Two-sided learning in New Keynesian models: Dynamics, (lack of) convergence and the value of information

This paper investigates the role of learning by private agents and the central bank(two-sided learning) in a New Keynesian framework in which both sides of the economyhave asymmetric and imperfect knowledge about the true data generating process. Weassume that all agents employ the data that they observe (which may be distinct fordifferent sets of agents) to form beliefs about unknown aspects of the true model ofthe economy, use their beliefs to decide on actions, and revise these beliefs througha statistical learning algorithm as new information becomes available. We study theshort-run dynamics of our model and derive its policy recommendations, particularlywith respect to central bank communications. We demonstrate that two-sided learningcan generate substantial increases in volatility and persistence, and alter the behaviorof the variables in the model in a significant way. Our simulations do not convergeto a symmetric rational expectations equilibrium and we highlight one source thatinvalidates the convergence results of Marcet and Sargent (1989). Finally, we identifya novel aspect of central bank communication in models of learning: communicationcan be harmful if the central bank's model is substantially mis-specified.

Cluster-based active learning for compact image classification

In this paper, we consider active sampling to label pixels grouped with hierarchical clustering. The objective of the method is to match the data relationships discovered by the clustering algorithm with the user's desired class semantics. The first is represented as a complete tree to be pruned and the second is iteratively provided by the user. The active learning algorithm proposed searches the pruning of the tree that best matches the labels of the sampled points. By choosing the part of the tree to sample from according to current pruning's uncertainty, sampling is focused on most uncertain clusters. This way, large clusters for which the class membership is already fixed are no longer queried and sampling is focused on division of clusters showing mixed labels. The model is tested on a VHR image in a multiclass classification setting. The method clearly outperforms random sampling in a transductive setting, but cannot generalize to unseen data, since it aims at optimizing the classification of a given cluster structure.

New kernel functions and learning methods for text and data mining

Recent advances in machine learning methods enable increasingly the automatic construction of various types of computer assisted methods that have been difficult or laborious to program by human experts. The tasks for which this kind of tools are needed arise in many areas, here especially in the fields of bioinformatics and natural language processing. The machine learning methods may not work satisfactorily if they are not appropriately tailored to the task in question. However, their learning performance can often be improved by taking advantage of deeper insight of the application domain or the learning problem at hand. This thesis considers developing kernel-based learning algorithms incorporating this kind of prior knowledge of the task in question in an advantageous way. Moreover, computationally efficient algorithms for training the learning machines for specific tasks are presented. In the context of kernel-based learning methods, the incorporation of prior knowledge is often done by designing appropriate kernel functions. Another well-known way is to develop cost functions that fit to the task under consideration. For disambiguation tasks in natural language, we develop kernel functions that take account of the positional information and the mutual similarities of words. It is shown that the use of this information significantly improves the disambiguation performance of the learning machine. Further, we design a new cost function that is better suitable for the task of information retrieval and for more general ranking problems than the cost functions designed for regression and classification. We also consider other applications of the kernel-based learning algorithms such as text categorization, and pattern recognition in differential display. We develop computationally efficient algorithms for training the considered learning machines with the proposed kernel functions. We also design a fast cross-validation algorithm for regularized least-squares type of learning algorithm. Further, an efficient version of the regularized least-squares algorithm that can be used together with the new cost function for preference learning and ranking tasks is proposed. In summary, we demonstrate that the incorporation of prior knowledge is possible and beneficial, and novel advanced kernels and cost functions can be used in algorithms efficiently.

