R. Sabourin - Academia.edu (original) (raw)
Uploads
Papers by R. Sabourin
5th International Conference on Imaging for Crime Detection and Prevention (ICDP 2013), 2013
In video surveillance, person re-identification refers to recognizing individuals of interest fro... more In video surveillance, person re-identification refers to recognizing individuals of interest from faces captured across a network of video cameras. Face recognition in such applications is challenging because faces are captured with limited spatial and temporal constraints. In addition, facial models for recognition are commonly designed using limited reference samples from faces captured under specific conditions. Given new reference samples, updating facial models may allow maintaining a high level of performance over time. Although adaptive ensembles have been successfully applied to robust modeling of an individual's face, reference data samples must be stored for validation. In this paper, a memory management strategy based on Kullback-Leiber (KL) divergence is proposed to rank and select the most relevant validation samples over time in adaptive individual-specific ensembles. When new reference data becomes available for an individual, updates to the corresponding ensembles are validated using a mixture of new and previously-stored samples. Only the samples with the highest KL divergence are preserved in memory for future adaptations. The strategy is compared with reference classifiers using videos from the FIA data set. Simulation results show that the proposed strategy tends to select samples of statistically different subjects (so-called "wolfs") for validation, thereby reducing the number of samples per individual by up to 80%, yet maintaining a high level of performance.
Abstract In this paper, we propose a novel approach for the multi-objective optimization of class... more Abstract In this paper, we propose a novel approach for the multi-objective optimization of classifier ensembles in the ROC space. We first evolve a pool of simple classifiers with NSGA-II using values of the ROC curves as the optimization objectives. These simple classifiers are then combined at the decision level using the Iterative Boolean Combination method (IBC). This method produces multiple ensembles of classifiers optimized for various operating conditions.
Lecture Notes in Computer Science, 1997
The purpose of this paper is the assessment of a family of shape factors for o -line signature ve... more The purpose of this paper is the assessment of a family of shape factors for o -line signature veri cation. The initial method, suggested in 1], which extracts geometric features, is modi ed to assess the performance of veri cation in the general case of forgery. W e c o ntribute to the information coded by adapting the coding method to the image and by i n tegrating a spatial distance into the shape factor denition. Moreover, we include the coding of an information related to the dynamic of the signature. To use these two t ypes of information, we propose a fuzzy technique to combine and then obtain one kind of information. We e v aluate this new coding operator with two t ypes of forgery : random forgery and photocopy s i m ulation with some adapted protocols.
IEEE Annual Meeting of the Fuzzy Information, 2004. Processing NAFIPS '04., 2004
The supervisory control strategy set points for an existing HVAC system could be optimized using ... more The supervisory control strategy set points for an existing HVAC system could be optimized using a two-objective evolutionary algorithm. The set points for the supply air temperature, the supply duct static pressure, the chilled water temperature, and the zone temperatures are the problem variables, while energy use and thermal comfort are the objective functions. Different evolutionary algorithm methods for two-objective optimization in HVAC systems are evaluated. It was concluded that controlled elitist non-dominated sorting genetic algorithms offer great potential for finding the Paretooptimal solutions of investigated problems. The results also showed that the on-line implementation of optimization process could save energy by 19.5%. The two-objective optimization could also help control daily energy use while bringing about further energy use savings as compared to a one-objective optimization.
5th International Conference on Imaging for Crime Detection and Prevention (ICDP 2013), 2013
In video surveillance, person re-identification refers to recognizing individuals of interest fro... more In video surveillance, person re-identification refers to recognizing individuals of interest from faces captured across a network of video cameras. Face recognition in such applications is challenging because faces are captured with limited spatial and temporal constraints. In addition, facial models for recognition are commonly designed using limited reference samples from faces captured under specific conditions. Given new reference samples, updating facial models may allow maintaining a high level of performance over time. Although adaptive ensembles have been successfully applied to robust modeling of an individual's face, reference data samples must be stored for validation. In this paper, a memory management strategy based on Kullback-Leiber (KL) divergence is proposed to rank and select the most relevant validation samples over time in adaptive individual-specific ensembles. When new reference data becomes available for an individual, updates to the corresponding ensembles are validated using a mixture of new and previously-stored samples. Only the samples with the highest KL divergence are preserved in memory for future adaptations. The strategy is compared with reference classifiers using videos from the FIA data set. Simulation results show that the proposed strategy tends to select samples of statistically different subjects (so-called "wolfs") for validation, thereby reducing the number of samples per individual by up to 80%, yet maintaining a high level of performance.
Abstract In this paper, we propose a novel approach for the multi-objective optimization of class... more Abstract In this paper, we propose a novel approach for the multi-objective optimization of classifier ensembles in the ROC space. We first evolve a pool of simple classifiers with NSGA-II using values of the ROC curves as the optimization objectives. These simple classifiers are then combined at the decision level using the Iterative Boolean Combination method (IBC). This method produces multiple ensembles of classifiers optimized for various operating conditions.
Lecture Notes in Computer Science, 1997
The purpose of this paper is the assessment of a family of shape factors for o -line signature ve... more The purpose of this paper is the assessment of a family of shape factors for o -line signature veri cation. The initial method, suggested in 1], which extracts geometric features, is modi ed to assess the performance of veri cation in the general case of forgery. W e c o ntribute to the information coded by adapting the coding method to the image and by i n tegrating a spatial distance into the shape factor denition. Moreover, we include the coding of an information related to the dynamic of the signature. To use these two t ypes of information, we propose a fuzzy technique to combine and then obtain one kind of information. We e v aluate this new coding operator with two t ypes of forgery : random forgery and photocopy s i m ulation with some adapted protocols.
IEEE Annual Meeting of the Fuzzy Information, 2004. Processing NAFIPS '04., 2004
The supervisory control strategy set points for an existing HVAC system could be optimized using ... more The supervisory control strategy set points for an existing HVAC system could be optimized using a two-objective evolutionary algorithm. The set points for the supply air temperature, the supply duct static pressure, the chilled water temperature, and the zone temperatures are the problem variables, while energy use and thermal comfort are the objective functions. Different evolutionary algorithm methods for two-objective optimization in HVAC systems are evaluated. It was concluded that controlled elitist non-dominated sorting genetic algorithms offer great potential for finding the Paretooptimal solutions of investigated problems. The results also showed that the on-line implementation of optimization process could save energy by 19.5%. The two-objective optimization could also help control daily energy use while bringing about further energy use savings as compared to a one-objective optimization.