Co-evolutionary and Reinforcement Learning Techniques Applied to Computer Go players

The objective of this thesis is model some processes from the nature as evolution and co-evolution, and proposing some techniques that can ensure that these learning process really happens and useful to solve some complex problems as Go game. The Go game is ancient and very complex game with simple rules which still is a challenge for the Artificial Intelligence. This dissertation cover some approaches that were applied to solve this problem, proposing solve this problem using competitive and cooperative co-evolutionary learning methods and other techniques proposed by the author. To study, implement and prove these methods were used some neural networks structures, a framework free available and coded many programs. The techniques proposed were coded by the author, performed many experiments to find the best configuration to ensure that co-evolution is progressing and discussed the results. Using co-evolutionary learning processes can be observed some pathologies which could impact co-evolution progress. In this dissertation is introduced some techniques to solve pathologies as loss of gradients, cycling dynamics and forgetting. According to some authors, one solution to solve these co-evolution pathologies is introduce more diversity in populations that are evolving. In this thesis is proposed some techniques to introduce more diversity and some diversity measurements for neural networks structures to monitor diversity during co-evolution. The genotype diversity evolved were analyzed in terms of its impact to global fitness of the strategies evolved and their generalization. Additionally, it was introduced a memory mechanism in the network neural structures to reinforce some strategies in the genes of the neurons evolved with the intention that some good strategies learned are not forgotten. In this dissertation is presented some works from other authors in which cooperative and competitive co-evolution has been applied. The Go board size used in this thesis was 9x9, but can be easily escalated to more bigger boards.The author believe that programs coded and techniques introduced in this dissertation can be used for other domains.

Diffusion Gradient Temporal Difference for Cooperative Reinforcement Learning with Linear Function Approximation

We introduce a diffusion-based algorithm in which multiple agents cooperate to predict a common and global statevalue function by sharing local estimates and local gradient information among neighbors. Our algorithm is a fully distributed implementation of the gradient temporal difference with linear function approximation, to make it applicable to multiagent settings. Simulations illustrate the benefit of cooperation in learning, as made possible by the proposed algorithm.

Linear Bayes policy for learning in contextual-bandits

Machine and Statistical Learning techniques are used in almost all online advertisement systems. The problem of discovering which content is more demanded (e.g. receive more clicks) can be modeled as a multi-armed bandit problem. Contextual bandits (i.e., bandits with covariates, side information or associative reinforcement learning) associate, to each specific content, several features that define the “context” in which it appears (e.g. user, web page, time, region). This problem can be studied in the stochastic/statistical setting by means of the conditional probability paradigm using the Bayes’ theorem. However, for very large contextual information and/or real-time constraints, the exact calculation of the Bayes’ rule is computationally infeasible. In this article, we present a method that is able to handle large contextual information for learning in contextual-bandits problems. This method was tested in the Challenge on Yahoo! dataset at ICML2012’s Workshop “new Challenges for Exploration & Exploitation 3”, obtaining the second place. Its basic exploration policy is deterministic in the sense that for the same input data (as a time-series) the same results are obtained. We address the deterministic exploration vs. exploitation issue, explaining the way in which the proposed method deterministically finds an effective dynamic trade-off based solely in the input-data, in contrast to other methods that use a random number generator.

Evolving and coevolving computer go players using neuroevolution.

The Go game is ancient very complex game with simple rules which still is a challenge for the AI.This work cover some neuroevolution techniques used in reinforcement learning applied to the GO game as SANE (Symbiotic Adaptive Neuro-Evolution) and presents a variation to this method with the intention of evolving better strategies in the game. The computer Go player based in SANE is evolved againts a knowed player which creates some problem as determinism for which is proposed the co-evolution. Finally, it is introduced an algorithm to co-evolve two populations of neurons to evolve better computer Go players.