Learning Preferences with Kernel-Based Methods

Learning of preference relations has recently received significant attention in machine learning community. It is closely related to the classification and regression analysis and can be reduced to these tasks. However, preference learning involves prediction of ordering of the data points rather than prediction of a single numerical value as in case of regression or a class label as in case of classification. Therefore, studying preference relations within a separate framework facilitates not only better theoretical understanding of the problem, but also motivates development of the efficient algorithms for the task. Preference learning has many applications in domains such as information retrieval, bioinformatics, natural language processing, etc. For example, algorithms that learn to rank are frequently used in search engines for ordering documents retrieved by the query. Preference learning methods have been also applied to collaborative filtering problems for predicting individual customer choices from the vast amount of user generated feedback. In this thesis we propose several algorithms for learning preference relations. These algorithms stem from well founded and robust class of regularized least-squares methods and have many attractive computational properties. In order to improve the performance of our methods, we introduce several non-linear kernel functions. Thus, contribution of this thesis is twofold: kernel functions for structured data that are used to take advantage of various non-vectorial data representations and the preference learning algorithms that are suitable for different tasks, namely efficient learning of preference relations, learning with large amount of training data, and semi-supervised preference learning. Proposed kernel-based algorithms and kernels are applied to the parse ranking task in natural language processing, document ranking in information retrieval, and remote homology detection in bioinformatics domain. Training of kernel-based ranking algorithms can be infeasible when the size of the training set is large. This problem is addressed by proposing a preference learning algorithm whose computation complexity scales linearly with the number of training data points. We also introduce sparse approximation of the algorithm that can be efficiently trained with large amount of data. For situations when small amount of labeled data but a large amount of unlabeled data is available, we propose a co-regularized preference learning algorithm. To conclude, the methods presented in this thesis address not only the problem of the efficient training of the algorithms but also fast regularization parameter selection, multiple output prediction, and cross-validation. Furthermore, proposed algorithms lead to notably better performance in many preference learning tasks considered.

Sequential Machine learning Approaches for Portfolio Management

Cette thèse envisage un ensemble de méthodes permettant aux algorithmes d'apprentissage statistique de mieux traiter la nature séquentielle des problèmes de gestion de portefeuilles financiers. Nous débutons par une considération du problème général de la composition d'algorithmes d'apprentissage devant gérer des tâches séquentielles, en particulier celui de la mise-à-jour efficace des ensembles d'apprentissage dans un cadre de validation séquentielle. Nous énumérons les desiderata que des primitives de composition doivent satisfaire, et faisons ressortir la difficulté de les atteindre de façon rigoureuse et efficace. Nous poursuivons en présentant un ensemble d'algorithmes qui atteignent ces objectifs et présentons une étude de cas d'un système complexe de prise de décision financière utilisant ces techniques. Nous décrivons ensuite une méthode générale permettant de transformer un problème de décision séquentielle non-Markovien en un problème d'apprentissage supervisé en employant un algorithme de recherche basé sur les K meilleurs chemins. Nous traitons d'une application en gestion de portefeuille où nous entraînons un algorithme d'apprentissage à optimiser directement un ratio de Sharpe (ou autre critère non-additif incorporant une aversion au risque). Nous illustrons l'approche par une étude expérimentale approfondie, proposant une architecture de réseaux de neurones spécialisée à la gestion de portefeuille et la comparant à plusieurs alternatives. Finalement, nous introduisons une représentation fonctionnelle de séries chronologiques permettant à des prévisions d'être effectuées sur un horizon variable, tout en utilisant un ensemble informationnel révélé de manière progressive. L'approche est basée sur l'utilisation des processus Gaussiens, lesquels fournissent une matrice de covariance complète entre tous les points pour lesquels une prévision est demandée. Cette information est utilisée à bon escient par un algorithme qui transige activement des écarts de cours (price spreads) entre des contrats à terme sur commodités. L'approche proposée produit, hors échantillon, un rendement ajusté pour le risque significatif, après frais de transactions, sur un portefeuille de 30 actifs.

Étude de la formation et de l'évolution de nanostructures par méthodes Monte Carlo

Cette thèse, composée de quatre articles scientifiques, porte sur les méthodes numériques atomistiques et leur application à des systèmes semi-conducteurs nanostructurés. Nous introduisons les méthodes accélérées conçues pour traiter les événements activés, faisant un survol des développements du domaine. Suit notre premier article, qui traite en détail de la technique d'activation-relaxation cinétique (ART-cinétique), un algorithme Monte Carlo cinétique hors-réseau autodidacte basé sur la technique de l'activation-relaxation nouveau (ARTn), dont le développement ouvre la voie au traitement exact des interactions élastiques tout en permettant la simulation de matériaux sur des plages de temps pouvant atteindre la seconde. Ce développement algorithmique, combiné à des données expérimentales récentes, ouvre la voie au second article. On y explique le relâchement de chaleur par le silicium cristallin suite à son implantation ionique avec des ions de Si à 3 keV. Grâce à nos simulations par ART-cinétique et l'analyse de données obtenues par nanocalorimétrie, nous montrons que la relaxation est décrite par un nouveau modèle en deux temps: "réinitialiser et relaxer" ("Replenish-and-Relax"). Ce modèle, assez général, peut potentiellement expliquer la relaxation dans d'autres matériaux désordonnés. Par la suite, nous poussons l'analyse plus loin. Le troisième article offre une analyse poussée des mécanismes atomistiques responsables de la relaxation lors du recuit. Nous montrons que les interactions élastiques entre des défauts ponctuels et des petits complexes de défauts contrôlent la relaxation, en net contraste avec la littérature qui postule que des "poches amorphes" jouent ce rôle. Nous étudions aussi certains sous-aspects de la croissance de boîtes quantiques de Ge sur Si (001). En effet, après une courte mise en contexte et une introduction méthodologique supplémentaire, le quatrième article décrit la structure de la couche de mouillage lors du dépôt de Ge sur Si (001) à l'aide d'une implémentation QM/MM du code BigDFT-ART. Nous caractérisons la structure de la reconstruction 2xN de la surface et abaissons le seuil de la température nécessaire pour la diffusion du Ge en sous-couche prédit théoriquement par plus de 100 K.

Modélisation des émotions de l’apprenant et interventions implicites pour les systèmes tutoriels intelligents

La modélisation de l’expérience de l’utilisateur dans les Interactions Homme-Machine est un enjeu important pour la conception et le développement des systèmes adaptatifs intelligents. Dans ce contexte, une attention particulière est portée sur les réactions émotionnelles de l’utilisateur, car elles ont une influence capitale sur ses aptitudes cognitives, comme la perception et la prise de décision. La modélisation des émotions est particulièrement pertinente pour les Systèmes Tutoriels Émotionnellement Intelligents (STEI). Ces systèmes cherchent à identifier les émotions de l’apprenant lors des sessions d’apprentissage, et à optimiser son expérience d’interaction en recourant à diverses stratégies d’interventions. Cette thèse vise à améliorer les méthodes de modélisation des émotions et les stratégies émotionnelles utilisées actuellement par les STEI pour agir sur les émotions de l’apprenant. Plus précisément, notre premier objectif a été de proposer une nouvelle méthode pour détecter l’état émotionnel de l’apprenant, en utilisant différentes sources d’informations qui permettent de mesurer les émotions de façon précise, tout en tenant compte des variables individuelles qui peuvent avoir un impact sur la manifestation des émotions. Pour ce faire, nous avons développé une approche multimodale combinant plusieurs mesures physiologiques (activité cérébrale, réactions galvaniques et rythme cardiaque) avec des variables individuelles, pour détecter une émotion très fréquemment observée lors des sessions d’apprentissage, à savoir l’incertitude. Dans un premier lieu, nous avons identifié les indicateurs physiologiques clés qui sont associés à cet état, ainsi que les caractéristiques individuelles qui contribuent à sa manifestation. Puis, nous avons développé des modèles prédictifs permettant de détecter automatiquement cet état à partir des différentes variables analysées, à travers l’entrainement d’algorithmes d’apprentissage machine. Notre deuxième objectif a été de proposer une approche unifiée pour reconnaître simultanément une combinaison de plusieurs émotions, et évaluer explicitement l’impact de ces émotions sur l’expérience d’interaction de l’apprenant. Pour cela, nous avons développé une plateforme hiérarchique, probabiliste et dynamique permettant de suivre les changements émotionnels de l'apprenant au fil du temps, et d’inférer automatiquement la tendance générale qui caractérise son expérience d’interaction à savoir : l’immersion, le blocage ou le décrochage. L’immersion correspond à une expérience optimale : un état dans lequel l'apprenant est complètement concentré et impliqué dans l’activité d’apprentissage. L’état de blocage correspond à une tendance d’interaction non optimale où l'apprenant a de la difficulté à se concentrer. Finalement, le décrochage correspond à un état extrêmement défavorable où l’apprenant n’est plus du tout impliqué dans l’activité d’apprentissage. La plateforme proposée intègre trois modalités de variables diagnostiques permettant d’évaluer l’expérience de l’apprenant à savoir : des variables physiologiques, des variables comportementales, et des mesures de performance, en combinaison avec des variables prédictives qui représentent le contexte courant de l’interaction et les caractéristiques personnelles de l'apprenant. Une étude a été réalisée pour valider notre approche à travers un protocole expérimental permettant de provoquer délibérément les trois tendances ciblées durant l’interaction des apprenants avec différents environnements d’apprentissage. Enfin, notre troisième objectif a été de proposer de nouvelles stratégies pour influencer positivement l’état émotionnel de l’apprenant, sans interrompre la dynamique de la session d’apprentissage. Nous avons à cette fin introduit le concept de stratégies émotionnelles implicites : une nouvelle approche pour agir subtilement sur les émotions de l’apprenant, dans le but d’améliorer son expérience d’apprentissage. Ces stratégies utilisent la perception subliminale, et plus précisément une technique connue sous le nom d’amorçage affectif. Cette technique permet de solliciter inconsciemment les émotions de l’apprenant, à travers la projection d’amorces comportant certaines connotations affectives. Nous avons mis en œuvre une stratégie émotionnelle implicite utilisant une forme particulière d’amorçage affectif à savoir : le conditionnement évaluatif, qui est destiné à améliorer de façon inconsciente l’estime de soi. Une étude expérimentale a été réalisée afin d’évaluer l’impact de cette stratégie sur les réactions émotionnelles et les performances des apprenants.

