Libo Ren

A MapReduce-based approach for shortest path problem in large-scale networks

The cloud computing allows to use virtually infinite resources, and seems to be a new promising opportunity to solve scientific computing problems. The MapReduce parallel programming model is a new framework favoring the design of algorithms for cloud computing. Such framework favors processing of problems across huge datasets using a large number of heterogeneous computers over the web. In this paper, we are interested in evaluating how the MapReduce framework can create an innovative way for solving operational research problems. We proposed a MapReduce-based approach for the shortest path problem in large-scale real-road networks. Such a problem is the cornerstone of any real-world routing problem including the dial-a-ride problem (DARP), the pickup and delivery problem (PDP) and its dynamic variants. Most of efficient methods dedicated to these routing problems have to use the shortest path algorithms to construct the distance matrix between each pair of nodes and it could be a time-consuming task on a large-scale network due to its size. We focus on the design of an efficient MapReduce-based approach since a classical shortest path algorithm is not suitable to accomplish efficiently such task. Our objective is not to guarantee the optimality but to provide high quality solutions in acceptable computational time. The proposed approach consists in partitioning the original graph into a set of subgraphs, then solving the shortest path on each subgraph in a parallel way to obtain a solution for the original graph. An iterative improvement procedure is introduced to improve the solution. It is benchmarked on a graph modeling French road networks extracted from OpenStreetMap. The results of the experiment show that such approach achieves significant gain of computational time.

A smartphone-driven methodology for estimating physical activities and energy expenditure in free living conditions

This paper introduces a function dedicated to the estimation of total energy expenditure (TEE) of daily activities based on data from accelerometers integrated into smartphones. The use of mass-market sensors such as accelerometers offers a promising solution for the general public due to the growing smartphone market over the last decade. The TEE estimation function quality was evaluated using data from intensive numerical experiments based, first, on 12 volunteers equipped with a smartphone and two research sensors (Armband and Actiheart) in controlled conditions (CC) and, then, on 30 other volunteers in free-living conditions (FLC). The TEE given by these two sensors in both conditions and estimated from the metabolic equivalent tasks (MET) in CC served as references during the creation and evaluation of the function. The TEE mean gap in absolute value between the function and the three references was 7.0%, 16.4% and 2.7% in CC, and 17.0% and 23.7% according to Armband and Actiheart, respectively, in FLC. This is the first step in the definition of a new feedback mechanism that promotes self-management and daily-efficiency evaluation of physical activity as part of an information system dedicated to the prevention of chronic diseases.

A Multi-Start Split based Path Relinking (MSSPR) approach for the vehicle routing problem with route balancing

This paper addresses the vehicle routing problem with route balancing (VRPRB), which aims at minimizing two criteria simultaneously: the total routing cost and the difference between the largest and smallest route cost. We propose a multi-start approach based on two search spaces each of them using a different solution presentation: a TSP tour that denotes an indirect solution based on a sequence of customers as in the Traveling Salesman Problem, and a VRPRB solution that denotes a complete solution containing a set of vehicle trips. Switching from an indirect to a complete solution is possible through an adaptation of a splitting algorithm considering both optimization criteria. More precisely, such an adaptation requires an acceptance criterion allowing the generation a set of non-dominated VRPRB solutions from a single TSP tour. A path relinking algorithm improves the set of obtained VRPRB solutions. The proposed method is evaluated on VRPRB instances derived from classical VRP instances and the results reveal the method as effective in comparison with the best published algorithms for the problem optimizing the total routing cost. Regarding both criteria, the method competes with a previous published method handling the VRPRB. In fact, it is able to provide similar results in shorter computational time and since no details are available on state-of-the-art fronts, no further conclusion can be made. A web page presents all the solutions on our fronts to favor future comparative studies. Furthermore, the proposed method allows tackling a variant of the problem ignored by the previous works on VRPRB, which integrates limitation on vehicle service time.

A GRASP×ELS approach for the job-shop with a web service paradigm packaging

The Job-Shop Scheduling Problem (JSSP) is well known for its complexity as an NP-hard disjunctive scheduling problem. The problem addressed in this paper is JSSPs with an objective of minimizing makespan while satisfying a number of hard constraints. An efficient GRASPxELS approach is introduced for solving this problem. The efficiency is evaluated using the widely known 40 Laurences instances which encompass medium and large scale instances. The computational results prove that the proposed method competes with the best published methods in both quality of results and computational time. Recently, Web services have generated great interest in researchers. Such application architecture is based on the client-server model using existing Internet protocols and open standards. It provides new approaches to optimization methods. The proposed GRASPxELS is packaged into a Web Service (WS), i.e., it offers for the research community an open access to our optimization approach. Moreover, the proposed web service can be even included in research future works with a very small programming effort. To favor utilization of the web service and to prove the facility in which the service could be used, we provide an example in Java proving that it is possible to obtain in less than 10min a client application using the different methods exposed by this web service. Such usage extends to classical library inclusion in program with the difference that a method is called in the client side and represents an execution on the server. The Web Service paradigm is a new approach in spreading algorithms and therefore this paper stands at the crossroads of optimization research community and the web service community expectations. The GRASPxELS provided in the web service, is a state of the art method which competes with previously published ones and which has the advantage of being available for free, in any languages, everywhere contributing in spreading operational research contribution.

An acceleration vector variance based method for energy expenditure estimation in real-life environment with a smartphone/smartwatch integration

Investigating the TEE evaluation by predictive functions using smart-{phone, watch}.Using a personalized MET value in regard of the characteristics of participants.New activities classification model to obtain an TEE estimation.Gap less than 4% on both activities classification and TEE.Free Research application available on Google Play. By combining embedded passive sensing technologies from both smartphone and smartwatch, it is possible to obtain a high quality detection of sedentary activities (sitting, reclining postureź), movements (walkingź) and periods of more intense body movements (runningź). Our research encompasses the definition of an energy-saving function for the total energy expenditure (TEE) estimation using accelerometry data. This topic is clearly at the crossroad of both computer science and medical research. The present contribution proposes an intelligent wearable system, which combines the use of two complementary devices: smartphone and smartwatch to collect accelerometry data. Together they can precisely discriminate real-world human sedentary and active behaviors and their duration and estimate energy expenditure in real time and in free-living conditions. The results of the study are expected to help subjects to handle their daily-living physical activity notably for being compliant with the physical activity international guidelines (150min of moderate intensity activity/week). It is also expected that the physical activity feedbacks using these popular devices can prove the effectiveness of such wearable objects to promote individually-adapted healthy behavioral changes. The performance of the proposed function was evaluated by comparing the energy expenditure given by the smartphone and smartwatch with that produced by Armbandź. The mean error of TEE between the proposed function and Armbandź was less than 4% for an average 6h period of daily-living activities. The main theoretical contribution is the definition of a new predictive mathematical function of energy expenditure, which competes with the non-public function used in dedicated costly devices such as Armbandź. In addition, this work demonstrates the potential of wearable technologies.

The eMouveRecherche application competes with research devices to evaluate energy expenditure, physical activity and still time in free-living conditions

The proliferation of smartphones is creating new opportunities to monitor and interact with human subjects in free-living conditions since smartphones are familiar to large segments of the population and facilitate data collection, transmission and analysis. From accelerometry data collected by smartphones, the present work aims to estimate time spent in different activity categories and the energy expenditure in free-living conditions. Our research encompasses the definition of an energy-saving function (PredEE) considering four physical categories of activities (still, light, moderate and vigorous), their duration and metabolic cost (MET). To create an efficient discrimination function, the method consists of classifying accelerometry-transformed signals into categories and of associating each category with corresponding Metabolic Equivalent Tasks. The performance of the PredEE function was compared with two previously published functions (f(η,d)aedes,f(η,d)nrjsi), and with two dedicated sensors (Armband® and Actiheart®) in free-living conditions over a 12-h monitoring period using 30 volunteers. Compared to the two previous functions, PredEE was the only one able to provide estimations of time spent in each activity category. In relative value, all the activity categories were evaluated similarly to those given by Armband®. Compared to Actiheart®, the function underestimated still activities by 10.1% and overestimated light- and moderate-intensity activities by 7.9% and 4.2%, respectively. The total energy expenditure error produced by PredEE compared to Armband® was lower than those given by the two previous functions (5.7% vs. 14.1% and 17.0%). PredEE provides the user with an accurate physical activity feedback which should help self-monitoring in free-living conditions.

eMouveRecherche: the first scientific application to promote light-intensity activity for the prevention of chronic diseases

Physical inactivity and long sedentary time are involved in the development of chronic diseases. The aim of this study was to compare the intensity of spontaneous physical activity in two population samples consisting of 30 normal weight and 30 overweight or obese adults. Physical activity on an ordinary day was evaluated using the eMouveRecherche application that collected and sent smartphone accelerometry data to the ActivCollector Web platform via Internet. The algorithms implemented on the platform can accurately discriminate between sedentary and active behaviors, including their duration expressed in minutes and in percentage of waking period, in real time and ecological conditions. Physical activities are divided into four categories by the algorithms according to their intensity: immobility, light, moderate and vigorous intensity. The data were collected in 2013 and 2014 in Clermont-Ferrand and analyzed in 2015-2017. Time spent in only two categories was found to differ between the two populations. Immobile activities were longer in overweight than in normal weight participants (652 min vs. 504 min, 81.4% vs. 65.0%, p<0.0001). In contrast, the light-intensity activities were more popular in normal weight than in overweight participants (215 min vs. 124 min, 29.5% vs. 15.4%, p<0.0001). No difference was observed for either the moderateor vigorous-intensity categories. BMI and waist circumference were positively correlated with immobility and negatively with light-intensity activities. The results provided additional interesting indications in terms of time spent in light-intensity activities associated with normal weight status and seem to support the positive effect of this activity category on health. Correspondence to: Sylvie Rousset, Université Clermont Auvergne, INRA, UNH, Unité de Nutrition Humaine, 63000 Clermont-Ferrand, France, Tel: +33 (0)4 73 62 46 79; E-mail: sylvie.rousset@inra.fr

Support Vector machine and Monte Carlo simulation for robust optimization of industrial processes

Goods-producing industries continuously search to improve the quality of final products. The main approach is to identify a correlation between the process settings and the quality of the final product. In this work, a three steps robust approach is presented to improve an industrial process. The first step consists in using a Support Vector machines Regression (SVR) method to build a model of the considered process. It is based on the historic process data defined by an output (a criterion on the product quality) and multiple inputs (various production line settings). Then an optimization step based on an iterative descent method is done on the obtained model to identify interesting settings. Finally the set of settings found is validated by a Monte Carlo simulation approach used to simulate and test settings close to the one found on the optimization step. The proposed regression and optimization methods are compared to existing methods from the literature on a fluidized bed combustion boiler in the context of paper industry. The experiments confirm the efficiency of our approach.

A Split based Approach for the Vehicle Routing Problem with Route Balancing

The vehicle routing problem with route balancing is a bi-objective routing problem, in which the total route length and the balance of routes (i.e. the difference between the maximal and minimal route length) are minimized. In this paper, we propose an approach based on using of two solution representations: a giant tour representing a sequencing of the all customers and a complete solution with a decomposition of the giant tour, combining with a split algorithm to alternate between them. It offres a particularly efficient way to explore the solution space. The originality here is to adapt the split algorithm considering both of two objectives. An evolutionary path relinking algorithm is embedded to improve the obtained solutions. The proposed approach is evaluated on classical vehicle routing problem instances and the results push us into accepting the method as competitive with the best published mono-objective methods. On a bi-objective point of view, our method is competitive with the lexicographic solutions reported in the literature in the sense that it provides similar or better results in comparable computational time.

Shortest Path Challenging Problem

The shortest path problem is a well know routing problem which received a considerable amount of attention for several decades. This problem is the cornerstone of any real-world routing problem including the VRP or the Hub Location. The majority of efficient methods dedicated to these problems consist in computing first the matric of shortest path between nodes. Furthermore, there has been a renaissance of interest in the shortest path problem in recent year for use in various transportation engineering applications. This paper relates to the conception of efficient routing algorithms tuned for mobility. More precisely, it is targeted to the field of pedestrian mobility in an urban environment. In a mobile environment, specific constraints as the treatment of wireless network traffic disturbances must be taken into account. The architecture that we tune for the project is based on an active monitoring system, which dynamically required new shortest path calculation using the exposed web service API. The web service is performed when a specific constraint appears or a new part of the path is required. Using of such architecture offers a new approach in spreading operational research algorithms and our contribution stands at the crossroads of optimization research community and the web service community expectations.