Browse ORBi by ORBi project

- Background
- Content
- Benefits and challenges
- Legal aspects
- Functions and services
- Team
- Help and tutorials

Clinical data based optimal STI strategies for HIV: a reinforcement learning approach Ernst, Damien ; ; et al in Proceedings of the 45th IEEE Conference on Decision and Control (CDC 2006) (2006) This paper addresses the problem of computing optimal structured treatment interruption strategies for HIV infected patients. We show that reinforcement learning may be useful to extract such strategies ... [more ▼] This paper addresses the problem of computing optimal structured treatment interruption strategies for HIV infected patients. We show that reinforcement learning may be useful to extract such strategies directly from clinical data, without the need of an accurate mathematical model of HIV infection dynamics. To support our claims, we report simulation results obtained by running a recently proposed batch-mode reinforcement learning algorithm, known as fitted Q iteration, on numerically generated data. [less ▲] Detailed reference viewed: 140 (9 ULg)Clinical data based optimal STI strategies for HIV: a reinforcement learning approach Ernst, Damien ; ; et al in Proceedings of the 15th Annual Machine Learning Conference of Belgium and The Netherlands (Benelearn 2006) (2006) This paper addresses the problem of computing optimal structured treatment interruption strategies for HIV infected patients. We show that reinforcement learning may be useful to extract such strategies ... [more ▼] This paper addresses the problem of computing optimal structured treatment interruption strategies for HIV infected patients. We show that reinforcement learning may be useful to extract such strategies directly from clinical data, without the need of an accurate mathematical model of HIV infection dynamics. To support our claims, we report simulation results obtained by running a recently proposed batch-mode reinforcement learning algorithm, known as tted Q iteration, on numerically generated data. [less ▲] Detailed reference viewed: 55 (20 ULg)Multi-area security assessment: results using efficient bounding method Wehenkel, Louis ; ; Ernst, Damien in Proceedings of the 38th North American Power Symposium (NAPS 2006) (2006) We present our recent results on using previously introduced framework for multi-area security assessment in large interconnections. The basic idea of the framework is exchanging just enough information ... [more ▼] We present our recent results on using previously introduced framework for multi-area security assessment in large interconnections. The basic idea of the framework is exchanging just enough information so that each operator can evaluate the impact in his control area of contingencies both internal and external to his area. We provide illustrations based on a localization concept known as efficient bounding method and recently introduced approximate DC model of the European interconnected system. In this paper we focus on four transmission system operators (within the approximate DC model): Belgium, France, Germany, and Netherlands (for both Summer and Winter-peak load conditions). [less ▲] Detailed reference viewed: 23 (3 ULg)Damping control by fusion of reinforcement learning and control Lyapunov functions ; Wehenkel, Louis ; Ernst, Damien in Proceedings of the 38th North American Power Symposium (NAPS 2006) (2006) The main idea behind the concept, proposed in the paper, is the opportunity to make control systems with increased capabilities by synergetic fusion of the domain-specific knowledge and the methodologies ... [more ▼] The main idea behind the concept, proposed in the paper, is the opportunity to make control systems with increased capabilities by synergetic fusion of the domain-specific knowledge and the methodologies from control theory and artificial intelligence. The particular approach considered combines Control Lyapunov Functions (CLF), a constructive control technique, an Reinforcement Learning (RL) in attempt to optimize a mix of system stability and performance. Two control schemes are proposed and the capabilities of the resulting controllers are illustrated on a control problem involving a Thyristor Controlled Series Capacitor (TCSC) for damping oscillations in a four-machine power system. [less ▲] Detailed reference viewed: 29 (2 ULg)Automatic learning of sequential decision strategies for dynamic security assessment and control Wehenkel, Louis ; ; Geurts, Pierre et al in Proceedings of the IEEE Power Engineering Society General Meeting 2006 (2006) This paper proposes to formulate security control as a sequential decision making problem and presents new developments in automatic learning of sequential decision making strategies from simulations and ... [more ▼] This paper proposes to formulate security control as a sequential decision making problem and presents new developments in automatic learning of sequential decision making strategies from simulations and/or information collected from real-life system measurements. The exploitation of these methods for the design of decision making strategies and control policies in the context of preventive and emergency mode dynamic security assessment and control