd1 mice clustered together when subjected to unsupervised community detection based on stimulus-response properties. Our data reveal an impressive ability for surviving circuitry to recreate a rich visual code following advanced retinal degeneration and are promising for regenerative medicine in the central nervous system.

Adolescent functional network connectivity prospectively predicts adult anxiety symptoms related to perceived COVID-19 economic adversity.

Stressful events, such as the COVID-19 pandemic, are major contributors to anxiety and depression, but only a subset of individuals develop psychopathology. In a population-based sample (N=174) with a high representation of marginalized individuals, this study examined adolescent functional network connectivity as a marker of susceptibility to anxiety and depression in the context of adverse experiences. Data-driven network-based subgroups were identified using an unsupervised community detection algorithm within functional neural connectivity. Neuroimaging data collected during emotion processing (age 15) were extracted from a priori regions of interest linked to anxiety and depression. Symptoms were self-reported at ages 15, 17, and 21 (during COVID-19). During COVID-19, participants reported on pandemic-related economic adversity. Differences across subgroup networks were first examined, then subgroup membership and subgroup-adversity interaction were tested to predict change in symptoms over time. Two subgroups were identified: Subgroup A, characterized by relatively greater neural network variation (i.e., heterogeneity) and density with more connections involving the amygdala, subgenual cingulate, and ventral striatum; and the more homogenous Subgroup B, with more connections involving the insula and dorsal anterior cingulate. Accounting for initial symptoms, subgroup A individuals had greater increases in symptoms across time (β =.138, p=.042), and this result remained after adjusting for additional covariates (β =.194, p=.023). Furthermore, there was a subgroup-adversity interaction: compared with Subgroup B, Subgroup A reported greater anxiety during the pandemic in response to reported economic adversity (β =.307, p=.006), and this remained after accounting for initial symptoms and many covariates (β =.237, p=.021). A subgrouping algorithm identified young adults who were susceptible to adversity using their personalized functional network profiles derived from a priori brain regions. These results highlight potential prospective neural signatures involving heterogeneous emotion networks that predict individuals at the greatest risk for anxiety when experiencing adverse events.

A Graph-Based Blocking Approach for Entity Matching Using Contrastively Learned Embeddings

Data integration is considered a crucial task in the entity matching process. In this process, redundant and cunning entries must be identified and eliminated to improve the data quality. To archive this, a comparison between all entities is performed. However, this has quadratic computational complexity. To avoid this, `blocking' limits comparisons to probable matches. This paper presents a k-nearest neighbor graph-based blocking approach utilizing state-of-the-art context-aware sentence embeddings from pre-trained transformers. Our approach maps each database tuple to a node and generates a graph where nodes are connected by edges if they are related. We then invoke unsupervised community detection techniques over this graph and treat blocking as a graph clustering problem. Our work is motivated by the scarcity of training data for entity matching in real-world scenarios and the limited scalability of blocking schemes in the presence of proliferating data. Additionally, we investigate the impact of contrastively trained embeddings on the above system and test its capabilities on four data sets exhibiting more than 6 million comparisons. We show that our block processing times on the target benchmarks vary owing to the efficient data structure of the k-nearest neighbor graph. Our results also show that our method achieves better performance in terms of F1 score when compared to current deep learning-based blocking solutions.

Unsupervised Community Detection Algorithm Based on Graph Convolution Network and Social Media

In view of the difficulty and low efficiency of most existing algorithms in detecting large-scale community networks, an unsupervised community detection algorithm based on graph convolution networks and social media is proposed. First, some positive and negative sample nodes are labeled according to the node similarity to complete the graph segmentation. Then, the improved graph convolution network model is used for training to obtain the local community where the given starting node is located. Finally, the local community is optimized by setting the threshold of membership degree, so as to further screen the nodes outside the community and obtain accurate community detection results. The experimental analysis of the proposed algorithm based on Flixster, Douban, and Yelp datasets shows that when the number of community divisions is 12, the modularity values on the three datasets are 0.59, 0.62, and 0.69, respectively, and the standard deviations of F1 are 0.044, 0.048, and 0.040, respectively. Overall, the proposed unsupervised community detection algorithm has better robustness.

Unsupervised community detection in attributed networks based on mutual information maximization

Community detection is of great significance for understanding network functions and behaviors. With the successful application of deep learning in network-based analyses, recent studies have turned to utilizing graph convolutional networks (GCNs) to this problem due to their capability in capturing network attributes. Nevertheless, most existing GCN-based community detection approaches are semi-supervised and local structure-aware, even though community detection is an unsupervised learning problem essentially. In this paper, we develop a novel GCN method for unsupervised community detection under the framework of mutual information (MI) maximization, called UCDMI. Specifically, a novel MI maximization mechanism is developed to capture more fine-grained information of the global network structure in an unsupervised manner. Moreover, a new aggregation function is proposed for GCN to distinguish the importance between different neighboring nodes, which enables our method to identify more high-quality node representations and improve the community detection performance. Our extensive experiments demonstrate the effectiveness of our proposed UCDMI compared with several state-of-the-art community detection methods.

