Partitional Clustering Research Articles

Cluster consensus of opinion dynamics with stubborn individuals

In this paper, we aim to study the cluster consensus of Friedkin–Johnsen (F-J) model with one or more stubborn nodes. Firstly, for F-J model with only one stubborn node, by algebra theory and graph theory, we not only establish consensus criterion for F-J model without oblivious agents but also establish some couple-cluster consensus criteria for F-J model with oblivious agents. Secondly, for F-J model with more than one stubborn individual, we present a novel network partition strategy to divide F-J model into many subnetworks, where every stubborn node and its pure-reachable nodes (a special reachable node without stubborn nodes except the starting node appearing in the walk) form a subnetwork, and all oblivious individuals of F-J model also constitute a subnetwork if they exist. By our network partition, some necessary and sufficient cluster consensus criteria are given. In addition, if there exist some oblivious individuals, our main results show that the influence of the initial opinion of stubborn root on the final opinion of non-oblivious individuals is proportional to the stubborn extent of the stubborn root to its initial opinion. Finally, some simulations are provided to illustrate the results.

Self-Supervised Object Localization with Joint Graph Partition

Object localization aims to generate a tight bounding box for the target object, which is a challenging problem that has been deeply studied in recent years. Since collecting bounding-box labels is time-consuming and laborious, many researchers focus on weakly supervised object localization (WSOL). As the recent appealing self-supervised learning technique shows its powerful function in visual tasks, in this paper, we take the early attempt to explore unsupervised object localization by self-supervision. Specifically, we adopt different geometric transformations to image and utilize their parameters as pseudo labels for self-supervised learning. Then, the class-agnostic activation map (CAAM) is used to highlight the target object potential regions. However, such attention maps merely focus on the most discriminative part of the objects, which will affect the quality of the predicted bounding box. Based on the motivation that the activation maps of different transformations of the same image should be equivariant, we further design a siamese network that encodes the paired images and propose a joint graph cluster partition mechanism in an unsupervised manner to enhance the object co-occurrent regions. To validate the effectiveness of the proposed method, extensive experiments are conducted on CUB-200-2011, Stanford Cars and FGVC-Aircraft datasets. Experimental results show that our method outperforms state-of-the-art methods using the same level of supervision, even outperforms some weakly-supervised methods.

A simple and effective iterated greedy algorithm for structural balance in signed networks

Recently, problems related to structural balance have received widespread attention in the research fields of signed networks, such as signed network partition, community detection and correlation clustering. Structural balance problems aim to find a partition of a signed network such that the frustration index, the sum of positive edges between subsets and negative edges within subsets, is minimized, which is known to be nondeterministic polynomial-time (NP)-complete and remains an open challenge. Many heuristic and meta-heuristic algorithms are developed to minimize the frustration index. However, most extant algorithms require quite a few parameters and are not efficient enough for large signed networks. In this study, we present a simple and effective iterated greedy (IG) algorithm for solving the structural balance problems with the objective of frustration minimization. In the algorithm, a constructive greedy heuristic is proposed to generate and reconstruct solutions to the problem with high efficiency. A two-stage local search procedure is designed to exploit the search space at different levels of granularity. An adaptive destruction method is developed to enhance the exploitation ability of the algorithm in the early stage and maintain the exploration ability later. Additionally, two acceleration methods are proposed to help the algorithm get a better performance in large signed networks. The experimental results indicate that the proposed IG algorithm outperforms other meta-heuristics in the literature especially in terms of the computational time.