Arquitectura de Control para la Conducción Autónoma de Vehículos en Entornos Urbanos y Autovías

Hoy en día, el desarrollo tecnológico en el campo de los sistemas inteligentes de transporte (ITS por sus siglas en inglés) ha permitido dotar a los vehículos con diversos sistemas de ayuda a la conducción (ADAS, del inglés advanced driver assistance system), mejorando la experiencia y seguridad de los pasajeros, en especial del conductor. La mayor parte de estos sistemas están pensados para advertir al conductor sobre ciertas situaciones de riesgo, como la salida involuntaria del carril o la proximidad de obstáculos en el camino. No obstante, también podemos encontrar sistemas que van un paso más allá y son capaces de cooperar con el conductor en el control del vehículo o incluso relegarlos de algunas tareas tediosas. Es en este último grupo donde se encuentran los sistemas de control electrónico de estabilidad (ESP - Electronic Stability Program), el antibloqueo de frenos (ABS - Anti-lock Braking System), el control de crucero (CC - Cruise Control) y los más recientes sistemas de aparcamiento asistido. Continuando con esta línea de desarrollo, el paso siguiente consiste en la supresión del conductor humano, desarrollando sistemas que sean capaces de conducir un vehículo de forma autónoma y con un rendimiento superior al del conductor. En este trabajo se presenta, en primer lugar, una arquitectura de control para la automatización de vehículos. Esta se compone de distintos componentes de hardware y software, agrupados de acuerdo a su función principal. El diseño de la arquitectura parte del trabajo previo desarrollado por el Programa AUTOPIA, aunque introduce notables aportaciones en cuanto a la eficiencia, robustez y escalabilidad del sistema. Ahondando un poco más en detalle, debemos resaltar el desarrollo de un algoritmo de localización basado en enjambres de partículas. Este está planteado como un método de filtrado y fusión de la información obtenida a partir de los distintos sensores embarcados en el vehículo, entre los que encontramos un receptor GPS (Global Positioning System), unidades de medición inercial (IMU – Inertial Measurement Unit) e información tomada directamente de los sensores embarcados por el fabricante, como la velocidad de las ruedas y posición del volante. Gracias a este método se ha conseguido resolver el problema de la localización, indispensable para el desarrollo de sistemas de conducción autónoma. Continuando con el trabajo de investigación, se ha estudiado la viabilidad de la aplicación de técnicas de aprendizaje y adaptación al diseño de controladores para el vehículo. Como punto de partida se emplea el método de Q-learning para la generación de un controlador borroso lateral sin ningún tipo de conocimiento previo. Posteriormente se presenta un método de ajuste on-line para la adaptación del control longitudinal ante perturbaciones impredecibles del entorno, como lo son los cambios en la inclinación del camino, fricción de las ruedas o peso de los ocupantes. Para finalizar, se presentan los resultados obtenidos durante un experimento de conducción autónoma en carreteras reales, el cual se llevó a cabo en el mes de Junio de 2012 desde la población de San Lorenzo de El Escorial hasta las instalaciones del Centro de Automática y Robótica (CAR) en Arganda del Rey. El principal objetivo tras esta demostración fue validar el funcionamiento, robustez y capacidad de la arquitectura propuesta para afrontar el problema de la conducción autónoma, bajo condiciones mucho más reales a las que se pueden alcanzar en las instalaciones de prueba. ABSTRACT Nowadays, the technological advances in the Intelligent Transportation Systems (ITS) field have led the development of several driving assistance systems (ADAS). These solutions are designed to improve the experience and security of all the passengers, especially the driver. For most of these systems, the main goal is to warn drivers about unexpected circumstances leading to risk situations such as involuntary lane departure or proximity to other vehicles. However, other ADAS go a step further, being able to cooperate with the driver in the control of the vehicle, or even overriding it on some tasks. Examples of this kind of systems are the anti-lock braking system (ABS), cruise control (CC) and the recently commercialised assisted parking systems. Within this research line, the next step is the development of systems able to replace the human drivers, improving the control and therefore, the safety and reliability of the vehicles. First of all, this dissertation presents a control architecture design for autonomous driving. It is made up of several hardware and software components, grouped according to their main function. The design of this architecture is based on the previous works carried out by the AUTOPIA Program, although notable improvements have been made regarding the efficiency, robustness and scalability of the system. It is also remarkable the work made on the development of a location algorithm for vehicles. The proposal is based on the emulation of the behaviour of biological swarms and its performance is similar to the well-known particle filters. The developed method combines information obtained from different sensors, including GPS, inertial measurement unit (IMU), and data from the original vehicle’s sensors on-board. Through this filtering algorithm the localization problem is properly managed, which is critical for the development of autonomous driving systems. The work deals also with the fuzzy control tuning system, a very time consuming task when done manually. An analysis of learning and adaptation techniques for the development of different controllers has been made. First, the Q-learning –a reinforcement learning method– has been applied to the generation of a lateral fuzzy controller from scratch. Subsequently, the development of an adaptation method for longitudinal control is presented. With this proposal, a final cruise control controller is able to deal with unpredictable environment disturbances, such as road slope, wheel’s friction or even occupants’ weight. As a testbed for the system, an autonomous driving experiment on real roads is presented. This experiment was carried out on June 2012, driving from San Lorenzo de El Escorial up to the Center for Automation and Robotics (CAR) facilities in Arganda del Rey. The main goal of the demonstration was validating the performance, robustness and viability of the proposed architecture to deal with the problem of autonomous driving under more demanding conditions than those achieved on closed test tracks.

