人們如何在重覆的囚犯困境賽局選擇策略是本文探討的議題，其中的賽局學習理論就是預測賽局的參與者(player)會選擇何種策略。本文使用的資料包括3個囚犯困境的實驗，各自有不同的實驗設定及配對程序，參加者都是政治大學的大學部學生，我們將使用這些資料比較不同的學習模型。除了常見的3個學習模型：增強學習模型(Reinforcement Learning model)、信念學習模型(Belief Learning model)及加權經驗吸引模型(Experience-Weighted Attraction model)，本文也提出一個延伸的增強學習模型(Extended reinforcement learning model)。接著將分析劃為Training (in-sample)及Testing (out-sample)，並比較各實驗間或模型間的結果。 　　雖然延伸增強學習模型(Extended reinforcement learning model)較原始的增強學習模型(Reinforcement learning model)多了一個參數，該模型(Extended reinforcement learning model)在Training(in-sample)及Testing(out-sample)表現多較之前的模型來得些許的好。 / How people choose strategies in a finite repeated prisoners’ dilemma game is of interest in Game Theory. The way to predict which strategies the people choose in a game is so-called game learning theory. The objective of this study is to find a proper learning model for the prisoners’ dilemma game data collected in National Cheng-Chi University. The game data consist of three experiments with different game and matching rules. Four learning models are considered, including Reinforcement learning model, Belief learning model, Experience Weighted Attraction learning model and a proposed model modified from reinforcement learning model. The data analysis was divided into 2 parts: training (in-sample) and testing (out-sample). The proposed learning model is slightly better than the original reinforcement learning model no matter when in training or testing prediction although one more parameter is added. The performances of prediction by model fitting are all better than guessing the decisions with equal chance.

Hierarchical task decompositions play an essential role in the design of complex simulation and decision systems, such as the ones that arise in video games. Game designers find it very natural to adopt a divide-and-conquer philosophy of specifying hierarchical policies, where decision modules can be constructed somewhat independently. The process of choosing the parameters of these modules manually is typically lengthy and tedious. The hierarchical reinforcement learning (HRL) field has produced elegant ways of decomposing policies and value functions using semi-Markov decision processes. However, there is still a lack of demonstrations in larger nonlinear systems with discrete and continuous variables. To narrow this gap between industrial practices and academic ideas, we address the problem of designing efficient algorithms to facilitate the deployment of HRL ideas in more realistic settings. In particular, we propose Bayesian active learning methods to learn the relevant aspects of either policies or value functions by focusing on the most relevant parts of the parameter and state spaces respectively. To demonstrate the scalability of our solution, we have applied it to The Open Racing Car Simulator (TORCS), a 3D game engine that implements complex vehicle dynamics. The environment is a large topological map roughly based on downtown Vancouver, British Columbia. Higher level abstract tasks are also learned in this process using a model-based extension of the MAXQ algorithm. Our solution demonstrates how HRL can be scaled to large applications with complex, discrete and continuous non-linear dynamics.

The past few years have witnessed a significant evolution in mobile computing and communications, in which new trends and applications have the traditional role of computer networks into that of distributed service providers. In this thesis we explore an alternative way to form wireless ad-hoc networks whose topologies can be customized as required by the users’ software applications. In particular, we investigate the applicability of mobile codes to networks created by devices equipped with Bluetooth technology. Computer simulations results suggest that our proposed approach can achieve this task effectively, while matching the level of efficiency seen in other salient proposals in this area. This thesis also addresses the issue of service discovery in mobile ad-hoc networks. We propose the use of a directory whose network location varies in an attempt to reduce traffic overhead driven by users’ hosts looking for service information. We refer to this scheme as the Service Directory Placement Algorithm, or SDPA. We formulate the directory relocation problem as a Markov Decision Process that is solved by using Q-learning. Performance evaluations through computer simulations reveal bandwidth overhead reductions that range between 40% and 48% when compared with a basic broadcast flooding approach for networks comprising hosts moving at pedestrian speeds. We then extend our proposed approach and introduce a multi-directory service discovery system called the Service Directory Placement Protocol, or SDPP. Our findings reveal bandwidth overhead reductions typically ranging from 15% to 75% in networks comprising slow-moving hosts with restricted memory availability. In the fourth and final part of this work, we present the design foundations and architecture of a middleware system that called WISEMAN – WIreless Sensors Employing Mobile Agents. We employ WISEMAN for dispatching and processing mobile programs in Wireless Sensor Networks (WSNs). Our proposed system enables the dynamic creation of semantic relationships between network nodes that cooperate to provide an aggregate service. We present discussions on the advantages of our proposed approach, and in particular, how WISEMAN facilitates the realization of service-oriented tasks in WSNs.

