Clark Graduate School of Geography, Clark University, Worcester, 01610, USA; Département de Géographie, Université de Montréal, Montréal, USA; Department of Natural Resource Sciences and McGill School of Environment, McGill University, Sainte-Anne-de-Bellevue, Canada; Geography and Urban Studies Department, Temple University, Philadelphia, 19122, USA; Department of International Development, Co...

Face detection in the wild needs to deal with various challenging conditions, which often leads to the situation where intraclass difference of faces exceeds interclass difference between faces and nonfaces. Based on this observation, in this paper we propose a locally rejected metric learning (LRML) based false positives filtering method. We firstly learn some prototype faces with affinity pro...

In this paper we tackle the issue of clustering trajectories of geolocalized observations. Using clustering technics based on the choice of a distance between the observations, we first provide a comprehensive review of the different distances used in the literature to compare trajectories. Then based on the limitations of these methods, we introduce a new distance : Symmetrized Segment-Path Di...

Data clustering is very useful in helping users visit the large scale of data in digit library. In this paper, we present an improved algorithm for clustering large scale of data set with dense relationship based on Affinity Propagation. First, the input data are divided into several groups and Affinity Propagation is applied to them respectively. Results from first step are grouped together in...

We analyze and exploit some scaling properties of the affinity propagation (AP) clustering algorithm proposed by Frey and Dueck [Science 315, 972 (2007)]. Following a divide and conquer strategy we setup an exact renormalization-based approach to address the question of clustering consistency, in particular, how many cluster are present in a given data set. We first observe that the divide and ...

Clustering is a fundamental problem in machine learning and has been approached in many ways. Two general and quite different approaches include iteratively fitting a mixture model (e.g., using EM) and linking together pairs of training cases that have high affinity (e.g., using spectral methods). Pair-wise clustering algorithms need not compute sufficient statistics and avoid poor solutions by...