This thesis develops flexible and principled nonparametric learning algorithms to explore, understand, and predict high dimensional and complex datasets. Such data appear frequently in modern scientific domains and lead to numerous important applications. For example, exploring high dimensional functional magnetic resonance imaging data helps us to better understand brain functionalities; inferring large-scale gene regulatory network is crucial for new drug design and development; detecting anomalies in high dimensional transaction databases is vital for corporate and government security. Our main results include a rigorous theoretical framework and efficient nonparametric learning algorithms that exploit hidden structures to overcome the curse of dimensionality when analyzing massive high dimensional datasets. These algorithms have strong theoretical guarantees and provide high dimensional nonparametric recipes for many important learning tasks, ranging from unsupervised exploratory data analysis to supervised predictive modeling. In this thesis, we address three aspects: 1 Understanding the statistical theories of high dimensional nonparametric inference, including risk, estimation, and model selection consistency; 2 Designing new methods for different data-analysis tasks, including regression, classification, density estimation, graphical model learning, multi-task learning, spatial-temporal adaptive learning; 3 Demonstrating the usefulness of these methods in scientific applications, including functional genomics, cognitive neuroscience, and meteorology. In the last part of this thesis, we also present the future vision of high dimensional and large-scale nonparametric inference.

In this thesis, research efforts are dedicated towards the development of practical adaptable techniques to be used in Smart Homes and Buildings, with the aim to improve energy management and conservation, while enhancing the learning capabilities of Programmable Communicating Thermostats (PCT) – “transforming” them into smart adaptable devices, i.e., “Smart Thermostats”. An Adaptable Hybrid Intelligent System utilizing Wireless Sensor Network (WSN) and Artificial Intelligence (AI) techniques is presented, based on which, a novel Adaptive Learning System (ALS) model utilizing WSN, a rule-based system and Adaptive Resonance Theory (ART) concepts is proposed. The main goal of the ALS is to adapt to the occupant’s pattern and/or schedule changes by providing comfort, while not ignoring the energy conservation aspect. The proposed ALS analytical model is a technique which enables PCTs to learn and adapt to user input pattern changes and/or other parameters of interest. A new algorithm for finding the global maximum in a predefined interval within a two dimensional space is proposed. The proposed algorithm is a synergy of reward/punish concepts from the reinforcement learning (RL) and agent-based technique, for use in small-scale embedded systems with limited memory and/or processing power, such as the wireless sensor/actuator nodes. An application is implemented to observe the algorithm at work and to demonstrate its main features. It was observed that the “RL and Agent-based Search”, versus the “RL only” technique, yielded better performance results with respect to the number of iterations and function evaluations needed to find the global maximum. Furthermore, a “House Simulator” is developed as a tool to simulate house heating/cooling systems and to assist in the practical implementation of the ALS model under different scenarios. The main building blocks of the simulator are the “House Simulator”, the “Smart Thermostat”, and a placeholder for the “Adaptive Learning Models”. As a result, a novel adaptive learning algorithm, “Observe, Learn and Adapt” (OLA) is proposed and demonstrated, reflecting the main features of the ALS model. Its evaluation is achieved with the aid of the “House Simulator”. OLA, with the use of sensors and the application of the ALS model learning technique, captures the essence of an actual PCT reflecting a smart and adaptable device. The experimental performance results indicate adaptability and potential energy savings of the single in comparison to the zone controlled scenarios with the OLA capabilities being enabled.

Esta tese é composta de três artigos relacionados à política monetária e inflação e possuem em comum a ênfase na importância das expectativas tanto para o desenho da política monetária como para a dinâmica inflacionária. No primeiro ensaio, contribuímos para o debate sobre a resposta apropriada de política monetária a flutuações de preços de ativos em um contexto de aprendizagem adaptativa. O modelo conta com dois tipos de regras de juros instrumentais como em Bullard e Mitra (2002), porém com um papel adicional para preços de ativos. Do ponto de vista da E-Estabilidade, conclui-se que uma resposta a preços de ativos não é desejável nem com a regra que utiliza expectativas futuras nem com a regra que responde a valores contemporâneos. Crenças heterogêneas a respeito da dinâmica das flutuações de preços de ativos, inflação e hiato do produto são introduzidas. Também é avaliada uma regra de política monetária ótima que inclui um peso para os preços de ativos. De forma geral, conclui-se que o princípio de Taylor é relevante para todas as regras de juros analisadas e que os bancos centrais devem agir com cautela ao considerar a introdução de preços de ativos na política monetária. No segundo ensaio, oferecemos estimativas recentes de persistência inflacionária no Brasil, com uma abordagem multivariada de componentes não-observados, na qual são consideradas as seguintes fontes que impactam na persistência da inflação: desvios das expectativas da meta real de inflação; persistência dos fatores que provocam inflação; e termos defasados da inflação. Dados de inflação, produto e taxas de juros são decompostos em componentes não-observados e, para simplificar a estimativa de um número grande de variáveis desconhecidas, utilizamos análise bayesiana, seguindo Dossche e Everaert (2005). Os resultados indicam que a persistência baseada em expectativas tem grande participação na persistência inflacionária no Brasil, que tem diminuído nos últimos anos. Tal resultado implica que apenas as tradicionais fricções no ajuste de preços usadas nos modelos macroeconômicos não são suficientes para representar a real persistência da inflação. No último capítulo estimamos diversas curvas de Phillips reduzidas com dados brasileiros recentes, numa abordagem de séries de tempo com componentes não-observados, que se apresenta como alternativa às tradicionais estimativas, baseadas em métodos GMM, de curvas de Phillips Novo-Keynesianas (NKPC), que raramente foram bem sucedidas empiricamente. A decomposição em tendência, sazonalidade e ciclo oferece, através do resultado gráfico, interpretação econômica direta. Diferentemente de Harvey (2011), incluímos expectativas de inflação nas estimações, assim como na NKPC habitual. A inflação no Brasil parece ter respondido cada vez menos às medidas de atividade econômica consideradas. Isso consiste em evidência de achatamento da curva de Phillips no Brasil, o que significa por um lado custos de desinflação mais altos, mas por outro lado menores pressões inflacionárias derivadas de crescimento do produto. / This thesis is composed of three essays on monetary policy and inflation that share particular emphasis on the importance of expectations for both monetary policy design and inflation dynamics. First we contribute to the debate on the appropriate response of monetary policy to asset price fluctuations in an adaptive learning context. Our model accounts for two types of instrumental rules in the spirit of Bullard and Mitra (2002), but with an additional role for asset prices. From the point of view of EStability, we find that a response to stock prices is not desirable under both a forward expectations policy rule and an interest rate rule responding to contemporaneous values. Heterogeneous beliefs about the dynamics of asset price fluctuations, inflation and the output gap are introduced. We also evaluate an optimal monetary policy rule including a weight on asset prices. Overall we find that the Taylor principle remain important over all interest rate rules analysed and that central banks should remain cautious when considering the introduction of stock prices in monetary policy. In the second essay, we provide recent estimates of inflation persistence in Brazil in a multivariate framework of unobserved components, whereby we account for the following sources affecting inflation persistence: First, deviations of expectations from the actual policy target; second, persistence of the factors driving inflation; and third, lagged inflation terms. Data on inflation, output and interest rates are decomposed into unobserved components and to simplify the estimation of a great number of unknown variables, we utilize bayesian analysis as in Dossche and Everaert (2005). Our results indicate that expectations-based persistence matters considerably for inflation persistence in Brazil, which has experienced an overall decrease in the last few years. This finding implies that traditional price-setting frictions used in macroeconomic models are not enough to represent actual inflation persistence. In the last chapter we estimate alternative reduced-form Phillips curves with recent Brazilian data, using a framework of time series with unobserved components, as an alternative to traditional GMM estimations of the New Keynesian Phillips Curve (NKPC), which have seldom been empirically successful. The decomposition into trend, seasonal and cycle features offers, through the graphical output, straightforward economic interpretations. Differently from Harvey (2011), we allow for inflation expectations as in the usual NKPC. Inflation in Brazil seems to have responded gradually less to measures of economic activity in recent years. This provides some evidence of a flattening of the Phillips curve in Brazil, which means higher costs of disinflation on the one hand, but also lower inflationary pressures derived from output growth, on the other.

Continuous change and development are essential aspects of evolving environments and applications, including, but not limited to, smart cities, military, medicine, nuclear reactors, self-driving cars, aviation, and aerospace. That is, the fundamental characteristics of such environments may evolve, and so cause dangerous consequences, e.g., putting people lives at stake, if no reaction is adopted. Therefore, learning systems need to apply intelligent algorithms to monitor evolvement in their environments and update themselves effectively. Further, we may experience fluctuations regarding the performance of learning algorithms due to the nature of incoming data as it continuously evolves. That is, the current efficient learning approach may become deprecated after a change in data or environment. Hence, the question 'how to have an efficient learning algorithm over time against evolving data?' has to be addressed. In this thesis, we have made two contributions to settle the challenges described above. In the machine learning literature, the phenomenon of (distributional) change in data is known as concept drift. Concept drift may shift decision boundaries, and cause a decline in accuracy. Learning algorithms, indeed, have to detect concept drift in evolving data streams and replace their predictive models accordingly. To address this challenge, adaptive learners have been devised which may utilize drift detection methods to locate the drift points in dynamic and changing data streams. A drift detection method able to discover the drift points quickly, with the lowest false positive and false negative rates, is preferred. False positive refers to incorrectly alarming for concept drift, and false negative refers to not alarming for concept drift. In this thesis, we introduce three algorithms, called as the Fast Hoeffding Drift Detection Method (FHDDM), the Stacking Fast Hoeffding Drift Detection Method (FHDDMS), and the McDiarmid Drift Detection Methods (MDDMs), for detecting drift points with the minimum delay, false positive, and false negative rates. FHDDM is a sliding window-based algorithm and applies Hoeffding’s inequality (Hoeffding, 1963) to detect concept drift. FHDDM slides its window over the prediction results, which are either 1 (for a correct prediction) or 0 (for a wrong prediction). Meanwhile, it compares the mean of elements inside the window with the maximum mean observed so far; subsequently, a significant difference between the two means, upper-bounded by the Hoeffding inequality, indicates the occurrence of concept drift. The FHDDMS extends the FHDDM algorithm by sliding multiple windows over its entries for a better drift detection regarding the detection delay and false negative rate. In contrast to FHDDM/S, the MDDM variants assign weights to their entries, i.e., higher weights are associated with the most recent entries in the sliding window, for faster detection of concept drift. The rationale is that recent examples reflect the ongoing situation adequately. Then, by putting higher weights on the latest entries, we may detect concept drift quickly. An MDDM algorithm bounds the difference between the weighted mean of elements in the sliding window and the maximum weighted mean seen so far, using McDiarmid’s inequality (McDiarmid, 1989). Eventually, it alarms for concept drift once a significant difference is experienced. We experimentally show that FHDDM/S and MDDMs outperform the state-of-the-art by representing promising results in terms of the adaptation and classification measures. Due to the evolving nature of data streams, the performance of an adaptive learner, which is defined by the classification, adaptation, and resource consumption measures, may fluctuate over time. In fact, a learning algorithm, in the form of a (classifier, detector) pair, may present a significant performance before a concept drift point, but not after. We define this problem by the question 'how can we ensure that an efficient classifier-detector pair is present at any time in an evolving environment?' To answer this, we have developed the Tornado framework which runs various kinds of learning algorithms simultaneously against evolving data streams. Each algorithm incrementally and independently trains a predictive model and updates the statistics of its drift detector. Meanwhile, our framework monitors the (classifier, detector) pairs, and recommends the efficient one, concerning the classification, adaptation, and resource consumption performance, to the user. We further define the holistic CAR measure that integrates the classification, adaptation, and resource consumption measures for evaluating the performance of adaptive learning algorithms. Our experiments confirm that the most efficient algorithm may differ over time because of the developing and evolving nature of data streams.