Developing an integrated model for planning the delivery of construction materials to post-disaster reconstruction projects

AbstractConstruction material delivery to post-disaster reconstruction projects is challenging because of the resource and time limitations that follow a large-scale disaster. There is compelling evidence that inadequate planning jeopardises the success of a large number of post-disaster reconstruction projects. Thus, the current study proposes an integrated approach to facilitate the procurement planning of construction materials following a large-scale disaster. The proposed approach clustered the location of construction projects using a differential evolution (DE)-K-prototypes, a new partitional clustering algorithm based on DE and K-prototypes, method. Then, using a permanent matrix prioritises cluster points based on route reliability-affecting factors. The model’s objectives are to minimise the total travel time, maximise the reliability of the route, and minimise the total weighted undelivered materials to projects. In the case of distribution of material through land vehicles, the possibility of breakdowns in the vehicle is considered, allowing for the determination of vehicle breakdown under various scenarios and the minimisation of undelivered materials to projects. As a result of the uncertain character of the disaster, the demands of construction projects are fuzzy, and Jimenez’s method is used to handle it. Due to the complexity of the problem, two algorithms are proposed, a multi-objective evolutionary algorithm based on decomposition (MOEA/D) and a non-dominated sorting genetic algorithm-II (NSGA-II). The results confirm that the proposed MOEA/D has a higher accuracy while NSGA-II has a shorter computational time. By providing new theoretical perspectives on disaster recovery strategies in the construction sector, this study contributes to the growing body of knowledge about disaster recovery strategies in the sector. The findings of this study can be employed to develop an integrated planning system for the delivery of construction materials to post-disaster reconstruction projects in disaster-prone countries.

P1317: TIME SERIES CLUSTERING OF T CELL SUBSETS DISSECTS HETEROGENEITY IN IMMUNE RECONSTITUTION AND SURVIVAL AMONG RECIPIENTS OF MUD-HCT WITH ATG OR PTCY

Background: Proliferating alloreactive T cells have a central role in the induction of acute GVHD (aGVHD) making them a promising target in preventing alloreactivity. T cell depleting regimens such as anti-T-lymphocyte globulin (ATG) or post-transplant cyclophosphamide (PTCy) effectively eliminate these cells and reduce aGVHD after hematopoietic cell transplantation (HCT) with marginal differences in clinical outcome. Comparative immune reconstitution analyses could contribute to answer the question, which agent would be most appropriate for an individual patient. However, not only effects relating to the different T cell depleting regimens, i.e. ATG or PTCy may be relevant but the general heterogeneity of T cell reconstitution. Aims: Here, we aimed at dissecting this heterogeneity with an approach called time-series clustering to better understand the impact of both regimens in individual patients and to consequently identify distinct patient subsets which benefit the most from each protocol. Methods: We retrospectively compared immune reconstitution data of 339 recipients of matched-unrelated donor (MUD) HCT with either ATG (n=304) or PTCy (n=35) as T cell depletion via conventional analysis on the cohort level and for its individual heterogeneity via time-series clustering. This analysis leveraged the approach of dynamic time warping (DTW) to determine the distance measure later used for the partitional clustering of individual patient time-series data. The performance of this methods was evaluated by the silhouette coefficient (Sil) which indicates a separation of clusters with values between 0 (overlapping clusters) and +1 (best separation); A 10-fold resampling was used to assess model robustness. Cluster information were used for subsequent analysis of clinical outcomes. All analysis were performed using R packages R stats, dtwclust, survival, survminer and cmprsk. Results: Comparative analysis of cellular reconstitution revealed distinct prominent T cell population after each protocol. While patients receiving PTCy as GVHD prophylaxis presented with higher levels of regulatory T cells, ATG patients had higher levels of γδ T- or NK-T cells. Time-series clustering of T cell subpopulations that associate with GVHD successfully dissected each population’s heterogeneity. For ATG patients the clustering revealed two distinct clusters with its optimal model configuration showing a good and robust silhouette coefficient (7_1: Sil=0.524) and a balanced patient distribution (Fig. 1A). In this model, clustering of ATG patients was driven by αβ- and activated T cells as those revealed higher absolute counts and a greater difference in shape over time as compared to regulatory- and γδ T cells. Patients in ATG cluster 1, showing higher absolute counts of αβ- and activated T cells compared to cluster 2 were associated to a significantly higher 1-year OS (98% vs. 79%, p=0.0023) due to lower NRM (p=0.032) and relapse (p=0.01). The clustering of PTCy patients distinguished two clusters with lower silhouette coefficients had a strong impact of αβ- and activated T cells (Fig.1B). Comparing the overall survival probability of both ATG clusters with the PTCy cohort revealed a significant difference driven by the ATG clusters (Fig 1C). Image:Summary/Conclusion: Beyond a differential impact of ATG and PTCy on immune reconstitution our analysis identified phenotypes that reproducibly associated with impaired clinical outcomes within the same T cell depletion platform. This provides guidance for individually choosing the most appropriate agent in the MUD setting.

