Maurice Tchuente

The Ross–Macdonald model in a patchy environment

We generalize to n patches the Ross-Macdonald model which describes the dynamics of malaria. We incorporate in our model the fact that some patches can be vector free. We assume that the hosts can migrate between patches, but not the vectors. The susceptible and infectious individuals have the same dispersal rate. We compute the basic reproduction ratio R(0). We prove that if R(0)1, then the disease-free equilibrium is globally asymptotically stable. When R(0)>1, we prove that there exists a unique endemic equilibrium, which is globally asymptotically stable on the biological domain minus the disease-free equilibrium.

Dynamic programming on two-dimensional systolic arrays

Given w(i, j), 1⩽i<j⩽n, and c(i, i+1), 1⩽i<n, the problem is to compute c(i, j) = w(i, j)+mini<k<j{c(i, k)+ c(k, j)}, for 1⩽i<j⩽n. We show that this dynamic programming problem can be solved in optimal time T = 2n on a systolic array whose size 18n2+O(n) is two times smaller than the number of processors required by the best previously known solution.

Churn Prediction in a Real Online Social Network Using Local CommunIty Analysis

Prediction of user behavior in Social Networks is important for a lot of applications, ranging from marketing to social community management. In this paper, we develop and test a model to estimate the propensity of a user to stop using the social platform in a near future. This problem is called churn prediction and has been extensively studied in telecommunication networks. We focus here on building a statistical model estimating the probability that a user will leave the social network in the near future. The model is based on graph attributes extracted in the users vicinity. We present a novel algorithm to accurately detect overlapping local communities in social graphs. Our algorithm outperforms the state of the art methods and is able to deal with pathological cases which can occur in real networks. We show that using attributes computed from the local community around the user allows to build a robust statistical model to predict churn. Our ideas are tested on one of the largest French social blog platform, Sky rock, where millions of teenagers interact daily.

WARIM: Wireless Sensor Networks Architecture for a Reliable Intersection Monitoring

A traffic light controller takes as input an estimation of the number of vehicles entering the intersection and produces as output a light plan, with the objective to reduce the traffic jam. The quality of the input traffic estimation is a key consideration on the performance of the traffic light controller. The advent of Wireless Sensor Networks, with their relatively low deployment and operation price, led to the development of several sensor-based architectures for intersection monitoring. We show in this paper that the solutions proposed in the literature are unrealistic in terms of communication possibilities and that they do not allow a measure of the vehicular queue length at a lane level. Based on extensive experimental results, we propose an energy efficient, low cost and lightweight multihop wireless sensor network architecture to measure with a good accuracy the vehicle queue length, in order to have a more precise vision of traffic at the intersection. Associated challenges are then discussed, such as self-configuration, routing and energy harvesting, which should be addressed in order to reduce the cost of the proposed solution and to improve the performance of the target application.

Low Birth Weight in Perinatally HIV-Exposed Uninfected Infants: Observations in Urban Settings in Cameroon

Background The consequences of maternal HIV infection for fetal growth are controversial. Here, we estimated the frequency of small for gestational age and gender (SGAG) among neonates born to HIV-infected or uninfected mothers and assessed the contribution, if any, of maternal HIV to the risk of SGAG. Methods The data used were obtained from the ANRS-Pediacam cohort in Cameroon. Pairs of newborns, one to a HIV-infected mother and the other to an uninfected mother, were identified during the first week of life, and matched on gender and recruitment site from 2007–2010. SGAG was defined in line with international recommendations as a birth weight Z-score adjusted for gestational age at delivery and gender more than two standard deviations below the mean (−2SD). Considering the matched design, logistic regression modeling was adjusted on site and gender to explore the effect of perinatal HIV exposure on SGAG. Results Among the 4104 mother-infant pairs originally enrolled, no data on birth weight and/or gestational age were available for 108; also, 259 were twins and were excluded. Of the remaining 3737 mother-infant pairs, the frequency of SGAG was 5.3% (95%CI: 4.6–6.0), and was significantly higher among HIV-infected infants (22.4% vs. 6.3%; p<.001) and lower among HIV-unexposed uninfected infants (3.5% vs. 6.3%; p<.001) than among HIV-exposed uninfected infants. Similarly, SGAG was significantly more frequent among HIV-infected infants (aOR: 4.1; 2.0–8.1) and less frequent among HIV-unexposed uninfected infants (aOR: 0.5; 0.4–0.8) than among HIV-exposed uninfected infants. Primiparity (aOR: 1.9; 1.3–2.7) and the presence of any disease during pregnancy (aOR: 1.4; 1.0–2.0) were identified as other contributors to SGAG. Conclusion Maternal HIV infection was independently associated with SGAG for HIV-exposed uninfected infants. This provides further evidence of the need for adapted monitoring of pregnancy in HIV-infected women, especially if they are symptomatic, to minimize additional risk factors for SGAG.

Scheduling of the dag associated with pipeline inversion of triangular matrices

We are interested in methods which compute the inverse of a triangular matrix A of order n by solving the n linear systems Ax=ei, i=1,…, n, where ei is the i-th element of the canonical basis of Rn. More precisely, we consider the dependence graph associated with algorithms where the entries of matrix A are read only once and used in pipeline for the solution of these systems. We exhibit a new scheduling which induces an algorithm with time complexity T*=2n−1. The number n2/8+O(n) of processors required by this scheduling improves the best previously known bound n2/6+O(n), and is quite close to the lower bound n2/8.5+O(n).

Parallelization of the Gaussian Elimination Algorithm on Systolic Arrays

We study the parallel implementation of the Gaussian elimination scheme on systolic arrays. We first show that the time (resp. area) complexity of the algorithm isT= 3n? 1 (resp.S= (n2/4) +O(n)), wherenis the size of the linear system. Then we exhibit three algorithms. The two first ones are optimal in time. The first one corresponds to an orthogonally connected array of size 3(n2/8) +O(n). The second network is smaller,S= 3(n2/10) +O(n), but two layers are necessary in order to obtain a regular layout with local communications. The third one is hexagonally connected, has size (n2/3) +O(n), and is almost optimal in time.

Systolic Gauss-Jordan elimination for dense linear systems

Abstract A systolic network for the Gauss-Jordan algorithms is presented for the solution of dense linear systems. This network compares favorably with Melhems network since its execution time (4 n − 1) is the same and the number of cells is decreased from n 2 2 to 3n 2 8 .

Cycles generated by sequential iterations

Abstract For any finite non-empty set Q, and for any positive integere n, we exhibit various classes of sequential transformations which realize a cyclic permutation of Qn. We also present some partial results on the longest cycles generated by monotone boolean functions of n variables.

Distance-Based Trace Diagnosis for Multimedia Applications: Help Me TED!

Execution traces have become essential resources that many developers analyze to debug their applications. Ideally, a developer wants to quickly detect whether there are anomalies on his application or not. However, in practice, the size of multimedia applications trace can reach gigabytes, which makes their exploitation very complex. Usually, developers use visualization tools before stating a hypothesis. In this paper, we argue that this solution is not satisfactory and propose to automatically provide a diagnosis by comparing execution traces. We use distance-based models and conduct a user case to show how TED, our automatic trace diagnosis tool, provides semantic added-value information to the developer. Performance evaluation over real world data shows that our approach is scalable.