Particle competition and cooperation for semi-supervised learning with label noise

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Semi-supervised learning guided by the modularity measure in complex networks

Semi-supervised learning techniques have gained increasing attention in the machine learning community, as a result of two main factors: (1) the available data is exponentially increasing; (2) the task of data labeling is cumbersome and expensive, involving human experts in the process. In this paper, we propose a network-based semi-supervised learning method inspired by the modularity greedy algorithm, which was originally applied for unsupervised learning. Changes have been made in the process of modularity maximization in a way to adapt the model to propagate labels throughout the network. Furthermore, a network reduction technique is introduced, as well as an extensive analysis of its impact on the network. Computer simulations are performed for artificial and real-world databases, providing a numerical quantitative basis for the performance of the proposed method.

Combining meta-learning and search techniques to select parameters for support vector machines

Support Vector Machines (SVMs) have achieved very good performance on different learning problems. However, the success of SVMs depends on the adequate choice of the values of a number of parameters (e.g., the kernel and regularization parameters). In the current work, we propose the combination of meta-learning and search algorithms to deal with the problem of SVM parameter selection. In this combination, given a new problem to be solved, meta-learning is employed to recommend SVM parameter values based on parameter configurations that have been successfully adopted in previous similar problems. The parameter values returned by meta-learning are then used as initial search points by a search technique, which will further explore the parameter space. In this proposal, we envisioned that the initial solutions provided by meta-learning are located in good regions of the search space (i.e. they are closer to optimum solutions). Hence, the search algorithm would need to evaluate a lower number of candidate solutions when looking for an adequate solution. In this work, we investigate the combination of meta-learning with two search algorithms: Particle Swarm Optimization and Tabu Search. The implemented hybrid algorithms were used to select the values of two SVM parameters in the regression domain. These combinations were compared with the use of the search algorithms without meta-learning. The experimental results on a set of 40 regression problems showed that, on average, the proposed hybrid methods obtained lower error rates when compared to their components applied in isolation.

Stochastic competitive learning in complex networks

Competitive learning is an important machine learning approach which is widely employed in artificial neural networks. In this paper, we present a rigorous definition of a new type of competitive learning scheme realized on large-scale networks. The model consists of several particles walking within the network and competing with each other to occupy as many nodes as possible, while attempting to reject intruder particles. The particle's walking rule is composed of a stochastic combination of random and preferential movements. The model has been applied to solve community detection and data clustering problems. Computer simulations reveal that the proposed technique presents high precision of community and cluster detections, as well as low computational complexity. Moreover, we have developed an efficient method for estimating the most likely number of clusters by using an evaluator index that monitors the information generated by the competition process itself. We hope this paper will provide an alternative way to the study of competitive learning.

Aktionenlernen mit Selbstorganisierenden Karten und Reinforcement Learning

Die vorliegende Arbeit beschäftigt sich mit der Entwicklung eines Funktionsapproximators und dessen Verwendung in Verfahren zum Lernen von diskreten und kontinuierlichen Aktionen: 1. Ein allgemeiner Funktionsapproximator – Locally Weighted Interpolating Growing Neural Gas (LWIGNG) – wird auf Basis eines Wachsenden Neuralen Gases (GNG) entwickelt. Die topologische Nachbarschaft in der Neuronenstruktur wird verwendet, um zwischen benachbarten Neuronen zu interpolieren und durch lokale Gewichtung die Approximation zu berechnen. Die Leistungsfähigkeit des Ansatzes, insbesondere in Hinsicht auf sich verändernde Zielfunktionen und sich verändernde Eingabeverteilungen, wird in verschiedenen Experimenten unter Beweis gestellt. 2. Zum Lernen diskreter Aktionen wird das LWIGNG-Verfahren mit Q-Learning zur Q-LWIGNG-Methode verbunden. Dafür muss der zugrunde liegende GNG-Algorithmus abgeändert werden, da die Eingabedaten beim Aktionenlernen eine bestimmte Reihenfolge haben. Q-LWIGNG erzielt sehr gute Ergebnisse beim Stabbalance- und beim Mountain-Car-Problem und gute Ergebnisse beim Acrobot-Problem. 3. Zum Lernen kontinuierlicher Aktionen wird ein REINFORCE-Algorithmus mit LWIGNG zur ReinforceGNG-Methode verbunden. Dabei wird eine Actor-Critic-Architektur eingesetzt, um aus zeitverzögerten Belohnungen zu lernen. LWIGNG approximiert sowohl die Zustands-Wertefunktion als auch die Politik, die in Form von situationsabhängigen Parametern einer Normalverteilung repräsentiert wird. ReinforceGNG wird erfolgreich zum Lernen von Bewegungen für einen simulierten 2-rädrigen Roboter eingesetzt, der einen rollenden Ball unter bestimmten Bedingungen abfangen soll.

