Biometric authentication methods are being increasingly used for many types of applications. Since such necessitate humans to interact with a device, effective implementation requires consideration the perceptions and responses end users. Towards this goal, we present in paper modality-independent evaluation methodology study users' acceptance satisfaction biometric systems. It is based on use questionnaire some data mining tools analysis. We have applied it two systems developed our...

We propose in this paper a low cost multimodal biometric system combining keystroke dynamics and 2D face recognition. The objective of the proposed is to be used while keeping mind: good performances, acceptability, espect privacy. Different fusion methods have been (min, max, mul, svm, weighted sum configured with genetic algorithms, and, programming) on scores three algorithms two recognition ones. This improves rate comparison each individual method. On chimeric database composed 100...

Mobile Money Transfer (MMT) services provided by mobile network operators enable funds transfers made on devices of end-users, using digital equivalent cash (electronic money) without any bank accounts involved. MMT simplifies banking relationships and facilitates financial inclusion, and, therefore, is rapidly expanding all around the world, especially in developing countries. systems are subject to same controls as those required for institutions, including detection Laundering (ML) - a...

Mobile payments become more and popular thus are very attractive targets for fraudsters. As the latter always find new ways to commit crimes avoid detection, research in field of fraud is evolving. However, transactional data feedback from existing services lacking. This article addresses this issue by proposing a synthetic generator. Our idea model behavior various actors generate testing that researchers can use evaluate approaches identifying fraudulent transactions. paper presents our...

This paper presents a modality-independent evaluation methodology to study users’ acceptance and satisfaction of biometric systems. It uses survey questionnaire for data collection, some data-mining tools their analysis. We have applied it on two systems developed in our research laboratory. The results from this show the necessity taking point view when designing evaluating A panel 100 volunteers was more satisfied keystroke system than face one. Users surprisingly considered that its...

This paper presents an overview of a first-year PhD thesis. The main idea can be expressed as community detection issue in real network containing data from telecommunications operator. More precisely, mobile money transactions will studied, with view to detecting fraudulent ones. For this purpose, we propose address the problem by studying different approaches. Indeed, network, communities are constituted between users sharing common habits, therefore may appear unusual behaviors....

This paper deals with a comparative study of metrics allowing the evaluation results provided by object localization algorithms. We particularly focus on bounding box representation. 26 are studied in this paper. A protocol is presented for creation ground truths and synthetic These permit to simulate several inaccuracies (translation, scale errors... ) behaviors face considered alterations. Finally, some conclusions perspectives given.

We propose in this paper to study different color spaces for representing an image the face authentication application. used a generic algorithm based on matching of keypoints using sift descriptors computed one component. Ten have been studied four large and significant benchmark databases (ENSIB, FACES94, AR FERET). show that all do not provide same efficiency use information allows interesting improvement verification results.

We propose in this paper a new evaluation metric that enables to quantify the quality of an image interpretation result.This takes into account priori knowledge used by algorithm and ground truth associated with original image.We combine two metrics evaluate localization recognition results each detected object.We show proposed fulfills some theoretical properties has correct behavior face empirical experiments on benchmark database.We think could be reliable reference for video...

This paper analyzes the benefit and limitations of using a particular biometric technology "namely face authentication" for banking applications. We present first general concepts banking. propose method in order to replace PIN code authentication by biometrics data. Biometric is then detailed. A recognition we developed presented revealing as itself candidate solution. show limits this approach be used real industrial context.

Many object descriptors have been proposed in the state of art. For many reasons (occlusion, point view, acquisition conditions...), local a better robustness for image understanding applications. The goal this paper is to make comparative study eight recent descriptors. objective here quantify their ability generate automatically an taxonomy. In order answer question, we use Caltech256 benchmark which provides large taxonomy used as reference. This shows that SIFT, differential invariants...

One of the main factors affecting performance biometric systems is quality acquired samples. Poor-quality samples increase enrollment failure, and decrease system performance. Therefore, it important for a to estimate Toward this goal, we present in paper multi-class SVM-based method predict sample quality. The proposed uses two types information: first one based on image second pattern-based using SIFT keypoints extracted from image. For experiments, use four large significant face...

We present in this paper a comparative study of invariant descriptors for face recognition. Despite their multiple advantages (pose and scale characterization), few methods from the state art exploit kind features. A support vector machine is used training recognition steps. Experimental results show efficiency such on AR database. With proposed method, correctly achieved 97.4% cases

Networks have provided a representation for wide range of real systems, including communication flow, money transfer or biological to mention just few. Communities represent fundamental structures understanding the organization real-world networks. Uncovering coherent groups in these networks is goal community detection. A mesoscopic structure with nodes heavily connected within their by comparison other groups. might also overlap as they may share one multiple nodes. Evaluating results...