Towards Non Invasive Diagnosis of Scoliosis Using Semi-supervised Learning Approach

In this paper, a new methodology for the prediction of scoliosis curve types from non invasive acquisitions of the back surface of the trunk is proposed. One hundred and fifty-nine scoliosis patients had their back surface acquired in 3D using an optical digitizer. Each surface is then characterized by 45 local measurements of the back surface rotation. Using a semi-supervised algorithm, the classifier is trained with only 32 labeled and 58 unlabeled data. Tested on 69 new samples, the classifier succeeded in classifying correctly 87.0% of the data. After reducing the number of labeled training samples to 12, the behavior of the resulting classifier tends to be similar to the reference case where the classifier is trained only with the maximum number of available labeled data. Moreover, the addition of unlabeled data guided the classifier towards more generalizable boundaries between the classes. Those results provide a proof of feasibility for using a semi-supervised learning algorithm to train a classifier for the prediction of a scoliosis curve type, when only a few training data are labeled. This constitutes a promising clinical finding since it will allow the diagnosis and the follow-up of scoliotic deformities without exposing the patient to X-ray radiations.

A Reinforcement Learning Approach to Economic Dispatch using Neural Networks

This paper presents a Reinforcement Learning (RL) approach to economic dispatch (ED) using Radial Basis Function neural network. We formulate the ED as an N stage decision making problem. We propose a novel architecture to store Qvalues and present a learning algorithm to learn the weights of the neural network. Even though many stochastic search techniques like simulated annealing, genetic algorithm and evolutionary programming have been applied to ED, they require searching for the optimal solution for each load demand. Also they find limitation in handling stochastic cost functions. In our approach once we learn the Q-values, we can find the dispatch for any load demand. We have recently proposed a RL approach to ED. In that approach, we could find only the optimum dispatch for a set of specified discrete values of power demand. The performance of the proposed algorithm is validated by taking IEEE 6 bus system, considering transmission losses

Hierarchical Mixtures of Experts and the EM Algorithm

We present a tree-structured architecture for supervised learning. The statistical model underlying the architecture is a hierarchical mixture model in which both the mixture coefficients and the mixture components are generalized linear models (GLIM's). Learning is treated as a maximum likelihood problem; in particular, we present an Expectation-Maximization (EM) algorithm for adjusting the parameters of the architecture. We also develop an on-line learning algorithm in which the parameters are updated incrementally. Comparative simulation results are presented in the robot dynamics domain.