Certificates for Incremental Type Checking

The central topic of this thesis is the study of algorithms for type checking, both from the programming language and from the proof-theoretic point of view. A type checking algorithm takes a program or a proof, represented as a syntactical object, and checks its validity with respect to a specification or a statement. It is a central piece of compilers and proof assistants. We postulate that since type checkers are at the interface between proof theory and program theory, their study can let these two fields mutually enrich each other. We argue by two main instances: first, starting from the problem of proof reuse, we develop an incremental type checker; secondly, starting from a type checking program, we evidence a novel correspondence between natural deduction and the sequent calculus.

Spatio-temporal credit assignment in population learning

Learning by reinforcement is important in shaping animal behavior, and in particular in behavioral decision making. Such decision making is likely to involve the integration of many synaptic events in space and time. However, using a single reinforcement signal to modulate synaptic plasticity, as suggested in classical reinforcement learning algorithms, a twofold problem arises. Different synapses will have contributed differently to the behavioral decision, and even for one and the same synapse, releases at different times may have had different effects. Here we present a plasticity rule which solves this spatio-temporal credit assignment problem in a population of spiking neurons. The learning rule is spike-time dependent and maximizes the expected reward by following its stochastic gradient. Synaptic plasticity is modulated not only by the reward, but also by a population feedback signal. While this additional signal solves the spatial component of the problem, the temporal one is solved by means of synaptic eligibility traces. In contrast to temporal difference (TD) based approaches to reinforcement learning, our rule is explicit with regard to the assumed biophysical mechanisms. Neurotransmitter concentrations determine plasticity and learning occurs fully online. Further, it works even if the task to be learned is non-Markovian, i.e. when reinforcement is not determined by the current state of the system but may also depend on past events. The performance of the model is assessed by studying three non-Markovian tasks. In the first task, the reward is delayed beyond the last action with non-related stimuli and actions appearing in between. The second task involves an action sequence which is itself extended in time and reward is only delivered at the last action, as it is the case in any type of board-game. The third task is the inspection game that has been studied in neuroeconomics, where an inspector tries to prevent a worker from shirking. Applying our algorithm to this game yields a learning behavior which is consistent with behavioral data from humans and monkeys, revealing themselves properties of a mixed Nash equilibrium. The examples show that our neuronal implementation of reward based learning copes with delayed and stochastic reward delivery, and also with the learning of mixed strategies in two-opponent games.

Towards Self-Learning Radio-Based Localization Systems

Location-awareness indoors will be an inseparable feature of mobile services/applications in future wireless networks. Its current ubiquitous availability is still obstructed by technological challenges and privacy issues. We propose an innovative approach towards the concept of indoor positioning with main goal to develop a system that is self-learning and able to adapt to various radio propagation environments. The approach combines estimation of propagation conditions, subsequent appropriate channel modelling and optimisation feedback to the used positioning algorithm. Main advantages of the proposal are decreased system set-up effort, automatic re-calibration and increased precision.