Cooperative off-policy prediction of markov decision processes in adaptive networks

We apply diffusion strategies to propose a cooperative reinforcement learning algorithm, in which agents in a network communicate with their neighbors to improve predictions about their environment. The algorithm is suitable to learn off-policy even in large state spaces. We provide a mean-square-error performance analysis under constant step-sizes. The gain of cooperation in the form of more stability and less bias and variance in the prediction error, is illustrated in the context of a classical model. We show that the improvement in performance is especially significant when the behavior policy of the agents is different from the target policy under evaluation.

Auto-Emergencia de Comunicación Sintáctica en Entornos Estáticos y Dinámicos para Grupos de Robots mediante Evolución Gramatical y Aprendizaje por Refuerzo

La idea de dotar a un grupo de robots o agentes artificiales de un lenguaje ha sido objeto de intenso estudio en las ultimas décadas. Como no podía ser de otra forma los primeros intentos se enfocaron hacia el estudio de la emergencia de vocabularios compartidos convencionalmente por el grupo de robots. Las ventajas que puede ofrecer un léxico común son evidentes, como también lo es que un lenguaje con una estructura más compleja, en la que se pudieran combinar palabras, sería todavía más beneficioso. Surgen así algunas propuestas enfocadas hacia la emergencia de un lenguaje consensuado que muestre una estructura sintáctica similar al lenguaje humano, entre las que se encuentra este trabajo. Tomar el lenguaje humano como modelo supone adoptar algunas de las hipótesis y teorías que disciplinas como la filosofía, la psicología o la lingüística entre otras se han encargado de proponer. Según estas aproximaciones teóricas el lenguaje presenta una doble dimension formal y funcional. En base a su dimensión formal parece claro que el lenguaje sigue unas reglas, por lo que el uso de una gramática se ha considerado esencial para su representación, pero también porque las gramáticas son un dispositivo muy sencillo y potente que permite generar fácilmente estructuras simbólicas. En cuanto a la dimension funcional se ha tenido en cuenta la teoría quizá más influyente de los últimos tiempos, que no es otra que la Teoría de los Actos del Habla. Esta teoría se basa en la idea de Wittgenstein por la que el significado reside en el uso del lenguaje, hasta el punto de que éste se entiende como una manera de actuar y de comportarse, en definitiva como una forma de vida. Teniendo presentes estas premisas en esta tesis se pretende experimentar con modelos computacionales que permitan a un grupo de robots alcanzar un lenguaje común de manera autónoma, simplemente mediante interacciones individuales entre los robots, en forma de juegos de lenguaje. Para ello se proponen tres modelos distintos de lenguaje: • Un modelo basado en gramáticas probabilísticas y aprendizaje por refuerzo en el que las interacciones y el uso del lenguaje son claves para su emergencia y que emplea una gramática generativa estática y diseñada de antemano. Este modelo se aplica a dos grupos distintos: uno formado exclusivamente por robots y otro que combina robots y un humano, de manera que en este segundo caso se plantea un aprendizaje supervisado por humanos. • Un modelo basado en evolución gramatical que permite estudiar no solo el consenso sintáctico, sino también cuestiones relativas a la génesis del lenguaje y que emplea una gramática universal a partir de la cual los robots pueden evolucionar por sí mismos la gramática más apropiada según la situación lingüística que traten en cada momento. • Un modelo basado en evolución gramatical y aprendizaje por refuerzo que toma aspectos de los anteriores y amplia las posibilidades de los robots al permitir desarrollar un lenguaje que se adapta a situaciones lingüísticas dinámicas que pueden cambiar en el tiempo y también posibilita la imposición de restricciones de orden muy frecuentes en las estructuras sintácticas complejas. Todos los modelos implican un planteamiento descentralizado y auto-organizado, de manera que ninguno de los robots es el dueño del lenguaje y todos deben cooperar y colaborar de forma coordinada para lograr el consenso sintáctico. En cada caso se plantean experimentos que tienen como objetivo validar los modelos propuestos, tanto en lo relativo al éxito en la emergencia del lenguaje como en lo relacionado con cuestiones paralelas de importancia, como la interacción hombre-máquina o la propia génesis del lenguaje. ABSTRACT The idea of giving a language to a group of robots or artificial agents has been the subject of intense study in recent decades. The first attempts have focused on the development and emergence of a conventionally