Catálogo de publicaciones - libros

The purpose of this paper is to present a typology of career paths in France with the Kohonen algorithm and its generalization to a clustering method of life history using Self Organizing Maps. Several methods have already been proposed to transform qualitative information into quantitative one such as being able to apply clustering algorithm based on the Euclidean distance such as SOM. In the case of life history, these methods generally ignore the longitudinal organization of the data. Our approach aims to deduce a quantitative encode from the labor market situation proximities across time. Using SOM, the topology preservation is also helpful to check when the new encoding keep particularities of the life history and our economic approach of careers. In final, this quantitative encoding can be easily generalized to a method of clustering life history and complete the set of methods generalizing the use of SOM to qualitative data.

Accurate forecasting of inter-urban traffic flow has been one of most important issues in the research on road traffic congestion. The traffic flow forecasting involves a rather complex nonlinear data pattern. Recently, support vector regression (SVR) model has been widely used to solve nonlinear regression and time series problems. This investigation presents a short-term traffic forecasting model which combines SVR model with continuous ant colony optimization (SVRCACO), to forecast inter-urban traffic flow. A numerical example of traffic flow values from northern Taiwan is employed to elucidate the forecasting performance of the proposed model. The simulation results indicate that the proposed model yields more accurate forecasting results than the seasonal autoregressive integrated moving average (SARIMA) time-series model.

In this paper we use Kohonen’s Self-Organizing Map (SOM) for surveying the financial status of Spanish companies. From it, we infer which are the most relevant variables, so that a fast diagnostic on their status can be reached and, besides, explained via a few rules of thumb extracted from the behavior of those variables. This map can be used as part of a decision making process, or as a first stage in an automatic classification tool. Results show that variables, identified in an easy and visual way (using SOM and U-Matrix graph), are in agreement with those obtained using parametric and non-parametric tests, which are more complex and difficult to apply.

Kernel methods are a class of algorithms whose importance has grown from the 90s in the machine learning field. Their most notable example are Support Vector Machines (SVMs), which are the state of the art for classification problems. Nevertheless, they are applicable to functional approximation problems and there are however several of them available: SVM for regression, Gaussian Process Regression and Least Squares SVM (LS-SVM) for instance. This paper applies and studies these algorithms to a number of Time Series Prediction problems and compares them with some more conventional techniques.

In previous woks we have introduced a distributed neural architecture for the generation of complex behaviors in evolutionary robotics. In this paper we show how this architecture is able to create its own categories about the sensed world of a robot by direct interaction of the body with the environment. The distributed elements of the architecture cooperate to express the categories on an inner world that is easily accessible from the outside. We conclude the paper with an explanation of how the inner world created by the robot can be used to gain some insight into the mind-body problem.

In this paper we suggest a biologically inspired approach to flexible behavior through emotion modeling. We consider emotion to emerge from relational interaction of body, nervous system and world, through sensory-motor attunement of internal parameters to concern-relevant relationships. We interpret such relationships with the notions of collective variable and control parameters. We introduce a simple robotic implementation of this model of appraisal, following the techniques of evolutionary neuro-robotics.

A Policy Gradient Reinforcement Learning (RL) technique is used to design the low level controllers that drives the joints of articulated mobile robots: A search in the controller’s parameters space. There is an unknown value function that measures the quality of the controller respect to the parameters of it. The search is orientated by the approximation of the gradient of the value function. The approximation is made by means of the robot experiences and then the behaviors emerge. This technique is employed in a structure that processes sensor information to achieve coordination. The structure is based on a modularization principle in which complex overall behavior is the result of the interaction of individual ‘simple’ components. The simple components used are standard low level controllers (PID) which output is combined, sharing information between articulations and therefore taking integrated control actions. Modularization and Learning are cognitive features, here we endow the robots with this features. Learning experiences in simulated robots are presented as demonstration.

This paper presents a new approach to shared control. It consists of combining orders from a mobile and a human. The weight of these orders is obtained by evaluating their corresponding efficiencies but also of his/her condition, which is estimated by continuously monitoring attached biosensors. We rely on a hierarchical architecture where the reactive layer provides a simple and adaptive combination of these sources. We have evaluated the resulting emergent behaviour for different tasks to measure their efficiencies under different circumstances. The system has been successfully tested in real environments from a quantitative and a qualitative point of view.

This paper presents a bio-inspired control model for humanoid robots manipulating objects. Humanoids face several genuine problems: 1) they are not fixed (to the ground) therefore extreme forces generate noisy vibrations on the whole platform (robot body) and 2) rigid control (to avoid dynamic modelling) requires high power to be accurate and dramatically limits their autonomy. We compare a velocity vs a position driven control scheme in the framework of object manipulation. The velocity driven control scheme helps smoother control (reducing the jerks). Furthermore, we use an artificial neural network (RBF) to extract some features of the dynamic model automatically complementing the control scheme. Its performance is evaluated using a real robot platform. Experiments were done using the robot’s arm and trajectory data was collected during different trials manipulating different objects in order to acquire the model and evaluate how to use it to improve control accuracy.

A neural architecture that makes possible the integration of a kinematic adaptive neuro-controller for trajectory tracking and an obstacle avoidance adaptive neuro-controller is proposed for nonholonomic mobile robots. The kinematic adaptive neuro-controller is a real-time, unsupervised neural network that learns to control a nonholonomic mobile robot in a nonstationary environment, which is termed Self-Organization Direction Mapping Network (SODMN), and combines associative learning and Vector Associative Map (VAM) learning to generate transformations between spatial and velocity coordinates. The transformations are learned in an unsupervised training phase, during which the robot moves as a result of randomly selected wheel velocities. The robot learns the relationship between these velocities and the resulting incremental movements. The obstacle avoidance adaptive neuro-controller is a neural network that learns to control avoidance behaviors in a mobile robot based on a form of animal learning known as operant conditioning. Learning, which requires no supervision, takes place as the robot moves around an cluttered environment with obstacles. The neural network requires no knowledge of the geometry of the robot or of the quality, number, or configuration of the robot’s sensors. The efficacy of the proposed neural architecture is tested experimentally by a differentially driven mobile robot.