Effects of instructional changes on student learning of electrochemistry in an IB chemistry course

This study investigated the effectiveness of incorporating several new instructional strategies into an International Baccalaureate (IB) chemistry course in terms of how they supported high school seniors’ understanding of electrochemistry. The three new methods used were (a) providing opportunities for visualization of particle movement by student manipulation of physical models and interactive computer simulations, (b) explicitly addressing common misconceptions identified in the literature, and (c) teaching an algorithmic, step-wise approach for determining the products of an aqueous solution electrolysis. Changes in student understanding were assessed through test scores on both internally and externally administered exams over a two-year period. It was found that visualization practice and explicit misconception instruction improved student understanding, but the effect was more apparent in the short-term. The data suggested that instruction time spent on algorithm practice was insufficient to cause significant test score improvement. There was, however, a substantial increase in the percentage of the experimental group students who chose to answer an optional electrochemistry-related external exam question, indicating an increase in student confidence. Implications for future instruction are discussed.

Improving Pediatric Basic Life Support Performance Through Blended Learning With Web-Based Virtual Patients: Randomized Controlled Trial

BACKGROUND E-learning and blended learning approaches gain more and more popularity in emergency medicine curricula. So far, little data is available on the impact of such approaches on procedural learning and skill acquisition and their comparison with traditional approaches. OBJECTIVE This study investigated the impact of a blended learning approach, including Web-based virtual patients (VPs) and standard pediatric basic life support (PBLS) training, on procedural knowledge, objective performance, and self-assessment. METHODS A total of 57 medical students were randomly assigned to an intervention group (n=30) and a control group (n=27). Both groups received paper handouts in preparation of simulation-based PBLS training. The intervention group additionally completed two Web-based VPs with embedded video clips. Measurements were taken at randomization (t0), after the preparation period (t1), and after hands-on training (t2). Clinical decision-making skills and procedural knowledge were assessed at t0 and t1. PBLS performance was scored regarding adherence to the correct algorithm, conformance to temporal demands, and the quality of procedural steps at t1 and t2. Participants' self-assessments were recorded in all three measurements. RESULTS Procedural knowledge of the intervention group was significantly superior to that of the control group at t1. At t2, the intervention group showed significantly better adherence to the algorithm and temporal demands, and better procedural quality of PBLS in objective measures than did the control group. These aspects differed between the groups even at t1 (after VPs, prior to practical training). Self-assessments differed significantly only at t1 in favor of the intervention group. CONCLUSIONS Training with VPs combined with hands-on training improves PBLS performance as judged by objective measures.

Arquitectura cognitiva basada en el gradiente sensorial y su aplicación a la robótica móvil

El objetivo de esta tesis es el desarrollo de un sistema completo de navegación, aprendizaje y planificación para un robot móvil. Dentro de los innumerables problemas que este gran objetivo plantea, hemos dedicado especial atención al problema del conocimiento autónomo del mundo. Nuestra mayor preocupación ha sido la de establecer mecanismos que permitan, a partir de información sensorial cruda, el desarrollo incremental de un modelo topológico del entorno en el que se mueve el robot. Estos mecanismos se apoyan invariablemente en un nuevo concepto propuesto en esta tesis: el gradiente sensorial. El gradiente sensorial es un dispositivo matemático que funciona como un detector de sucesos interesantes para el sistema. Una vez detectado uno de estos sucesos, el robot puede identificar su situación en un mapa topológico y actuar en consecuencia. Hemos denominado a estas situaciones especiales lugares sensorialmente relevantes, ya que (a) captan la atención del sistema y (b) pueden ser identificadas utilizando la información sensorial. Para explotar convenientemente los modelos construidos, hemos desarrollado un algoritmo capaz de elaborar planes internalizados, estableciendo una red de sugerencias en los lugares sensorialmente relevantes, de modo que el robot encuentra en estos puntos una dirección recomendada de navegación. Finalmente, hemos implementado un sistema de navegación robusto con habilidades para interpretar y adecuar los planes internalizados a las circunstancias concretas del momento. Nuestro sistema de navegación está basado en la teoría de campos de potencial artificial, a la que hemos incorporado la posibilidad de añadir cargas ficticias como ayuda a la evitación de mínimos locales. Como aportación adicional de esta tesis al campo genérico de la ciencia cognitiva, todos estos elementos se integran en una arquitectura centrada en la memoria, lo que pretende resaltar la importancia de ésta en los procesos cognitivos de los seres vivos y aporta un giro conceptual al punto de vista tradicional, centrado en los procesos. The general objective of this thesis is the development of a global navigation system endowed with planning and learning features for a mobile robot. Within this general objective we have devoted a special effort to the autonomous learning problem. Our main concern has been to establish the necessary mechanisms for the incremental development of a topological model of the robot’s environment using the sensory information. These mechanisms are based on a new concept proposed in the thesis: the sensory gradient. The sensory gradient is a mathematical device which works like a detector of “interesting” environment’s events. Once a particular event has been detected the robot can identify its situation in the topological map and to react accordingly. We have called these special situations relevant sensory places because (a) they capture the system’s attention and (b) they can be identified using the sensory information. To conveniently exploit the built-in models we have developed an algorithm able to make internalized plans, establishing a suggestion network in the sensory relevant places in such way that the robot can find at those places a recommended navigation direction. It has been also developed a robust navigation system able to navigate by means of interpreting and adapting the internalized plans to the concrete circumstances at each instant, i.e. a reactive navigation system. This reactive system is based on the artificial potential field approach with the additional feature introduced in the thesis of what we call fictitious charges as an aid to avoid local minima. As a general contribution of the thesis to the cognitive science field all the above described elements are integrated in a memory-based architecture, emphasizing the important role played by the memory in the cognitive processes of living beings and giving a conceptual turn in the usual process-based approach.

