Options
Stoffel, Kilian
Nom
Stoffel, Kilian
Affiliation principale
Fonction
Professeur.e ordinaire
Identifiants
Résultat de la recherche
10 Résultats
Voici les éléments 1 - 10 sur 10
- PublicationAccès libreInvestigating Hyperparameter Optimization and Transferability for ES-HyperNEAT: A TPE Approach(2024-08)
; ; Neuroevolution of Augmenting Topologies (NEAT) and its advanced version, Evolvable-Substrate HyperNEAT (ES-HyperNEAT), have shown great potential in developing neural networks. However, their effectiveness heavily depends on the selection of hyperparameters. This study investigates the optimization of ES-HyperNEAT hyperparameters using the Tree-structured Parzen Estimator (TPE) on the MNIST classification task, exploring a search space of over 3 billion potential combinations. TPE effectively navigates this vast space, significantly outperforming random search in terms of mean, median, and best accuracy. During the validation process, the best hyperparameter configuration found by TPE achieves an accuracy of 29.00% on MNIST, surpassing previous studies while using a smaller population size and fewer generations. The transferability of the optimized hyperparameters is explored in logic operations and Fashion-MNIST tasks, revealing successful transfer to the more complex Fashion-MNIST problem but limited to simpler logic operations. This study emphasizes a method to unlock the full potential of neuroevolutionary algorithms and provides insights into the hyperparameters’ transferability across tasks of varying complexity. - PublicationAccès libre
- PublicationAccès libreObject-oriented Software Modeling with Ontologies Around(2018-7-1)
; Despite the many integration tools proposed for mapping between OWL ontologies and the object-oriented paradigm, developers are still reluctant to incorporate ontologies into their code repositories. In this paper we survey existing approaches for OWL to OOP mapping trying to identify reasons for this shy adoption of ontologies among conventional software developers. We present a classification of the surveyed approaches and tools based on the characteristics of their resulting artifacts. We finally provide our own reflection for other potential reasons beyond those addressed in the literature. - PublicationAccès libreUsing association rules to guide evolutionary search in solving constraint satisfaction(2015-5-25)
; - PublicationAccès libreThe CriLiM Methodology: Crime Linkage with a Fuzzy MCDM Approach(: IEEE, 2013-8-12)
; ; - PublicationAccès libreCrime Linkage: a Fuzzy MCDM Approach(: IEEE, 2013-6-4)
; ; - PublicationAccès libreFrom Police Reports to Data Marts: a Step Towards a Crime Analysis FrameworkNowadays, crime analyses are often conducted with computational methods. These methods, using several different systems (such as decision support systems), need to handle forensic data in a specific way. In this paper we present a methodology to structure police report data for crime analysis. The proposed artifact is mainly about applying data warehousing concepts to forensic data in a crime analysis perspective. Moreover, a proof of concept is carried out with real forensic data to illustrate and evaluate our methodology. These experiments highlight the need of such framework for crime analysis.
- PublicationAccès libreFuzzy Extended BPMN for Modelling Crime Analysis Processes(2011-5)
; In the frame of an overall project concerning the development of an intelligent process-driven framework for crime analysis, the modelling phase of crime analysis processes requires formal approaches being able to capture both the vague nature of forensic data and the uncertainties and conjectures characterizing the inference structures of this domain. In this context, a first analysis on the feasibility of a fuzzy embedded BPMN using the extensibility mechanism introduced by BPMN 2.0 specification is considered. - PublicationAccès libreFuzzy methods for forensic data analysis(2010-12)
; ; In this paper we describe a methodology and an automatic procedure for inferring accurate and easily understandable expert-system-like rules from forensic data. This methodology is based on the fuzzy set theory. The algorithms we used are described in detail, and were tested on forensic data sets. We also present in detail some examples, which are representative for the obtained results. - PublicationAccès libreOptimal Scene Interpretation: Semantic Management of 3-D Object from a Point Cloud Scene(2009-5)
; ; This paper presents the main concepts of a project under development concerning the analysis process of a scene containing a large number of objects, represented as unstructured point clouds. To achieve what we called the ―optimal scene interpretation‖ (the shortest scene description satisfying the MDL principle) we follow an approach for managing 3-D objects based on a semantic framework based on ontologies for adding and sharing conceptual knowledge about spatial objects.