Renzo Massobrio

Support Vector Machine Acceleration for Intel Xeon Phi Manycore Processors

Support vector machines are widely used for classification and regression tasks. However, sequential implementations for support vector machines are usually unable to deal with the increasing size of current real-world learning problems. In this context, Intel®Xeon PhiTM processors allow easily incorporating high performance computing strategies to improve execution times. This article proposes a parallel implementation of the popular LIBSVM library, specially adapted to the Intel®Xeon PhiTM architecture. The proposed implementation is evaluated using publicly available datasets corresponding to classification and regression tasks. Results show that the proposed parallel version computes the same results than the original LIBSVM while reducing the time needed for training by up to a factor of 4.81.

Single and Multiobjective Evolutionary Algorithms for Clustering Biomedical Information with Unknown Number of Clusters

This article presents single and multiobjective evolutionary approaches for solving the clustering problem with unknown number of clusters. Simple and ad-hoc operators are proposed, aiming to keep the evolutionary search as simple as possible in order to scale up for solving large instances. The experimental evaluation is performed considering a set of real problem instances, including a real-life problem of analyzing biomedical information in the Parkinson’s disease map project. The main results demonstrate that the proposed evolutionary approaches are able to compute accurate trade-off solutions and efficiently handle the problem instance involving biomedical information.

Multiobjective Vehicle-type Scheduling in Urban Public Transport

In this paper, we study the problem of vehicle scheduling in urban public transport systems taking into account the vehicle-type (different capacity and operating cost) known as VTSP. It is modeled as a multiobjective optimization problem (MOP). We propose a heuristic based on MOCell (Multi-Objective Cellular evolutionary algorithm) to solve the problem considering restrictions of government agencies in context of smart cities to improve the Intelligent Transportation Systems (ITS). A set of non-dominated solutions represents different assignments of vehicles to cover trips of a specific route. The conflicting objectives of provider and users (passenger) are to minimize the total operating cost, and maximize the quality of service, reducing the waiting time and congestion in buses. We present experimental analysis and conclude that the proposed heuristic provides a good performance and competitive results in terms of convergence and diversity of the solutions along the Pareto front.

Infrastructure Deployment in Vehicular Communication Networks Using a Parallel Multiobjective Evolutionary Algorithm: INFRASTRUCTURE DEPLOYMENT IN VANETs USING A PARALLEL MOEA

This article describes the application of a multiobjective evolutionary algorithm for locating roadside infrastructure for vehicular communication networks over realistic urban areas. A multiobjective formulation of the problem is introduced, considering quality‐of‐service and cost objectives. The experimental analysis is performed over a real map of Málaga, using real traffic information and antennas, and scenarios that model different combinations of traffic patterns and applications (text/audio/video) in the communications. The proposed multiobjective evolutionary algorithm computes accurate trade‐off solutions, significantly improving over state‐of‐the‐art algorithms previously applied to the problem.

Distributed Big Data Analysis for Mobility Estimation in Intelligent Transportation Systems

This article describes the application of distributed computing techniques for the analysis of big data information from Intelligent Transportation Systems. Extracting useful mobility information from large volumes of data is crucial to improve decision-making processes in smart cities. We study the problem of estimating demand and origin-destination matrices based on ticket sales and location of buses in the city. We introduce a framework for mobility analysis in smart cities, including two algorithms for the efficient processing of large mobility data from the public transportation in Montevideo, Uruguay. Parallel versions are proposed for distributed memory (e.g., cluster, grid, cloud) infrastructures and a cluster implementation is presented. The experimental analysis performed using realistic datasets demonstrate that significatively speedup values, up to 16.41, are obtained.

Towards a Cloud Computing Paradigm for Big Data Analysis in Smart Cities

In this paper, we present a Big Data analysis paradigm related to smart cities using cloud computing infrastructures. The proposed architecture follows the MapReduce parallel model implemented using the Hadoop framework. We analyse two case studies: a quality-of-service assessment of public transportation system using historical bus location data, and a passenger-mobility estimation using ticket sales data from smartcards. Both case studies use real data from the transportation system of Montevideo, Uruguay. The experimental evaluation demonstrates that the proposed model allows processing large volumes of data efficiently.

Multiobjective evolutionary algorithms for the taxi sharing problem

Transportation planning plays a central role in the design and development of smart cities. In particular, the concept of sharing economy applied to urban transportation is gaining massive public attention in recent years. This article presents the application of two multiobjective evolutionary algorithms to the problem of distributing passengers travelling from the same origin to different destinations in several taxis. A new problem formulation is presented, accounting for two quality of service metrics from the point of view of taxi users: the total cost of the trips and the delay experienced by each passenger. Two multiobjective evolutionary algorithms are proposed: a parallel microevolutionary algorithm following a linear aggregation approach to combine the problem objectives and one well-known algorithm from the literature following a full multiobjective approach based on Pareto dominance. Both algorithms are compared against each other and against two greedy heuristics based on the ideas presented in the related literature. The experimental evaluation is performed over a set of 88 problem instances generated using real GPS taxi data. Results show that the proposed algorithms are able to efficiently reach significant improvements in both problem objectives over the greedy heuristics in short execution times.