Learning an L1-regularized Gaussian Bayesian Network in the Equivalence Class Space

Learning the structure of a graphical model from data is a common task in a wide range of practical applications. In this paper, we focus on Gaussian Bayesian networks, i.e., on continuous data and directed acyclic graphs with a joint probability density of all variables given by a Gaussian. We propose to work in an equivalence class search space, specifically using the k-greedy equivalence search algorithm. This, combined with regularization techniques to guide the structure search, can learn sparse networks close to the one that generated the data. We provide results on some synthetic networks and on modeling the gene network of the two biological pathways regulating the biosynthesis of isoprenoids for the Arabidopsis thaliana plant

Optimized algorithms for the incremental analysis of logic programs

Global analysis of logic programs can be performed effectively by the use of one of several existing efficient algorithms. However, the traditional global analysis scheme in which all the program code is known in advance and no previous analysis information is available is unsatisfactory in many situations. Incrementa! analysis of logic programs has been shown to be feasible and much more efficient in certain contexts than traditional (non-incremental) global analysis. However, incremental analysis poses additional requirements on the fixpoint algorithm used. In this work we identify these requirements, present an important class of strategies meeting the requirements, present sufficient a priori conditions for such strategies, and propose, implement, and evalúate experimentally a novel algorithm for incremental analysis based on these ideas. The experimental results show that the proposed algorithm performs very efficiently in the incremental case while being comparable to (and, in some cases, considerably better than) other state-of-the-art analysis algorithms even for the non-incremental case. We argüe that our discussions, results, and experiments also shed light on some of the many tradeoffs involved in the design of algorithms for logic program analysis.