Performance Evaluation of Simple K-Mean and Parallel K-Mean Clustering Algorithms: Big Data Business Process Management Concept

Data is the most valuable asset in any firm. As time passes, the data expands at a breakneck speed. A major research issue is the extraction of meaningful information from a complex and huge data source. Clustering is one of the data extraction methods. The basic K-Mean and Parallel K-Mean partition clustering algorithms work by picking random starting centroids. The basic and K-Mean parallel clustering methods are investigated in this work using two different datasets with sizes of 10000 and 5000, respectively. The findings of the Simple K-Mean clustering algorithms alter throughout numerous runs or iterations, according to the study, and so iterations differ for each run or execution. In some circumstances, the clustering algorithms’ outcomes are always different, and the algorithms separate and identify unique properties of the K-Mean Simple clustering algorithm from the K-Mean Parallel clustering algorithm. Differentiating these features will improve cluster quality, lapsed time, and iterations. Experiments are designed to show that parallel algorithms considerably improve the Simple K-Mean techniques. The findings of the parallel techniques are also consistent; however, the Simple K-Mean algorithm’s results vary from run to run. Both the 10,000 and 5000 data item datasets are divided into ten subdatasets for ten different client systems. Clusters are generated in two iterations, i.e., the time it takes for all client systems to complete one iteration (mentioned in chapter number 4). In the first execution, Client No. 5 has the longest elapsed time (8 ms), whereas the longest elapsed time in the following iterations is 6 ms, for a total elapsed time of 12 ms for the K-Mean clustering technique. In addition, the Parallel algorithms reduce the number of executions and the time it takes to complete a task.

A Clustering Ensemble Framework with Integration of Data Characteristics and Structure Information: A Graph Neural Networks Approach

Clustering ensemble is a research hotspot of data mining that aggregates several base clustering results to generate a single output clustering with improved robustness and stability. However, the validity of the ensemble result is usually affected by unreliability in the generation and integration of base clusterings. In order to address this issue, we develop a clustering ensemble framework viewed from graph neural networks that generates an ensemble result by integrating data characteristics and structure information. In this framework, we extract structure information from base clustering results of the data set by using a coupling affinity measure After that, we combine structure information with data characteristics by using a graph neural network (GNN) to learn their joint embeddings in latent space. Then, we employ a Gaussian mixture model (GMM) to predict the final cluster assignment in the latent space. Finally, we construct the GNN and GMM as a unified optimization model to integrate the objectives of graph embedding and consensus clustering. Our framework can not only elegantly combine information in feature space and structure space, but can also achieve suitable representations for final cluster partitioning. Thus, it can produce an outstanding result. Experimental results on six synthetic benchmark data sets and six real world data sets show that the proposed framework yields a better performance compared to 12 reference algorithms that are developed based on either clustering ensemble architecture or a deep clustering strategy.

Efficient Optimization of Partition Scan Statistics via the Consecutive Partitions Property