Esta tese é composta de três artigos relacionados à política monetária e inflação e possuem em comum a ênfase na importância das expectativas tanto para o desenho da política monetária como para a dinâmica inflacionária. No primeiro ensaio, contribuímos para o debate sobre a resposta apropriada de política monetária a flutuações de preços de ativos em um contexto de aprendizagem adaptativa. O modelo conta com dois tipos de regras de juros instrumentais como em Bullard e Mitra (2002), porém com um papel adicional para preços de ativos. Do ponto de vista da E-Estabilidade, conclui-se que uma resposta a preços de ativos não é desejável nem com a regra que utiliza expectativas futuras nem com a regra que responde a valores contemporâneos. Crenças heterogêneas a respeito da dinâmica das flutuações de preços de ativos, inflação e hiato do produto são introduzidas. Também é avaliada uma regra de política monetária ótima que inclui um peso para os preços de ativos. De forma geral, conclui-se que o princípio de Taylor é relevante para todas as regras de juros analisadas e que os bancos centrais devem agir com cautela ao considerar a introdução de preços de ativos na política monetária. No segundo ensaio, oferecemos estimativas recentes de persistência inflacionária no Brasil, com uma abordagem multivariada de componentes não-observados, na qual são consideradas as seguintes fontes que impactam na persistência da inflação: desvios das expectativas da meta real de inflação; persistência dos fatores que provocam inflação; e termos defasados da inflação. Dados de inflação, produto e taxas de juros são decompostos em componentes não-observados e, para simplificar a estimativa de um número grande de variáveis desconhecidas, utilizamos análise bayesiana, seguindo Dossche e Everaert (2005). Os resultados indicam que a persistência baseada em expectativas tem grande participação na persistência inflacionária no Brasil, que tem diminuído nos últimos anos. Tal resultado implica que apenas as tradicionais fricções no ajuste de preços usadas nos modelos macroeconômicos não são suficientes para representar a real persistência da inflação. No último capítulo estimamos diversas curvas de Phillips reduzidas com dados brasileiros recentes, numa abordagem de séries de tempo com componentes não-observados, que se apresenta como alternativa às tradicionais estimativas, baseadas em métodos GMM, de curvas de Phillips Novo-Keynesianas (NKPC), que raramente foram bem sucedidas empiricamente. A decomposição em tendência, sazonalidade e ciclo oferece, através do resultado gráfico, interpretação econômica direta. Diferentemente de Harvey (2011), incluímos expectativas de inflação nas estimações, assim como na NKPC habitual. A inflação no Brasil parece ter respondido cada vez menos às medidas de atividade econômica consideradas. Isso consiste em evidência de achatamento da curva de Phillips no Brasil, o que significa por um lado custos de desinflação mais altos, mas por outro lado menores pressões inflacionárias derivadas de crescimento do produto. / This thesis is composed of three essays on monetary policy and inflation that share particular emphasis on the importance of expectations for both monetary policy design and inflation dynamics. First we contribute to the debate on the appropriate response of monetary policy to asset price fluctuations in an adaptive learning context. Our model accounts for two types of instrumental rules in the spirit of Bullard and Mitra (2002), but with an additional role for asset prices. From the point of view of EStability, we find that a response to stock prices is not desirable under both a forward expectations policy rule and an interest rate rule responding to contemporaneous values. Heterogeneous beliefs about the dynamics of asset price fluctuations, inflation and the output gap are introduced. We also evaluate an optimal monetary policy rule including a weight on asset prices. Overall we find that the Taylor principle remain important over all interest rate rules analysed and that central banks should remain cautious when considering the introduction of stock prices in monetary policy. In the second essay, we provide recent estimates of inflation