This Ph.D. thesis presents the design and implementation of a biologically inspired four-phase walking strategy using behaviours for a four legged walking robot. In particular, the walking strategy addresses the balance issue, including both static and dynamic balance that were triggered non-deterministically based on the robot’s realtime interaction with the environment. Four parallel Subsumption Architectures (SA) and a simple Central Pattern Producer (CPP) are employed in the physical implementation of the walking strategy. An implementation framework for such a parallel Subsumption Architecture is also proposed to facilitate the reusability of the system. A Reinforcement Learning (RL) method was integrated into the CPP to allow the robot to learn the optimal walking cycle interval (OWCI), appropriate for the robot walking on various terrain conditions. Experimental results demonstrate that the robot employs the proposed walking strategy and can successfully carry out its walking behaviours under various experimental terrain conditions, such as flat ground, incline, decline and uneven ground. Interactions of all the behaviours of the robot enable it to exhibit a combination of both preset and emergent walking behaviours.

Thesis (M.S.)--University of Missouri--Rolla, 2007. / Vita. The entire thesis text is included in file. Title from title screen of thesis/dissertation PDF file (viewed May 16, 2007) Includes bibliographical references.

Research in mobile ad-hoc networks has focused on situations in whichnodes have no control over their movements. We investigate animportant but overlooked domain in which nodes do have controlover their movements. Reinforcement learning methods can be used tocontrol both packet routing decisions and node mobility, dramaticallyimproving the connectivity of the network. We first motivate theproblem by presenting theoretical bounds for the connectivityimprovement of partially mobile networks and then present superiorempirical results under a variety of different scenarios in which themobile nodes in our ad-hoc network are embedded with adaptive routingpolicies and learned movement policies.

We perceive the environment via sensor arrays and interact with it through motor outputs. The work of this thesis concerns how the brain selects actions given the information about the perceived state of the world and how it learns and adapts these selections to changes in this environment. Reinforcement learning theories suggest that an action will be more or less likely to be selected if the outcome has been better or worse than expected. A group of subcortical structures, the basal ganglia (BG), is critically involved in both the selection and the reward prediction. We developed and investigated a computational model of the BG. We implemented a Bayesian-Hebbian learning rule, which computes the weights between two units based on the probability of their activations. We were able test how various configurations of the represented pathways impacted the performance in several reinforcement learning and conditioning tasks. Then, following the development of a more biologically plausible version with spiking neurons, we simulated lesions in the different pathways and assessed how they affected learning and selection. We observed that the evolution of the weights and the performance of the models resembled qualitatively experimental data. The absence of an unique best way to configure the model over all the learning paradigms tested indicates that an agent could dynamically configure its action selection mode, mainly by including or not the reward prediction values in the selection process. We present hypotheses on possible biological substrates for the reward prediction pathway. We base these on the functional requirements for successful learning and on an analysis of the experimental data. We further simulate a loss of dopaminergic neurons similar to that reported in Parkinson’s disease. We suggest that the associated motor symptoms are mostly causedby an impairment of the pathway promoting actions, while the pathway suppressing them seems to remain functional. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 3: Manuscript.</p><p> </p>

Neste trabalho será descrita uma arquitetura de agente robótico autônomo projetada para ser capaz de criar uma representação de estado do ambiente e de realizar o aprendizado de tarefas simples em cima desta representação. A rede GTSOM (BASTOS, 2007) foi selecionada como método para classificação de estados. Sua tarefa é transformar os dados multidimensionais e contínuos lidos dos sensores em uma representação discreta, permitindo o uso de aprendizado por reforço convencional. Algumas modificações no algoritmo da rede foram necessárias para que pudesse ser aplicada neste contexto. