Neighbor-based Method Research Articles

Multi-Layer Stochastic Block Interaction driven by Logistic Regression (MLSBI-LR) for Efficient Link Recommendation in Intra-Layer Linkage Graphs

Link Recommendation (LR) in complex networks has attracted huge interest in the social and computer science communities. Numerous networks, such as recommendation systems and social networks (which facilitate user contact), are probabilistic rather than deterministic due to the uncertainty surrounding the presence of links. Evaluating the various characteristics of such networks has frequently been tricky as the Intra-layer linkage graph requires at least two nodes to be in the same layer. Moreover, many existing LR methods mainly operate well on Single-Layer Graphs (SLGs) compared to Multi-Layer Graphs (MLGs) when nodes traverse multi-layers in a network of Intra-layer linkages. Considering this drawback, this paper proposes a Multi-Layer Stochastic Block Interaction method driven by Logistic Regression (MLSBI-LR) to exploit the bi-directional resources associated with Intra-layer linkages. Its inherent dependence on knowledge-based systems uses multi-criteria recommender systems to accommodate additional criteria and can modify neighborhood-based approaches. A multi-criteria network with relations over the same set of nodes is used since the modified neighbor-based method can exhibit rich dependence between entities and have available experimental data sets in MLGs to recommend links that would efficiently enrich users' experience. The accuracy and robustness of the proposed MLSBI-LR method compared to existing LR methods were extensively investigated using three distinct benchmark data sets and four evaluation metrics. Based on our experimental results across the databases and metrics, the proposed MLSBI-LR method performed significantly better (recording up to 17% increment in accuracy), recommending potential links in MLGs. Consequently, the proposed method may revolutionize link recommendation tasks in social networks by improving users' overall experience.

Measuring Traffic in Cities Through a Large-Scale Online Platform

Online real-time traffic data services could effectively deliver traffic information to people all over the world and provide large benefits to the society and research about cities. Yet, city-wide road network traffic data are often hard to come by on a large scale over a longer period of time. We collect, describe, and analyze traffic data for 45 cities from HERE, a major online real-time traffic information provider. We sampled the online platform for city traffic data every 5 min during 1 year, in total more than 5 million samples covering more than 300 thousand road segments. Our aim is to describe some of the practical issues surrounding the data that we experienced in working with this type of data source, as well as to explore the data patterns and see how this data source provides information to study traffic in cities. We focus on data availability to characterize how traffic information is available for different cities; it measures the share of road segments with real-time traffic information at a given time for a given city. We describe the patterns of real-time data availability, and evaluate methods to handle filling in missing speed data for road segments when real-time information was not available. We conduct a validation case study based on Swedish traffic sensor data and point out challenges for future validation. Our findings include (i) a case study of validating the HERE data against ground truth available for roads and lanes in a Swedish city, showing that real-time traffic data tends to follow dips in travel speed but miss instantaneous higher speed measured in some sensors, typically at times when there are fewer vehicles on the road; (ii) using time series clustering, we identify four clusters of cities with different types of measurement patterns; and (iii) a k-nearest neighbor-based method consistently outperforms other methods to fill in missing real-time traffic speeds. We illustrate how to work with this kind of traffic data source that is increasingly available to researchers, travellers, and city planners. Future work is needed to broaden the scope of validation, and to apply these methods to use online data for improving our knowledge of traffic in cities.

A reliable neighbor-based method for identifying essential proteins by integrating gene expressions, orthology, and subcellular localization information

Essential proteins are those necessary for the survival or reproduction of species and discovering such essential proteins is fundamental for understanding the minimal requirements for cellular life, which is also meaningful to the disease study and drug design. With the development of high-throughput techniques, a large number of Protein-Protein Interactions (PPIs) can be used to identify essential proteins at the network level. Up to now, though a series of network-based computational methods have been proposed, it is still a challenge to improve the prediction precision as the high false positives in PPI networks. In this paper, we propose a new method GOS to identify essential proteins by integrating the Gene expressions, Orthology, and Subcellular localization information. The gene expressions and subcellular localization information are used to determine whether a neighbor in the PPI network is reliable. Only reliable neighbors are considered when we analyze the topological characteristics of a protein in a PPI network. We also analyze the orthologous attributes of each protein to reflect its conservative features, and use a random walk model to integrate a protein's topological characteristics and its orthology. The experimental results on the yeast PPI network show that the proposed method GOS outperforms the ten existing methods DC, BC, CC, SC, EC, IC, NC, PeC, ION, and CSC.

Snapshot and continuous points-based trajectory search

Trajectory data capture the traveling history of moving objects such as people or vehicles. With the proliferation of GPS and tracking technologies, huge volumes of trajectories are rapidly generated and collected. Under this, applications such as route recommendation and traveling behavior mining call for efficient trajectory retrieval. In this paper, we first focus on distance-to-points trajectory search; given a collection of trajectories and a set query points, the goal is to retrieve the top-k trajectories that pass as close as possible to all query points. We advance the state-of-the-art by combining existing approaches to a hybrid nearest neighbor-based method while also proposing an alternative, more efficient spatial range-based approach. Second, we investigate the continuous counterpart of distance-to-points trajectory search where the query is long-standing and the set of returned trajectories needs to be maintained whenever updates occur to the query and/or the data. Third, we propose and study two practical variants of distance-to-points trajectory search, which take into account the temporal characteristics of the searched trajectories. Through an extensive experimental analysis with real trajectory data, we show that our range-based approach outperforms previous methods by at least one order of magnitude for the snapshot and up to several times for the continuous version of the queries.

Hidden Information Revealed by Optimal Community Structure from a Protein-Complex Bipartite Network Improves Protein Function Prediction

The task of extracting the maximal amount of information from a biological network has drawn much attention from researchers, for example, predicting the function of a protein from a protein-protein interaction (PPI) network. It is well known that biological networks consist of modules/communities, a set of nodes that are more densely inter-connected among themselves than with the rest of the network. However, practical applications of utilizing the community information have been rather limited. For protein function prediction on a network, it has been shown that none of the existing community-based protein function prediction methods outperform a simple neighbor-based method. Recently, we have shown that proper utilization of a highly optimal modularity community structure for protein function prediction can outperform neighbor-assisted methods. In this study, we propose two function prediction approaches on bipartite networks that consider the community structure information as well as the neighbor information from the network: 1) a simple screening method and 2) a random forest based method. We demonstrate that our community-assisted methods outperform neighbor-assisted methods and the random forest method yields the best performance. In addition, we show that using the optimal community structure information is essential for more accurate function prediction for the protein-complex bipartite network of Saccharomyces cerevisiae. Community detection can be carried out either using a modified modularity for dealing with the original bipartite network or first projecting the network into a single-mode network (i.e., PPI network) and then applying community detection to the reduced network. We find that the projection leads to the loss of information in a significant way. Since our prediction methods rely only on the network topology, they can be applied to various fields where an efficient network-based analysis is required.

λ-diverse nearest neighbors browsing for multidimensional data

Traditional search methods try to obtain the most relevant information and rank it according to the degree of similarity to the queries. Diversity in query results is also preferred by a variety of applications since results very similar to each other cannot capture all aspects of the queried topic. In this paper, we focus on the \lambda-diverse k-nearest neighbor search problem on spatial and multidimensional data. Unlike the approach of diversifying query results in a postprocessing step, we naturally obtain diverse results with the proposed geometric and index-based methods. We first make an analogy with the concept of Natural Neighbors (NatN) and propose a natural neighbor-based method for 2D and 3D data and an incremental browsing algorithm based on Gabriel graphs for higher dimensional spaces. We then introduce a diverse browsing method based on the distance browsing feature of spatial index structures, such as R-trees. The algorithm maintains a Priority Queue with mindivdist of the objects depending on both relevancy and angular diversity and efficiently prunes nondiverse items and nodes. We experiment with a number of spatial and high-dimensional data sets, including Factual's (http://www.factual.com/) US points-of-interest data set of 13M entries. On the experimental setup, the diverse browsing method is shown to be more efficient (regarding disk accesses) than k-NN search on R-trees, and more effective (regarding Maximal Marginal Relevance (MMR)) than the diverse nearest neighbor search techniques found in the literature.

<i>k</i>近傍法に基づく予測を用いた人物追跡のための複数エージェントの経路計画法

This paper deals with a multi-agent path planning problem for tracking humans in order to obtain detail information such like human behavior and characteristics. To achieve this, paths of agents is planned based on similarity between the predicted intensity of humans positions and the agents' field of view in the future. The positions of humans are predicted by a k-Nearest Neighbor-based method which allows to predict complicated human movements in a real environment. The number of steps of the planned path is determined according to the consistency between the current prediction and the previous prediction of the future human positions to avoid the path planning using less reliable predictions. We conducted computer simulations and results showed that agents can follow human trajectories observed in a real environment, i.e., concourse of a station by our path planning method.

A K-nearest neighbor-based method for the restoration of damaged images

The present paper deals with the problem of restoring a digital image starting from its random sampling or its damaged version. The method is adaptive and uses local evaluation of the kernel parameters. Applications of the method on simulated and real data are presented and its performance on the reconstruction of very sparse images is shown.