persistence in Brazil in a multivariate framework of unobserved components, whereby we account for the following sources affecting inflation persistence: First, deviations of expectations from the actual policy target; second, persistence of the factors driving inflation; and third, lagged inflation terms. Data on inflation, output and interest rates are decomposed into unobserved components and to simplify the estimation of a great number of unknown variables, we utilize bayesian analysis as in Dossche and Everaert (2005). Our results indicate that expectations-based persistence matters considerably for inflation persistence in Brazil, which has experienced an overall decrease in the last few years. This finding implies that traditional price-setting frictions used in macroeconomic models are not enough to represent actual inflation persistence. In the last chapter we estimate alternative reduced-form Phillips curves with recent Brazilian data, using a framework of time series with unobserved components, as an alternative to traditional GMM estimations of the New Keynesian Phillips Curve (NKPC), which have seldom been empirically successful. The decomposition into trend, seasonal and cycle features offers, through the graphical output, straightforward economic interpretations. Differently from Harvey (2011), we allow for inflation expectations as in the usual NKPC. Inflation in Brazil seems to have responded gradually less to measures of economic activity in recent years. This provides some evidence of a flattening of the Phillips curve in Brazil, which means higher costs of disinflation on the one hand, but also lower inflationary pressures derived from output growth, on the other.

The purpose of this project is to develop a novel proof-of-concept system in attempt to measure affective states during learning-tasks and investigate whether machine learning models trained with this data has the potential to enhance the learning experience for an individual. By considering biometric signals from a user during a learning session, the affective states anxiety, engagement and boredom will be classified using different signal transformation methods and finally using machine-learning models from the Weka Java API. Data is collected using an Empatica E4 Wristband which gathers skin- and heart related biometric data which is streamed to an Android application via Bluetooth for processing. Several machine-learning algorithms and features were evaluated for best performance.

Esta tese é composta de três artigos relacionados à política monetária e inflação e possuem em comum a ênfase na importância das expectativas tanto para o desenho da política monetária como para a dinâmica inflacionária. No primeiro ensaio, contribuímos para o debate sobre a resposta apropriada de política monetária a flutuações de preços de ativos em um contexto de aprendizagem adaptativa. O modelo conta com dois tipos de regras de juros instrumentais como em Bullard e Mitra (2002), porém com um papel adicional para preços de ativos. Do ponto de vista da E-Estabilidade, conclui-se que uma resposta a preços de ativos não é desejável nem com a regra que utiliza expectativas futuras nem com a regra que responde a valores contemporâneos. Crenças heterogêneas a respeito da dinâmica das flutuações de preços de ativos, inflação e hiato do produto são introduzidas. Também é avaliada uma regra de política monetária ótima que inclui um peso para os preços de ativos. De forma geral, conclui-se que o princípio de Taylor é relevante para todas as regras de juros analisadas e que os bancos centrais devem agir com cautela ao considerar a introdução de preços de ativos na política monetária. No segundo ensaio, oferecemos estimativas recentes de persistência inflacionária no Brasil, com uma abordagem multivariada de componentes não-observados, na qual são consideradas as seguintes fontes que impactam na persistência da inflação: desvios das expectativas da meta real de inflação; persistência dos fatores que provocam inflação; e termos defasados da inflação. Dados de inflação, produto e taxas de juros são decompostos em componentes não-observados e, para simplificar a estimativa de um número grande de