We generalize the spatial and subset scan statistics from the single to the multiple subset case. The two main approaches to defining the log-likelihood ratio statistic in the single subset case – the population-based and expectation-based scan statistics – are considered, leading to risk partitioning and multiple cluster detection scan statistics, respectively. We show that, for distributions in a separable exponential family, the risk partitioning scan statistic can be expressed as a scaled f-divergence of the normalized count and baseline vectors, and the multiple cluster detection scan statistic as a sum of scaled Bregman divergences. In either case, however, maximization of the scan statistic by exhaustive search over all partitionings of the data requires exponential time. To make this optimization computationally feasible, we prove sufficient conditions under which the optimal partitioning is guaranteed to be consecutive. This Consecutive Partitions Property generalizes the linear-time subset scanning property from two partitions (the detected subset and the remaining data elements) to the multiple partition case. While the number of consecutive partitionings of n elements into t partitions scales as , making it computationally expensive for large t, we present a dynamic programming approach which identifies the optimal consecutive partitioning in time, thus allowing for the exact and efficient solution of large-scale risk partitioning and multiple cluster detection problems. Finally, we demonstrate the detection performance and practical utility of partition scan statistics using simulated and real-world data.

Regional Flood Frequency Analysis Using the FCM-ANFIS Algorithm: A Case Study in South-Eastern Australia

Regional flood frequency analysis (RFFA) is widely used to estimate design floods in ungauged catchments. Both linear and non-linear methods are adopted in RFFA. The development of the non-linear RFFA method Adaptive Neuro-fuzzy Inference System (ANFIS) using data from 181 gauged catchments in south-eastern Australia is presented in this study. Three different types of ANFIS models, Fuzzy C-mean (FCM), Subtractive Clustering (SC), and Grid Partitioning (GP) were adopted, and the results were compared with the Quantile Regression Technique (QRT). It was found that FCM performs better (with relative error (RE) values in the range of 38–60%) than the SC (RE of 44–69%) and GP (RE of 42–78%) models. The FCM performs better for smaller to medium ARIs (2 to 20 years) (ARI of five years having the best performance), and in New South Wales, over Victoria. In many aspects, the QRT and FCM models perform very similarly. These developed RFFA models can be used in south-eastern Australia to derive more accurate flood quantiles. The developed method can easily be adapted to other parts of Australia and other countries. The results of this study will assist in updating the Australian Rainfall Runoff (national guide)-recommended RFFA technique.

Proposed Modification of K-Means Clustering Algorithm with Distance Calculation Based on Correlation

Clustering is a technique in data mining that groups a set of data into groups (clusters) of similar data. In general, there are two methods of clustering, namely the hierarchical method and the partition method. One of the most commonly used partition clustering methods in clustering is K-Means. The use of K-means method has been widely used in various fields with various purposes. Many research has been carried out to improve the performance of the K-Means method, for example, by modifying the method of determining the initial centroid or determining the appropriate number of clusters. In this research, the modification of the K-Means algorithm was carried out in calculating the distance by considering the correlation value between attributes. Attributes that have a high correlation value are assumed to have similar characteristics so that they determine the location of data in a particular cluster. The steps of the proposed method are: calculating the correlation value between attributes, determining the cluster centroid, calculating the distance by considering the value of correlation, and determining the data into certain clusters. The first contribution of this research is to propose a new distance calculation technique in the K-Means algorithm by considering correlation and the second contribution is to apply the proposed algorithm to a specific dataset, namely Iris dataset. In this research, the performance calculation of the modified algorithm was also carried out. From the experimental results using the Iris dataset, the proposed modification of the K-Means algorithm has fewer iterations than the original K-Means method, so that it requires less processing time. The original K-Means method requires 8 iterations, while the proposed method requires only 6 iterations. The proposed method also produces a higher accuracy rate of 89.33% than the original K-Means method, which is 82.67%.

Driving pattern recognition using mobile telematics

Mobile Telematics is an application that enables it’s users to detect driving behavior through sensors that already exist on a common smart phone device. Sensors such as Gyroscopes, GPS, Radio-Positioning, Existing Cameras, Location APIs, etc will help detect and identify valuable pieces of information, this can then be used to create unsupervised learning models on the driver’s driving behavior and display feedback for the driver to identify and correct in order to minimize the risk of a potential accident. Current telematics models are not efficient and accurate enough to reduce the chances of road accidents, the sensors required to allow for users to access the needed data may also be expensive. Thus, we have found that our system is not only improved but also relatively inexpensive since we are using devices most of our user base already have -- Smartphones. Our framework consists of three main components -- Self Organizing Maps, Nine-Layers Deep Auto-Encoder, Partitive Clustering Algorithm. The self organizing maps simplifies complex pieces of data into consumable information, the deep auto encoder takes into consideration anomalies of said data and extracts its features, and the partitive clustering algorithms groups the data based on a set of driving behavior classifications.

Adaptive Clustering Algorithm for IIoT Based Mobile Opportunistic Networks

The clustering algorithms play a crucial role for energy saving solutions in mobile opportunistic networks. If the selection of cluster head is made appropriately, then the energy can be consumed optimally. The existing clustering algorithms do not consider the optimal selection of the cluster head resulting in low survival rates and high energy consumption rates in nodes. The adaptive clustering is required in Industrial Internet of Things (IIoT) based sophisticated networks where seamless connectivity is imperative for rapid communication. In order to meet this research gap, an adaptive clustering algorithm for mobile opportunistic networks is proposed within military bases that uses a heuristic algorithm (GA) for adaptive clustering. An analysis of the opportunity network at the connected military base is carried out and the mobile opportunity model is constructed using the adaptive clustering for the similar traffic. In a mobile machine network, the next hop node is determined by the node clustering principle. A LEACH clustering protocol enables communication between cluster heads and base stations based on single-hop and multihop cluster nodes. In order to perform adaptive clustering of mobile network nodes based on network partitioning and scheduling of clusters, genetic algorithms are used. The proposed approach can be applied to the IIoT systems in places where adaptive clustering is required to optimize energy consumption, to reduce latency rates, and to enhance the throughput of mobile networks. The experimental findings suggest that the proposed adaptive strategy is capable of optimizing energy consumption rates, reducing network latency, and boosting efficiency by increasing throughput in mobile opportunistic networks.

Cluster Partition-Based Zonal Voltage Control for Distribution Networks With High Penetrated PVs

With the high penetrated photovoltaics (PVs) accessing distribution networks (DNs) in the future, the overvoltage of distribution network will be more serious, and the voltage control will be more complex. To solve this problem, a cluster partition-based zonal voltage control method for DN is proposed in this study. First, a cluster partition method using a comprehensive performance index is proposed: in terms of DN structure, a global density quality function index is introduced to measure the coupling degree of nodes in clusters. In terms of cluster function, the inconsistency coefficient index is presented to reflect the PV output characteristics, and a node membership function index is presented to limit the scale of the cluster. Then, the DN cluster partition is carried out under the fast Newman (FN) algorithm. Second, a second-order cone programming (SOCP) model of voltage control is established to maximize the PV consumption in each cluster. To achieve the coordinated optimization of clusters, an iterative optimization strategy among clusters is proposed. Finally, the effectiveness and rationality of the proposed method are verified in a 10 kV actual feeder system in Zhejiang Province, China.

Khaya revisited: Genetic markers and morphological analysis reveal six species in the widespread taxon K. anthotheca

AbstractKhaya (Meliaceae) is a widespread genus of trees distributed from the tropical forests of Africa to Madagascar and the Comoros. Khaya species are very close morphologically and some contain considerable phenotypic and ecological diversity, raising the question of their delimitation: this is the case of Khaya anthotheca s.l. (sensu lato, including K. nyasica), which is often lumped with K. grandifoliola and K. ivorensis. We tested species delimitation within K. anthotheca by combining population‐genetic and morphological data. First, a visual inspection of over a thousand specimens suggested the existence of five morphogroups. Second, after genotyping 498 samples at 71 nuclear SNPs, Bayesian clustering analyses (STRUCTURE) first identified five genetic clusters (K1–K5), while further analyses subdivided K5 into two clusters. The six genetic clusters obtained are distributed in parapatry or allopatry from West Africa to East Africa, except for cluster K4, which is in sympatry with K3 in Central Africa. The partitioning of genetic clusters is also largely congruent with the morphogroups identified. The morphological differentiation of these clusters was confirmed by statistical analyses of 27 leaf characters. The clear‐cut genetic differentiation between clusters, with few admixed genotypes even in contact zones, argues for the recognition of distinct species following the biological species concept. Our analyses highlight that the current taxonomic treatment of Khaya underestimates by one‐half the total number of species. We identify two new species, circumscribe K. anthotheca and support the rehabilitation at the rank of species of three taxa considered to be synonyms of K. anthotheca (K. agboensis, K. euryphylla, K. nyasica). These results illustrate the strength of combining population‐genetic and morphological data to delineate an African rainforest tree species complex.