shared vocabulary. The advantages that can provide a common vocabulary are evident and therefore a more complex language that combines words would be even more beneficial. Thus some proposals are put forward towards the emergence of a consensual language with a sintactical structure in similar terms to the human language. This work follows this trend. Taking the human language as a model means taking some of the assumptions and theories that disciplines such as philosophy, psychology or linguistics among others have provided. According to these theoretical positions language has a double formal and functional dimension. Based on its formal dimension it seems clear that language follows rules, so that the use of a grammar has been considered essential for representation, but also because grammars are a very simple and powerful device that easily generates these symbolic structures. As for the functional dimension perhaps the most influential theory of recent times, the Theory of Speech Acts has been taken into account. This theory is based on the Wittgenstein’s idea about that the meaning lies in the use of language, to the extent that it is understood as a way of acting and behaving. Having into account these issues this work implements some computational models in order to test if they allow a group of robots to reach in an autonomous way a shared language by means of individual interaction among them, that is by means of language games. Specifically, three different models of language for robots are proposed: • A reinforcement learning based model in which interactions and language use are key to its emergence. This model uses a static probabilistic generative grammar which is designed beforehand. The model is applied to two different groups: one formed exclusively by robots and other combining robots and a human. Therefore, in the second case the learning process is supervised by the human. • A model based on grammatical evolution that allows us to study not only the syntactic consensus, but also the very genesis of language. This model uses a universal grammar that allows robots to evolve for themselves the most appropriate grammar according to the current linguistic situation they deal with. • A model based on grammatical evolution and reinforcement learning that takes aspects of the previous models and increases their possibilities. This model allows robots to develop a language in order to adapt to dynamic language situations that can change over time and also allows the imposition of syntactical order restrictions which are very common in complex syntactic structures. All models involve a decentralized and self-organized approach so that none of the robots is the language’s owner and everyone must cooperate and work together in a coordinated manner to achieve syntactic consensus. In each case experiments are presented in order to validate the proposed models, both in terms of success about the emergence of language and it relates to the study of important parallel issues, such as human-computer interaction or the very genesis of language.

Fusion of probabilistic knowledge-based classification rules and learning automata for automatic recognition of digital images

In this paper, the fusion of probabilistic knowledge-based classification rules and learning automata theory is proposed and as a result we present a set of probabilistic classification rules with self-learning capability. The probabilities of the classification rules change dynamically guided by a supervised reinforcement process aimed at obtaining an optimum classification accuracy. This novel classifier is applied to the automatic recognition of digital images corresponding to visual landmarks for the autonomous navigation of an unmanned aerial vehicle (UAV) developed by the authors. The classification accuracy of the proposed classifier and its comparison with well-established pattern recognition methods is finally reported.

The online case study development for the teaching of building structures reinforcement

The Reinforcement of Building Structures is one of the topics of the Master in Building Innovation Technology (MBIT) of Universidad Politécnica de Madrid (UPM). Since the beginning of the delivery of this master, case studies have been chosen as the teaching methodology. For the 2011-2012 course the online education of this subject was implemented, instead of the classical learning based on attendance. Through ICT’s (Information and Communication Technologies) students are provided with much more and more selective information than through the classical learning. ICT’s can be used for search, enquiries and reporting. Using the online tools has been proved, through the results obtained and based on the surveys made amongst students, to be a successful experience.