variáveis desconhecidas, utilizamos análise bayesiana, seguindo Dossche e Everaert (2005). Os resultados indicam que a persistência baseada em expectativas tem grande participação na persistência inflacionária no Brasil, que tem diminuído nos últimos anos. Tal resultado implica que apenas as tradicionais fricções no ajuste de preços usadas nos modelos macroeconômicos não são suficientes para representar a real persistência da inflação. No último capítulo estimamos diversas curvas de Phillips reduzidas com dados brasileiros recentes, numa abordagem de séries de tempo com componentes não-observados, que se apresenta como alternativa às tradicionais estimativas, baseadas em métodos GMM, de curvas de Phillips Novo-Keynesianas (NKPC), que raramente foram bem sucedidas empiricamente. A decomposição em tendência, sazonalidade e ciclo oferece, através do resultado gráfico, interpretação econômica direta. Diferentemente de Harvey (2011), incluímos expectativas de inflação nas estimações, assim como na NKPC habitual. A inflação no Brasil parece ter respondido cada vez menos às medidas de atividade econômica consideradas. Isso consiste em evidência de achatamento da curva de Phillips no Brasil, o que significa por um lado custos de desinflação mais altos, mas por outro lado menores pressões inflacionárias derivadas de crescimento do produto. / This thesis is composed of three essays on monetary policy and inflation that share particular emphasis on the importance of expectations for both monetary policy design and inflation dynamics. First we contribute to the debate on the appropriate response of monetary policy to asset price fluctuations in an adaptive learning context. Our model accounts for two types of instrumental rules in the spirit of Bullard and Mitra (2002), but with an additional role for asset prices. From the point of view of EStability, we find that a response to stock prices is not desirable under both a forward expectations policy rule and an interest rate rule responding to contemporaneous values. Heterogeneous beliefs about the dynamics of asset price fluctuations, inflation and the output gap are introduced. We also evaluate an optimal monetary policy rule including a weight on asset prices. Overall we find that the Taylor principle remain important over all interest rate rules analysed and that central banks should remain cautious when considering the introduction of stock prices in monetary policy. In the second essay, we provide recent estimates of inflation persistence in Brazil in a multivariate framework of unobserved components, whereby we account for the following sources affecting inflation persistence: First, deviations of expectations from the actual policy target; second, persistence of the factors driving inflation; and third, lagged inflation terms. Data on inflation, output and interest rates are decomposed into unobserved components and to simplify the estimation of a great number of unknown variables, we utilize bayesian analysis as in Dossche and Everaert (2005). Our results indicate that expectations-based persistence matters considerably for inflation persistence in Brazil, which has experienced an overall decrease in the last few years. This finding implies that traditional price-setting frictions used in macroeconomic models are not enough to represent actual inflation persistence. In the last chapter we estimate alternative reduced-form Phillips curves with recent Brazilian data, using a framework of time series with unobserved components, as an alternative to traditional GMM estimations of the New Keynesian Phillips Curve (NKPC), which have seldom been empirically successful. The decomposition into trend, seasonal and cycle features offers, through the graphical output, straightforward economic interpretations. Differently from Harvey (2011), we allow for inflation expectations as in the usual NKPC. Inflation in Brazil seems to have responded gradually less to measures of economic activity in recent years. This provides some evidence of a flattening of the Phillips curve in Brazil, which means higher costs of disinflation on the one hand, but also lower inflationary pressures derived from output growth, on the other.