Partition harvesting of a column-comb litchi harvester based on 3D clustering

Accurate recognition of litchi fruits in orchard environments and acquisition of their coordinate position information are key for realizing successful harvesting using litchi harvesters. However, the existing detection methods are often aimed at large and relatively sparse fruits and thus are inappropriate for small and densely distributed litchi fruits. Therefore, at present, litchi fruit are typically manually harvested, resulting in a low harvesting efficiency that cannot meet the needs of growers. To improve the efficiency of litchi harvesting, this study proposes a column-comb litchi harvesting method based on K-means 3D clustering partitioning, which includes four main steps: (1) Litchi image acquisition and labeling methods are developed. (2) An improved version of the present YOLOv3-tiny network model structure is developed named the YOLOv3-tiny-Litchi network model, and the litchi fruit detection results of five kinds of neural networks, namely, YOLOv3-tiny, YOLOv3-tiny-Litchi, YOLOv4, YOLOv5x and Faster R-CNN, are compared. (3) A depth camera is used to obtain the 3D coordinates of litchi fruits, and the K-means clustering algorithm is used to divide the litchi harvesting area to obtain the optimal partitioning results. (4) Field experiments on litchi harvesting are reported. The experimental results show that the improved YOLOv3-tiny-Litchi model can recognize litchi fruits more accurately; the recall rate is 78.99%, the precision rate is 87.43%, and the F1 score is 0.83. The results of 3D clustering partitioning show that when K is equal to 6, the optimal harvesting rate is 90.03%, which satisfies the theoretical requirements of litchi harvesters. The field experiments show that when the theoretical number of partitions is 6, the average harvesting rate of the litchi harvester is 91.15% and the recognition rate is 88.39%, and the harvesting efficiency of mechanical partition harvesting is 1.4 times that of mechanical harvesting without partitioning and 2 times that of manual harvesting, essentially meeting the operating requirements. These research results can serve as a reference for the mechanization of litchi harvesting.

A Study of Total-Factor Energy Efficiency for Regional Sustainable Development in China: An Application of Bootstrapped DEA and Clustering Approach

Total-factor energy efficiency (TFEE) is widely used to measure energy efficiency under the data envelopment analysis (DEA) framework, but the efficiencies obtained from DEA are structurally biased upward, and thus TFEE tends to overestimate energy efficiency. This research thus applies the bootstrapped DEA approach to correct the bias of TFEE. Using a dataset consisting of 30 provinces of China in the period 2016–2019, the bootstrapped-based test supports technology with variable returns to scale. The biased-corrected TFEE also indicates that energy consumption on average can be scaled down by 42.36%, rather than the biased value of 19.4%. The bootstrapped clustering partitions provinces into three groups: Cluster 1, with Guizhou as the representative medoid, includes half of the superior coastal provinces in terms of actual energy consumption and TFEE and half of the competitive inland provinces, whereas Cluster 3 outperforms Cluster 2 in terms of TFEE, but the actual energy consumption is higher, with Shandong and Hebei as the representative medoids, respectively. Lastly, empirical results imply that the northeast and central regions need more government attention and resources to practice sustainable development and improve TFEE.