is discussed and further opportunities for research in this area are highlighted. [less ▲] Detailed reference viewed: 29 (4 ULg)Ensembles of extremely randomized trees and some generic applications Wehenkel, Louis ; Ernst, Damien ; Geurts, Pierre in Proceedings of Robust Methods for Power System State Estimation and Load Forecasting (2006) In this paper we present a new tree-based ensemble method called “Extra-Trees”. This algorithm averages predictions of trees obtained by partitioning the inputspace with randomly generated splits, leading ... [more ▼] In this paper we present a new tree-based ensemble method called “Extra-Trees”. This algorithm averages predictions of trees obtained by partitioning the inputspace with randomly generated splits, leading to significant improvements of precision, and various algorithmic advantages, in particular reduced computational complexity and scalability. We also discuss two generic applications of this algorithm, namely for time-series classification and for the automatic inference of near-optimal sequential decision policies from experimental data. [less ▲] Detailed reference viewed: 91 (5 ULg)On multi-area security assessment of large interconnected power systems Wehenkel, Louis ; ; Ernst, Damien in Proceedings of the Second Carnegie Mellon Conference in Electric Power Systems: Monitoring, Sensing, Software and its Valuation for the Changing electric Power Industry (2006) The paper introduces a framework for information exchange and coordination of security assessment suitable for distributed multi-area control in large interconnections operated by a team of transmission ... [more ▼] The paper introduces a framework for information exchange and coordination of security assessment suitable for distributed multi-area control in large interconnections operated by a team of transmission system operators. The basic idea of the proposed framework consists of exchanging just enough information so that each operator can evaluate the impact in his control area of contingencies both internal and external to his area. The framework has been thought out with the European perspective in mind where it is presently not possible to set up a transnational security coordinator that would have authority to handle security control over the whole or part of the European interconnection. Nevertheless, it can also be considered as an approach to handle security control in North-American Mega-RTOs, where it could help to circumvent problems of scalability of algorithms and maintainability of data by distributing them over the TSOs under the authority of the RTO. [less ▲] Detailed reference viewed: 54 (0 ULg)About automatic learning for advanced sensing, monitoring and control of electric power systems Wehenkel, Louis ; ; Geurts, Pierre et al in Proceedings of the Second Carnegie Mellon Conference in Electric Power Systems: Monitoring, Sensing, Software and its Valuation for the Changing electric Power Industry (2006) The paper considers the possible uses of automatic learning for improving power system performance by software methodologies. Automatic learning per se is first reviewed and recent developements of the ... [more ▼] The paper considers the possible uses of automatic learning for improving power system performance by software methodologies. Automatic learning per se is first reviewed and recent developements of the field are highlighted. Then the authors’ views of its main actual or potential applications related to power system operation and control are described, and in each application present status and needs for further developments are discussed. [less ▲] Detailed reference viewed: 66 (4 ULg)Biological Image Classification with Random Subwindows and Extra-Trees Marée, Raphaël ; Geurts, Pierre ; Wehenkel, Louis Conference (2006) We illustrate the potential of our image classification method on three datasets of images at different imaging modalities/scales, from subcellular locations up to human body regions. The method is based ... [more ▼] We illustrate the potential of our image classification method on three datasets of images at different imaging modalities/scales, from subcellular locations up to human body regions. The method is based on random subwindows extraction and the combination of their classification using ensembles of extremely randomized decision trees. [less ▲] Detailed reference viewed: 64 (6 ULg)OK3: Méthode d’arbres à sortie noyau pour la prédiction de sorties structurées et l’apprentissage de noyau Geurts, Pierre ; Wehenkel, Louis ; in Proc. of CAP (Conférence francophone d'apprentissage) (2006) Detailed reference viewed: 17 (2 ULg)Kernelizing the output of tree-based methods Geurts, Pierre ; Wehenkel, Louis ; in Proceedings of the 23rd International Conference on Machine Learning (2006) We extend tree-based methods to the prediction of structured outputs using a kernelization of the algorithm that allows one to grow trees as soon as a kernel can be deﬁned on the output space. The ... [more ▼] We extend tree-based methods to the prediction of structured outputs using a kernelization of the algorithm that allows one to grow trees as soon as a kernel can be deﬁned on the output space. The resulting algorithm, called output kernel