Cette thèse étudie deux grandes problématiques économiques étant étroitement liées. D’une part, il est question d’analyser à quelles conditions les taux de change présentent des relations de long terme communes. D’autre part, une analyse en profondeur concernant les investissements sur devises connus sous le terme anglais de « carry trades » est proposée. Le taux de change étant un des déterminants du rendement de ces investissements, le lien entre les deux problématiques apparaît clairement. Ces problématiques sont traitées à travers la mobilisation d’outils théoriques et empiriques. Ce travail aboutit à plusieurs conclusions. Concernant les mouvements communs de long terme entre les taux de change, ils dépendent du degré d’intégration des économies ainsi que de la similarité de leurs politiques monétaires. Concernant les investissements sur devises, cette thèse démontre que les banques centrales des petites économies ouvertes ont tout intérêt à fixer une cible d’inflation ainsi qu’une cible d’afflux de capitaux afin d’éviter l’effet déstabilisateur des « carry trades ». Cette politique sera efficace uniquement si la banque centrale est transparente concernant ses cibles de long terme. Pour finir, après la crise financière de 2008, la banque centrale Néo-Zélandaise a changé de comportement vis-à-vis des « carry trades » en provenance du Japon. En effet, après la crise, la banque centrale y a répondu de manière à stabiliser l’économie. Cependant, les investissements en provenance des Etats-Unis sont toujours déstabilisateurs pour l’économie Néo-Zélandaise, surtout lorsque les Etats-Unis utilisent une politique d’assouplissement quantitatif. / This thesis investigates two main issues in economics. On the one hand, we investigate under which conditions cointegration between exchange rates is likely to appear. On the other hand, this thesis proposes to investigate how carry trades affect small open economies. Given that the exchange rate is a main determinant of carry trades’ returns, these two topics are obviously linked. These two issues are investigated both through theoretical and empirical tools. Concerning long run comovements between exchange rates, this thesis reveals that they depend on the degree of linkages between two economies and on the way central banks set their monetary policies. Concerning carry trades, this work sheds light on the fact that small open economies central banks should have both an inflation and a capital inflows target to suppress the destabilizing effect of carry trades. Moreover, such a policy would be efficient only if the central banks are transparent concerning their long run targets. Finally, in this thesis we show that the Reserve Bank of New Zealand (RBNZ) has changed its reaction to Japan-sourced carry trades after the 2008 global financial crisis (GFC). Indeed, after the GFC, the RBNZ responded in a stabilizing way to Japan-sourced carry trades. However, after the GFC, the RBNZ still responded in a destabilizing way to US-sourced carry trades. Our work also reveals that carry trades destabilize even more New-Zealand’s economy when the US are engaged in a quantitative easing policy.

In this thesis, research efforts are dedicated towards the development of practical adaptable techniques to be used in Smart Homes and Buildings, with the aim to improve energy management and conservation, while enhancing the learning capabilities of Programmable Communicating Thermostats (PCT) – “transforming” them into smart adaptable devices, i.e., “Smart Thermostats”. An Adaptable Hybrid Intelligent System utilizing Wireless Sensor Network (WSN) and Artificial Intelligence (AI) techniques is presented, based on which, a novel Adaptive Learning System (ALS) model utilizing WSN, a rule-based system and Adaptive Resonance Theory (ART) concepts is proposed. The main goal of the ALS is to adapt to the occupant’s pattern and/or schedule changes by providing comfort, while not ignoring the energy conservation aspect. The proposed ALS analytical model is a technique which enables PCTs to learn and adapt to user input pattern changes and/or other parameters of interest. A new algorithm for finding the global maximum in a predefined interval within a two dimensional space is proposed. The proposed algorithm is a synergy of reward/punish concepts from the reinforcement learning (RL) and agent-based technique, for use in small-scale embedded systems with limited memory and/or processing power, such as the wireless sensor/actuator nodes. An application is implemented to observe the algorithm at work and to demonstrate its main features. It was observed that the “RL and Agent-based Search”, versus the “RL only” technique, yielded better performance results with respect to the number of iterations and function evaluations needed to find the global maximum. Furthermore, a “House Simulator” is developed as a tool to simulate house heating/cooling systems and to assist in the practical implementation of the ALS model under different scenarios. The main building blocks of the simulator are the “House Simulator”, the “Smart Thermostat”, and a placeholder for the “Adaptive Learning Models”. As a result, a novel adaptive learning algorithm, “Observe, Learn and Adapt” (OLA) is proposed and demonstrated, reflecting the main features of the ALS model. Its evaluation is achieved with the aid of the “House Simulator”. OLA, with the use of sensors and the application of the ALS model learning technique, captures the essence of an actual PCT reflecting a smart and adaptable device. The experimental performance results indicate adaptability and potential energy savings of the single in comparison to the zone controlled scenarios with the OLA capabilities being enabled.

No description available.