Seafood freshness: e-nose data for classification purposes

The aim of this study was to develop an electronic nose system useful for the inspection of different seafood products and specifically designed to be applied as an auxiliary device for the management of seafood in the large-scale distribution chain. Since the freshness is a critical issue in determining the quality of fish it is of great importance to have a simple and rapid analytical method representing a useful tool for assessing the acceptability of the seafood products in the distribution centers or in the supermarkets. In this work a portable e-nose, composed by four selected metal oxide semiconductor sensors, a photoionization detector and two electrochemical cells, was applied to test three different seafood products - sole (Solea senegalensis) fillets, red mullet (Mullus barbatus) fillets and cuttlefish (Seppia officinalis) - during their shelf life from the day of arrival and packaging at the distribution center until beyond the expiration date. The K-means partitional clustering method was applied to group samples into three freshness classes, then confirmed by microbial analysis; two classification models based on k-nearest neighbors (K-NN) and partial least square-discriminant analysis were developed to classify the freshness of the seafood regardless the species and in particular, the KNN model provided 100% overall sensitivity, specificity and precision in prediction thus confirming the applicability of the system to inform retailers on product safety and quality.

A MULTI-OBJECTIVE GWO BASED CLUSTERING APPROACH FOR EFFICIENT DOCUMENT MINING

Learning and data mining systems often use clustering. Such applications require efficient and fast clustering algorithms. However, clustering a dataset is often a difficult task, especially for complex datasets such as text classification. Clustering is a strategy of grouping data points based on their similarities. Data clustering can be done using any clustering algorithm. However, conventional partitioning clustering algorithms greatly depend on initial points and can drift towards inefficient clustering on improper initialization. Furthermore, these algorithms are less flexible for achieving multiple objectives required for clustering the complex overlapping classes. Therefore, in this work, we proposed a Gray Wolf Optimization (GWO) based multiobjective optimization algorithm for document clustering. GWO is a metaheuristic algorithm that works in the same way that grey wolves do when hunting prey. The proposed algorithm has the advantage of easily modifying clusters based on the objectives. The performance of the proposed algorithm on the Reuters-21578 dataset demonstrates that it outperforms k-means and affinity propagation algorithms.

Predicting Australia’s Domestic Airline Passenger Demand using an Anfis Approach

Abstract The forecasting of future airline passenger demand is critical task for airline management. The objective of the present study was to develop an adaptive neuro-fuzzy inference system (ANFIS) for predicting Australia’s domestic airline passenger demand. The ANFIS model was trained, tested, and validated in the study. Sugeno fuzzy rules were used in the ANFIS structure and Gaussian membership function, and linear membership functions were also developed. The hybrid learning algorithm and the subtractive clustering partition method were used to generate the optimum ANFIS models. The results found that the mean absolute percentage error (MAPE) for the overall data set of the ANFIS model was 3.25% demonstrating that the ANFIS model has high predictive capabilities. The ANFIS model could be used in other domestic air travel markets.

Deep linear graph attention model for attributed graph clustering

Attributed graph clustering is an important task for grouping the nodes in a graph. In recent years, the algorithms based on graph convolutional networks (GCN) have achieved promising performance. However, almost all existing methods ignore that the nonlinearity between two consecutive GCN layers is unnecessary for improving the performance of attributed graph clustering, and may even harm the efficiency of the model. In this paper, we propose a novel deep linear graph attention model for attributed graph clustering (DLGAMC), which consists of an attention-based aggregation module and a similarity preserve module. Specifically, we simply exploit cosine similarity to construct the attention for aggregation, which does not need to learn extra attention parameters. It is worth noting that the attention we designed not only explicitly considers the similarity of attribute information, but also implicitly takes into account the local graph structure. To select the proper order of aggregation, we propose an adaptive strategy to evaluate the smoothness of node representations, where intra-cluster distance and inter-cluster distance are the key indicators in this process. To learn node representations for clustering, we design a similarity preserve module to preserve local similarity and global dissimilarity of the smooth features obtained by multiple aggregations, which is different from the ideas in reconstruction-based methods. Finally, k-means is performed on the learned representations to obtain the cluster partition. Experiments on several datasets show that our algorithm achieves great performance in attributed graph clustering tasks.