trees (OK3), generalizes classiﬁcation and regression trees as well as tree-based ensemble methods in a principled way. It inherits several features of these methods such as interpretability, robustness to irrelevant variables, and input scalability. When only the Gram matrix over the outputs of the learning sample is given, it learns the output kernel as a function of inputs. We show that the proposed algorithm works well on an image reconstruction task and on a biological network inference problem. [less ▲] Detailed reference viewed: 109 (15 ULg)A hybrid optimization technique coupling an evolutionary and a local search algorithm Kelner, Vincent ; Capitanescu, Florin ; Léonard, Olivier et al in Journal of Computational & Applied Mathematics (2006), 215(2), 281-287 Evolutionary algorithms are robust and powerful global optimization techniques for solving large-scale problems that have many local optima. However, they require high CPU times, and they are very poor in ... [more ▼] Evolutionary algorithms are robust and powerful global optimization techniques for solving large-scale problems that have many local optima. However, they require high CPU times, and they are very poor in terms of convergence performance. On the other hand, local search algorithms can converge in a few iterations but lack a global perspective. The combination of global and local search procedures should offer the advantages of both optimization methods while offsetting their disadvantages. This paper proposes a new hybrid optimization technique that merges a genetic algorithm with a local search strategy based on the interior point method. The efficiency of this hybrid approach is demonstrated by solving a constrained multi-objective mathematical test-case. [less ▲] Detailed reference viewed: 45 (9 ULg)Segment and combine: a generic approach for supervised learning of invariant classifiers from topologically structured data Geurts, Pierre ; Marée, Raphaël ; Wehenkel, Louis in Proceedings of the Machine Learning Conference of Belgium and The Netherlands (Benelearn) (2006) A generic method for supervised classification of structured objects is presented. The approach induces a classifier by (i) deriving a surrogate dataset from a pre-classified dataset of structured objects ... [more ▼] A generic method for supervised classification of structured objects is presented. The approach induces a classifier by (i) deriving a surrogate dataset from a pre-classified dataset of structured objects, by segmenting them into pieces, (ii) learning a model relating pieces to object-classes, (iii) classifying structured objects by combining predictions made for their pieces. The segmentation allows to exploit local information and can be adapted to inject invariances into the resulting classifier. The framework is illustrated on practical sequence, time-series and image classification problems. [less ▲] Detailed reference viewed: 162 (14 ULg)A Semi-Algebraic Description of Discrete Naive Bayes Models with Two Hidden Classes Auvray, Vincent ; Geurts, Pierre ; Wehenkel, Louis in Proc. Ninth International Symposium on Artificial Intelligence and Mathematics (2006) Detailed reference viewed: 21 (3 ULg)Experience with the multiple centrality corrections interior point algorithm for optimal power flow Capitanescu, Florin ; ; Wehenkel, Louis in CEE 05 conference (2005, October) This paper analyzes the ability of the Multiple Centrality Corrections (MCC) interior point algorithm to solve various classical optimal power flow (OPF) variants, namely: the minimization of generation ... [more ▼] This paper analyzes the ability of the Multiple Centrality Corrections (MCC) interior point algorithm to solve various classical optimal power flow (OPF) variants, namely: the minimization of generation cost, the minimization of active power losses, the maximization of power system loadability and the minimization of the amount of load curtailment. The performances of the MCC approach are assessed with respect to the predictor-corrector algorithm, which is widely recognized as the best interior point method based optimizer to date. Illustrative examples on three test systems up to 300 buses are provided. [less ▲] Detailed reference viewed: 27 (0 ULg)Applications of an interior point method based optimal power flow Capitanescu, Florin ; ; Wehenkel, Louis in CEE 05 conference (2005, October) This paper tackles the complex problem of an Optimal Power Flow (OPF) by the Interior Point Method (IPM). Two interior point algorithms are presented and compared, namely the pure primal-dual and the ... [more ▼] This paper tackles the complex problem of an Optimal Power Flow (OPF) by the Interior Point Method (IPM). Two interior point algorithms are presented and compared, namely the pure primal-dual and the predictor-corrector respectively. Among various OPF objectives, emphasis is put on two classical ones: the maximization of power system loadability and the minimization of the amount of load curtailment. Illustrative examples on three test systems of 60, 118 and 300 buses are provided. [less ▲] Detailed reference viewed: 26 (2 ULg)An interior point method based optimal power flow Capitanescu, Florin ; ; Wehenkel, Louis (2005, June) This paper deals with the