Modeling Stock Order Flows and Learning Market-Making from Data

Stock markets employ specialized traders, market-makers, designed to provide liquidity and volume to the market by constantly supplying both supply and demand. In this paper, we demonstrate a novel method for modeling the market as a dynamic system and a reinforcement learning algorithm that learns profitable market-making strategies when run on this model. The sequence of buys and sells for a particular stock, the order flow, we model as an Input-Output Hidden Markov Model fit to historical data. When combined with the dynamics of the order book, this creates a highly non-linear and difficult dynamic system. Our reinforcement learning algorithm, based on likelihood ratios, is run on this partially-observable environment. We demonstrate learning results for two separate real stocks.

A forward constrained selection algorithm for probabilistic neural network

A new probabilistic neural network (PNN) learning algorithm based on forward constrained selection (PNN-FCS) is proposed. An incremental learning scheme is adopted such that at each step, new neurons, one for each class, are selected from the training samples arid the weights of the neurons are estimated so as to minimize the overall misclassification error rate. In this manner, only the most significant training samples are used as the neurons. It is shown by simulation that the resultant networks of PNN-FCS have good classification performance compared to other types of classifiers, but much smaller model sizes than conventional PNN.

Neurofuzzy mixture of experts network parallel learning and model construction algorithms

A connection between a fuzzy neural network model with the mixture of experts network (MEN) modelling approach is established. Based on this linkage, two new neuro-fuzzy MEN construction algorithms are proposed to overcome the curse of dimensionality that is inherent in the majority of associative memory networks and/or other rule based systems. The first construction algorithm employs a function selection manager module in an MEN system. The second construction algorithm is based on a new parallel learning algorithm in which each model rule is trained independently, for which the parameter convergence property of the new learning method is established. As with the first approach, an expert selection criterion is utilised in this algorithm. These two construction methods are equivalent in their effectiveness in overcoming the curse of dimensionality by reducing the dimensionality of the regression vector, but the latter has the additional computational advantage of parallel processing. The proposed algorithms are analysed for effectiveness followed by numerical examples to illustrate their efficacy for some difficult data based modelling problems.

Adaptive modelling with tunable RBF network using multi-innovation RLS algorithm assisted by swarm intelligence

In this paper, we propose a new on-line learning algorithm for the non-linear system identification: the swarm intelligence aided multi-innovation recursive least squares (SI-MRLS) algorithm. The SI-MRLS algorithm applies the particle swarm optimization (PSO) to construct a flexible radial basis function (RBF) model so that both the model structure and output weights can be adapted. By replacing an insignificant RBF node with a new one based on the increment of error variance criterion at every iteration, the model remains at a limited size. The multi-innovation RLS algorithm is used to update the RBF output weights which are known to have better accuracy than the classic RLS. The proposed method can produces a parsimonious model with good performance. Simulation result are also shown to verify the SI-MRLS algorithm.

Application of logistic regression for fault analysis in an industrial printing process

A statistical technique for fault analysis in industrial printing is reported. The method specifically deals with binary data, for which the results of the production process fall into two categories, rejected or accepted. The method is referred to as logistic regression, and is capable of predicting future fault occurrences by the analysis of current measurements from machine parts sensors. Individual analysis of each type of fault can determine which parts of the plant have a significant influence on the occurrence of such faults; it is also possible to infer which measurable process parameters have no significant influence on the generation of these faults. Information derived from the analysis can be helpful in the operator's interpretation of the current state of the plant. Appropriate actions may then be taken to prevent potential faults from occurring. The algorithm is being implemented as part of an applied self-learning expert system.