Louis Wehenkel

Extremely randomized trees

This paper proposes a new tree-based ensemble method for supervised classification and regression problems. It essentially consists of randomizing strongly both attribute and cut-point choice while splitting a tree node. In the extreme case, it builds totally randomized trees whose structures are independent of the output values of the learning sample. The strength of the randomization can be tuned to problem specifics by the appropriate choice of a parameter. We evaluate the robustness of the default choice of this parameter, and we also provide insight on how to adjust it in particular situations. Besides accuracy, the main strength of the resulting algorithm is computational efficiency. A bias/variance analysis of the Extra-Trees algorithm is also provided as well as a geometrical and a kernel characterization of the models induced.

A complete fuzzy decision tree technique

In this paper, a new method of fuzzy decision trees called soft decision trees (SDT) is presented. This method combines tree growing and pruning, to determine the structure of the soft decision tree, with refitting and backfitting, to improve its generalization capabilities. The method is explained and motivated and its behavior is first analyzed empirically on 3 large databases in terms of classification error rate, model complexity and CPU time. A comparative study on 11 standard UCI Repository databases then shows that the soft decision trees produced by this method are significantly more accurate than standard decision trees. Moreover, a global model variance study shows a much lower variance for soft decision trees than for standard trees as a direct cause of the improved accuracy.

Random subwindows for robust image classification

We present a novel, generic image classification method based on a recent machine learning algorithm (ensembles of extremely randomized decision trees). Images are classified using randomly extracted subwindows that are suitably normalized to yield robustness to certain image transformations. Our method is evaluated on four very different, publicly available datasets (COIL-100, ZuBuD, ETH-80, WANG). Our results show that our automatic approach is generic and robust to illumination, scale, and viewpoint changes. An extension of the method is proposed to improve its robustness with respect to rotation changes.

An Artificial Intelligence Framework for On-Line Transient Stability Assessment of Power Systems

Transient stability assessment (TSA) of a power system pursues a twofold objective: first to appraise the systems capability to withstand major contingencies, and second to suggest remedial actions, i.e. means to enhance this capability, whenever needed. The first objective is the concern of analysis, the second is a matter of control. For the time being, the on-line TSA is still a totally open question. Indeed, none of the existing two broad classes of methods (the time domain and the direct methods) are able to meet the on-line requirements of the analysis aspects, nor are they in the least appropriate to tackle control aspects. The methodology we are introducing aims at solving the above stated on-line problem by making use of decision rules, preconstructed off-line. To this end, an inductive inference method is developed, able to provide decision rules in the form of binary trees expressing relationships between static, pre-fault operating conditions of a power system and its robustness to withstand assumed disturbances. This paper concentrates on this latter problem, which is the most difficult task, and also the kernel of the overall methodology. The proposed inductive inference (II) method pertains to a particular family of Machine Learning from examples. It derives from ID3 by Quinlan [1], tailored to our problem, where the examples are provided by numeric (load flow and stability) programs [2, 3]. According to the method, a decision tree (DT) is built on the basis of a preanalyzed learning set (LS), composed of states or operating points (OPs).

Proteomic mass spectra classification using decision tree based ensemble methods

MOTIVATION Modern mass spectrometry allows the determination of proteomic fingerprints of body fluids like serum, saliva or urine. These measurements can be used in many medical applications in order to diagnose the current state or predict the evolution of a disease. Recent developments in machine learning allow one to exploit such datasets, characterized by small numbers of very high-dimensional samples. RESULTS We propose a systematic approach based on decision tree ensemble methods, which is used to automatically determine proteomic biomarkers and predictive models. The approach is validated on two datasets of surface-enhanced laser desorption/ionization time of flight measurements, for the diagnosis of rheumatoid arthritis and inflammatory bowel diseases. The results suggest that the methodology can handle a broad class of similar problems.

Power systems stability control: reinforcement learning framework

In this paper, we explore how a computational approach to learning from interactions, called reinforcement learning (RL), can be applied to control power systems. We describe some challenges in power system control and discuss how some of those challenges could be met by using these RL methods. The difficulties associated with their application to control power systems are described and discussed as well as strategies that can be adopted to overcome them. Two reinforcement learning modes are considered: the online mode in which the interaction occurs with the real power system and the offline mode in which the interaction occurs with a simulation model of the real power system. We present two case studies made on a four-machine power system model. The first one concerns the design by means of RL algorithms used in offline mode of a dynamic brake controller. The second concerns RL methods used in online mode when applied to control a thyristor controlled series capacitor (TCSC) aimed to damp power system oscillations.

SIME: A hybrid approach to fast transient stability assessment and contingency selection

Abstract We propose an integrated scheme for transient stability assessment which in a sequence screens contingencies and scrutinizes only the selected ones. This scheme is based on a hybrid method, called SIME for SIngle Machine Equivalent. SIME relies on a particular direct method coupled with time-domain programs so as to combine the strengths of both, namely: the flexibility with respect to power system modelling of time-domain methods; the speed and richer information of the direct method. This paper lays the foundations of SIME, devises appropriate techniques for transient stability assessment per se and for contingency screening, and finally integrates these two techniques in a fully general function, i.e. able to comply with any power system modelling and stability scenario, and to assess any type of stability limits (critical clearing times or power limits). Throughout, real-world examples illustrate the proposed techniques and highlight their performances.

Contingency Filtering Techniques for Preventive Security-Constrained Optimal Power Flow

This paper focuses on contingency filtering to accelerate the iterative solution of preventive security-constrained optimal power flow (PSCOPF) problems. To this end, we propose two novel filtering techniques relying on the comparison at an intermediate PSCOPF solution of post-contingency constraint violations among postulated contingencies. We assess these techniques by comparing them with severity index-based filtering schemes, on a 60-and a 118-bus system. Our results show that the proposed contingency filtering techniques lead to faster solution of the PSCOPF, while being more robust and meaningful, than severity-index based ones.

A New Iterative Approach to the Corrective Security-Constrained Optimal Power Flow Problem

This paper deals with techniques to solve the corrective security-constrained optimal power flow (CSCOPF) problem. To this end, we propose a new iterative approach that comprises four modules: a CSCOPF which considers only a subset of potentially binding contingencies among the postulated contingencies, a (steady-state) security analysis (SSSA), a contingency filtering (CF) technique, and an OPF variant to check post-contingency state feasibility when taking into account post-contingency corrective actions. We compare performances of our approach and its possible variants with classical CSCOPF approaches such as the direct approach and Benders decomposition (BD), on three systems of 60, 118, and 1203 buses.

Machine learning approaches to power-system security assessment

The paper discusses a framework that uses machine learning and other automatic-learning methods to assess power-system security. The framework exploits simulation models in parallel to screen diverse simulation scenarios of a system, yielding a large database. Using data mining techniques, the framework extracts synthetic information about the simulated systems main features from this database.