solution of an optimal power flow (OPF) problem by the interior point method (IPM). The latter is a very appealing approach to this nonlinear programming problem due to its speed ... [more ▼] This paper deals with the solution of an optimal power flow (OPF) problem by the interior point method (IPM). The latter is a very appealing approach to this nonlinear programming problem due to its speed of convergence and ease of handling inequality constraints. Two interior point algorithms are presented and compared: the pure primal-dual and the predictor-corrector. Several implementation aspects of these IPM algorithms are also discussed. The OPF is formulated in rectangular coordinates which confers some significant advantages because generally its objective and constraints are quadratic functions. Among the large variety of OPF objectives, emphasis is put on two classical ones: the minimization of generation cost and the minimization of transmission active power losses. The solution obtained by both algorithms proves to be robust for the two OPF sub-problems (optimization of active power flows and reactive power flows) as well as for a full OPF applied to the former objective, which is unanimously recognized as the hardest problem to solve. Finally, numerical results on three test systems ranging from 60 to 300 buses are provided. [less ▲] Detailed reference viewed: 73 (0 ULg)A reinforcement learning based discrete supplementary control for power system transient stability enhancement ; Ernst, Damien ; Wehenkel, Louis in Engineering Intelligent Systems for Electrical Engineering and Communications (2005), 13(2 Sp. Iss. SI), 81-88 This paper proposes an application of a Reinforcement Learning (RL) method to the control of a dynamic brake aimed to enhance power system transient stability. The control law of the resistive brake is in ... [more ▼] This paper proposes an application of a Reinforcement Learning (RL) method to the control of a dynamic brake aimed to enhance power system transient stability. The control law of the resistive brake is in the form of switching strategies. In particular, the paper focuses on the application of a model based RL method, known as prioritized sweeping, a method proven to be suitable in applications in which computation is considered to be cheap. The curse of dimensionality problem is resolved by the system state dimensionality reduction based on the One Machine Infinite Bus (OMIB) transformation. Results obtained by using a synthetic four-machine power system are given to illustrate the performances of the proposed methodology. [less ▲] Detailed reference viewed: 54 (2 ULg)Tree-based batch mode reinforcement learning Ernst, Damien ; Geurts, Pierre ; Wehenkel, Louis in Journal of Machine Learning Research (2005), 6 Reinforcement learning aims to determine an optimal control policy from interaction with a system or from observations gathered from a system. In batch mode, it can be achieved by approximating the so ... [more ▼] Reinforcement learning aims to determine an optimal control policy from interaction with a system or from observations gathered from a system. In batch mode, it can be achieved by approximating the so-called Q-function based on a set of four-tuples (x(t), u(t), r(t), x(t+1)) where x(t) denotes the system state at time t, u(t) the control action taken, r(t) the instantaneous reward obtained and x(t+1) the successor state of the system, and by determining the control policy from this Q-function. The Q-function approximation may be obtained from the limit of a sequence of (batch mode) supervised learning problems. Within this framework we describe the use of several classical tree-based supervised learning methods (CART, Kd-tree, tree bagging) and two newly proposed ensemble algorithms, namely extremely and totally randomized trees. We study their performances on several examples and find that the ensemble methods based on regression trees perform well in extracting relevant information about the optimal control policy from sets of four-tuples. In particular, the totally randomized trees give good results while ensuring the convergence of the sequence, whereas by relaxing the convergence constraint even better accuracy results are provided by the extremely randomized trees. [less ▲] Detailed reference viewed: 469 (53 ULg)Combining a stability and a performance-oriented control in power systems ; Ernst, Damien ; Wehenkel, Louis in IEEE Transactions on Power Systems (2005), 20(1), 525-526 This paper suggests that the appropriate combination of a stability-oriented and a performance-oriented control technique is a promising way to implement advanced control schemes in power systems. The ... [more ▼] This paper suggests that the appropriate combination of a stability-oriented and a performance-oriented control technique is a promising way to implement advanced control schemes in power systems. The particular approach considered combines control Lyapunov functions (CLF) and reinforcement learning. The capabilities of the resulting controller are illustrated on a control problem involving a thyristor-controlled series capacitor (TCSC) device for damping oscillations in a four-machine power system. [less ▲] Detailed reference viewed: 26 (1 ULg) |
||