Regularization for sparsity in statistical analysis and machine learning

Pragmatism is the leading motivation of regularization. We can understand regularization as a modification of the maximum-likelihood estimator so that a reasonable answer could be given in an unstable or ill-posed situation. To mention some typical examples, this happens when fitting parametric or non-parametric models with more parameters than data or when estimating large covariance matrices. Regularization is usually used, in addition, to improve the bias-variance tradeoff of an estimation. Then, the definition of regularization is quite general, and, although the introduction of a penalty is probably the most popular type, it is just one out of multiple forms of regularization. In this dissertation, we focus on the applications of regularization for obtaining sparse or parsimonious representations, where only a subset of the inputs is used. A particular form of regularization, L1-regularization, plays a key role for reaching sparsity. Most of the contributions presented here revolve around L1-regularization, although other forms of regularization are explored (also pursuing sparsity in some sense). In addition to present a compact review of L1-regularization and its applications in statistical and machine learning, we devise methodology for regression, supervised classification and structure induction of graphical models. Within the regression paradigm, we focus on kernel smoothing learning, proposing techniques for kernel design that are suitable for high dimensional settings and sparse regression functions. We also present an application of regularized regression techniques for modeling the response of biological neurons. Supervised classification advances deal, on the one hand, with the application of regularization for obtaining a na¨ıve Bayes classifier and, on the other hand, with a novel algorithm for brain-computer interface design that uses group regularization in an efficient manner. Finally, we present a heuristic for inducing structures of Gaussian Bayesian networks using L1-regularization as a filter. El pragmatismo es la principal motivación de la regularización. Podemos entender la regularización como una modificación del estimador de máxima verosimilitud, de tal manera que se pueda dar una respuesta cuando la configuración del problema es inestable. A modo de ejemplo, podemos mencionar el ajuste de modelos paramétricos o no paramétricos cuando hay más parámetros que casos en el conjunto de datos, o la estimación de grandes matrices de covarianzas. Se suele recurrir a la regularización, además, para mejorar el compromiso sesgo-varianza en una estimación. Por tanto, la definición de regularización es muy general y, aunque la introducción de una función de penalización es probablemente el método más popular, éste es sólo uno de entre varias posibilidades. En esta tesis se ha trabajado en aplicaciones de regularización para obtener representaciones dispersas, donde sólo se usa un subconjunto de las entradas. En particular, la regularización L1 juega un papel clave en la búsqueda de dicha dispersión. La mayor parte de las contribuciones presentadas en la tesis giran alrededor de la regularización L1, aunque también se exploran otras formas de regularización (que igualmente persiguen un modelo disperso). Además de presentar una revisión de la regularización L1 y sus aplicaciones en estadística y aprendizaje de máquina, se ha desarrollado metodología para regresión, clasificación supervisada y aprendizaje de estructura en modelos gráficos. Dentro de la regresión, se ha trabajado principalmente en métodos de regresión local, proponiendo técnicas de diseño del kernel que sean adecuadas a configuraciones de alta dimensionalidad y funciones de regresión dispersas. También se presenta una aplicación de las técnicas de regresión regularizada para modelar la respuesta de neuronas reales. Los avances en clasificación supervisada tratan, por una parte, con el uso de regularización para obtener un clasificador naive Bayes y, por otra parte, con el desarrollo de un algoritmo que usa regularización por grupos de una manera eficiente y que se ha aplicado al diseño de interfaces cerebromáquina. Finalmente, se presenta una heurística para inducir la estructura de redes Bayesianas Gaussianas usando regularización L1 a modo de filtro.

Markov blanket-based approach for learning multi-dimensional Bayesian network classifiers: An application to predict the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson's Disease Questionnaire (PDQ-39)

Multi-dimensional Bayesian network classifiers (MBCs) are probabilistic graphical models recently proposed to deal with multi-dimensional classification problems, where each instance in the data set has to be assigned to more than one class variable. In this paper, we propose a Markov blanket-based approach for learning MBCs from data. Basically, it consists of determining the Markov blanket around each class variable using the HITON algorithm, then specifying the directionality over the MBC subgraphs. Our approach is applied to the prediction problem of the European Quality of Life-5 Dimensions (EQ-5D) from the 39-item Parkinson’s Disease Questionnaire (PDQ-39) in order to estimate the health-related quality of life of Parkinson’s patients. Fivefold cross-validation experiments were carried out on randomly generated synthetic data sets, Yeast data set, as well as on a real-world Parkinson’s disease data set containing 488 patients. The experimental study, including comparison with additional Bayesian network-based approaches, back propagation for multi-label learning, multi-label k-nearest neighbor, multinomial logistic regression, ordinary least squares, and censored least absolute deviations, shows encouraging results in terms of predictive accuracy as well as the identification of dependence relationships among class and feature variables.