Physiological Network From Anthropometric and Blood Test Biomarkers.

Currently, research in physiology focuses on molecular mechanisms underlying the functioning of living organisms. Reductionist strategies are used to decompose systems into their components and to measure changes of physiological variables between experimental conditions. However, how these isolated physiological variables translate into the emergence -and collapse- of biological functions of the organism as a whole is often a less tractable question. To generate a useful representation of physiology as a system, known and unknown interactions between heterogeneous physiological components must be taken into account. In this work we use a Complex Inference Networks approach to build physiological networks from biomarkers. We employ two unrelated databases to generate Spearman correlation matrices of 81 and 54 physiological variables, respectively, including endocrine, mechanic, biochemical, anthropometric, physiological, and cellular variables. From these correlation matrices we generated physiological networks by selecting a p-value threshold indicating statistically significant links. We compared the networks from both samples to show which features are robust and representative for physiology in health. We found that although network topology is sensitive to the p-value threshold, an optimal value may be defined by combining criteria of stability of topological features and network connectedness. Unsupervised community detection algorithms allowed to obtain functional clusters that correlate well with current medical knowledge. Finally, we describe the topology of the physiological networks, which lie between random and ordered structural features, and may reflect system robustness and adaptability. Modularity of physiological networks allows to explore functional clusters that are consistent even when considering different physiological variables. Altogether Complex Inference Networks from biomarkers provide an efficient implementation of a systems biology approach that is visually understandable and robust. We hypothesize that physiological networks allow to translate concepts such as homeostasis into quantifiable properties of biological systems useful for determination and quantification of health and disease.

Abstract IA05: Microenvironmental signatures reveal biologic and clinical subtypes of diffuse large B-cell lymphoma

Abstract Diffuse large B-cell lymphoma (DLBCL) is a biologically and clinically heterogenous disease almost invariably treated with combinatorial chemoimmunotherapy resulting in a ~65% cure rate in first line. Further improvement of treatment outcome relies on elucidating the biology that underlies the clinical behavior of this heterogenous disease. Transcriptomic and genetic characterization of DLBCL have increased our understanding of its pathogenesis and provided potential novel therapeutic targets. Over the past few years, the role of the microenvironment in DLBCL biology has been increasingly recognized. A guiding framework is necessary to effectively and safely translate research evidence into clinical practice. Towards this goal, we characterized the DLBCL microenvironment based on the transcriptional footprint of microenvironment cells and processes in a large cohort of patients. We first developed transcriptional signatures reflecting distinct cellular subtypes of the microenvironment, biologic processes, and canonical signaling pathway activation. We then used these signatures to virtually reconstruct the lymphoma microenvironment (LME) by applying an unsupervised community detection algorithm in 4,656 DLBCL cases. As a result, we described four classes of LMEs that reflect distinct enrichment of associated signatures. 1. “Germinal center-like” (GC-like) due to the presence of signatures from cell types commonly found in germinal centers including follicular dendritic cells, lymphatic endothelial cells, total T cells, and several CD4+ T-cell subpopulations. 2. “Mesenchymal” (MS) for the abundance of signatures from stromal cells such as cancer-associated fibroblasts, fibroblastic reticular cells, vascular endothelial cells, and extracellular matrix pathways. The MS-LME class was enriched in mutations affecting antigen presentation. These lymphomas also presented with high activity of the TGFB/SMAD and HIF pathways, previously associated with favorable prognosis in DLBCL. 3. “Inflammatory” (IN) for the presence of signatures from macrophages, neutrophils, NK cells, and cytotoxic T cells. The IN-LME class was enriched for immune suppressive/pro-lymphoma cytokines, expression of the immune checkpoint molecules PD-L1 and IDO1, and higher activity of the NF-kB, JAK/STAT, and TNF signaling pathways. 4. A “depleted” (DP) form that, contrasting with the other LMEs, was characterized by an overall lower presence of microenvironment-derived signatures. We analyzed the association of LME classes with overall survival (OS) and progression-free survival (PFS) in patient cohorts of, respectively, 2,646 and 2,189 DLBCLs treated with rituximab-based chemoimmunotherapy. Analysis of OS by LME indicated significant differences in prognosis from better to poor as follows: GC-like, MS (p =0.03, vs. GC-like), IN (p = 0.006, vs. MS), and DP (p = 0.008, vs. IN) LMEs, whereas GC-like and MS have similarly favorable PFS curves (p = 0.9) followed by IN (p < 0.01) and DP (p = 0.03, vs. IN), which presented with the poorest PFS. In multivariate Cox-proportional hazard analysis controlled for C.O.O. and I.P.I., the LME subtypes remained informative with GC-like and MS LMEs associated with better outcome. When segregated by C.O.O., the DP-LME retained the poorest OS and PFS in both C.O.O. subtypes. However, in ABC-DLBCL, the LME with the best PFS and OS was MS, and in GCB-DLBCL, the best LME was GC-like, suggesting that the biologic impact of the microenvironment may be different depending on the lymphoma subtype. DHL and high-grade lymphomas (HGL) modified their outcome when LME was considered. DHL/HGL with the favorable prognosis LMEs GC-like and MS had significantly better PFS and OS than DHL/HGL with the unfavorable prognosis LMEs IN and DP. DLBCLs with a DP-LME were characterized by a minimal presence of LME signatures and tumor cells exhibited higher clonality, were enriched for mutations affecting TP53 and cell proliferation genes, as well as presented higher levels of aberrant hypermethylation of immune-related and TGFB/SMAD1 genes. Preclinical murine models and patient data demonstrated that some of these features can be pharmacologically reversed with hypomethylating agents, indicating that cytosine methylation patterning of lymphoma cells may contribute to a depleted LME. In summary, we categorized the DLBCL LME into four major transcriptionally defined subtypes with distinct biologic properties and clinical behavior, which complement genetically defined subtypes of DLBCL in guiding the development of rational therapeutic approaches for these patients. Citation Format: Leandro Cerchietti. Microenvironmental signatures reveal biologic and clinical subtypes of diffuse large B-cell lymphoma [abstract]. In: Proceedings of the AACR Virtual Meeting: Advances in Malignant Lymphoma; 2020 Aug 17-19. Philadelphia (PA): AACR; Blood Cancer Discov 2020;1(3_Suppl):Abstract nr IA05.

Efficient and exact sampling of transition path ensembles on Markovian networks.

The problem of flickering trajectories in standard kinetic Monte Carlo (kMC) simulations prohibits sampling of the transition path ensembles (TPEs) on Markovian networks representing many slow dynamical processes of interest. In the present contribution, we overcome this problem using knowledge of the metastable macrostates, determined by an unsupervised community detection algorithm, to perform enhanced sampling kMC simulations. We implement two accelerated kMC methods to simulate the nonequilibrium stochastic dynamics on arbitrary Markovian networks, namely, weighted ensemble (WE) sampling and kinetic path sampling (kPS). WE-kMC utilizes resampling in pathway space to maintain an ensemble of representative trajectories covering the state space, and kPS utilizes graph transformation to simplify the description of an escape trajectory from a trapping energy basin. Both methods sample individual trajectories governed by the linear master equation with the correct statistical frequency. We demonstrate that they allow for efficient estimation of the time-dependent occupation probability distributions for the metastable macrostates, and of TPE statistics, such as committor functions and first passage time distributions. kPS is particularly attractive, since its efficiency is essentially independent of the degree of metastability, and we suggest how the algorithm could be coupled with other enhanced sampling methodologies. We illustrate our approach with results for a network representing the folding transition of a tryptophan zipper peptide, which exhibits a separation of characteristic timescales. We highlight some salient features of the dynamics, most notably, strong deviations from two-state behavior, and the existence of multiple competing mechanisms.

Active Learning in the Geometric Block Model

The geometric block model is a recently proposed generative model for random graphs that is able to capture the inherent geometric properties of many community detection problems, providing more accurate characterizations of practical community structures compared with the popular stochastic block model. Galhotra et al. recently proposed a motif-counting algorithm for unsupervised community detection in the geometric block model that is proved to be near-optimal. They also characterized the regimes of the model parameters for which the proposed algorithm can achieve exact recovery. In this work, we initiate the study of active learning in the geometric block model. That is, we are interested in the problem of exactly recovering the community structure of random graphs following the geometric block model under arbitrary model parameters, by possibly querying the labels of a limited number of chosen nodes. We propose two active learning algorithms that combine the use of motif-counting with two different label query policies. Our main contribution is to show that sampling the labels of a vanishingly small fraction of nodes (sub-linear in the total number of nodes) is sufficient to achieve exact recovery in the regimes under which the state-of-the-art unsupervised method fails. We validate the superior performance of our algorithms via numerical simulations on both real and synthetic datasets.

Using Node Identifiers and Community Prior for Graph-Based Classification

With widely available large-scale network data, one hot topic is how to adopt traditional classification algorithms to predict the most probable labels of nodes in a partially labeled network. In this article, we propose a new algorithm called identifier-based relational neighbor classifier (IDRN) to solve the within-network multi-label classification problem. We use the node identifiers in the egocentric networks as features and propose a within-network classification model by incorporating community structure information to predict the most probable classes for unlabeled nodes. We demonstrate the effectiveness of our approach on several publicly available datasets. First, taking a semi-supervised approach, IDRN without any community prior is applied in community detection experiments, and it outperforms most existing unsupervised community detection algorithms. After that, in large-scale graph-based multi-label classification tasks, our approaches perform well in both fully labeled and partially labeled networks in most cases. To evaluate the scalability of our algorithm, we also show a scalability test to evaluate the running time of our algorithm in different networks. The experiment results show that our approach is quite efficient and suitable for large-scale real-world classification tasks.

Alexithymia and personality traits of patients with inflammatory bowel disease

Psychological factors, specific lifestyles and environmental stressors may influence etiopathogenesis and evolution of chronic diseases. We investigate the association between Chronic Inflammatory Bowel Diseases (IBD) and psychological dimensions such as personality traits, defence mechanisms, and Alexithymia, i.e. deficits of emotional awareness with inability to give a name to emotional states. We analyzed a survey of 100 patients with IBD and a control group of 66 healthy individuals. The survey involved filling out clinical and anamnestic forms and administering five psychological tests. These were then analyzed by using a network representation of the system by considering it as a bipartite network in which elements of one set are the 166 individuals, while the elements of the other set are the outcome of the survey. We then run an unsupervised community detection algorithm providing a partition of the 166 participants into clusters. That allowed us to determine a statistically significant association between psychological factors and IBD. We find clusters of patients characterized by high neuroticism, alexithymia, impulsivity and severe physical conditions and being of female gender. We therefore hypothesize that in a population of alexithymic patients, females are inclined to develop psychosomatic diseases like IBD while males might eventually develop behavioral disorders.

Identification and Classification of Functional Modules in the Brain

Event Abstract Back to Event Identification and Classification of Functional Modules in the Brain Jane Wang1*, Luis Amaral2 and James Booth1 1 Northwestern University, United States 2 Northwestern University, HHMI, United States The complexity of the human brain is surpassed only by the large amounts of data being acquired through modern brain mapping techniques. This deluge of data necessitates novel tools of analysis which take into account the vast interdependencies and relationships within and between anatomical regions. Functional connectivity analysis visualizes dynamic patterns of activity acquired through functional magnetic resonance imaging (fMRI) as networks of temporal correlation and takes advantage of multivariate statistical methods to extract information not available with univariate linear modeling. Time series analysis involving coherence and other spectral methods have also been applied to characterize networks of interactions between brain regions (Muller 2001, Sun 2004). At the same time, the mathematical tools of complex network theory can be employed in analyzing these interactions to reveal either local or global topological characteristics thought to be indicative of modes of function or information processing. We examine changes in language processing over the course of development in children (age 9 to age 15) participating in phonological judgment (rhyming) tasks of visually presented words (Cao 2007, Bitan 2008). We are able to recover and characterize functional modules at multiple scales without the presupposition of previously defined anatomical regions using a combination of multivariate time series analysis techniques as well as graph theoretical measures such as shortest path length, clustering coefficient, and unsupervised community detection. Importantly, our results show that task performance can be predicted by network-level measures at both a global and a local (single link) scale. References Bitan, T., Cheon, J., Lu, D., Burman D. D., & Booth, J. R. (2008) Developmental increase in top-down and bottom-up processing in a phonological task: An effective connectivity fMRI study. Journal of Cognitive Neuroscience, 21 (6): 1135-1145. Cao, F., Bitan, T., & Booth, J. R. (2007) Effective brain connectivity in children with reading difficulties during phonological processing. Brain and Language, 107: 91-101. Muller, K., Lohmann, G., Bosch, V., & von Cramon, D. Y. (2001). On multivariate spectral analysis of fMRI time series. NeuroImage, 14: 347-356. Sun, F. T., Lee, M. M., & D’Esposito, M. (2004) Measuring interregional functional connectivity using coherence and partial coherence analyses of fMRI data. NeuroImage, 21: 647-658. Figure 1 Keywords: General neuroinformatics, Neuroimaging Conference: 4th INCF Congress of Neuroinformatics, Boston, United States, 4 Sep - 6 Sep, 2011. Presentation Type: Poster Presentation Topic: General neuroinformatics Citation: Wang J, Amaral L and Booth J (2011). Identification and Classification of Functional Modules in the Brain. Front. Neuroinform. Conference Abstract: 4th INCF Congress of Neuroinformatics. doi: 10.3389/conf.fninf.2011.08.00053 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 17 Oct 2011; Published Online: 19 Oct 2011. * Correspondence: Dr. Jane Wang, Northwestern University, Evanston, United States, janewang@northwestern.edu Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Jane Wang Luis Amaral James Booth Google Jane Wang Luis Amaral James Booth Google Scholar Jane Wang Luis Amaral James Booth PubMed Jane Wang Luis Amaral James Booth Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.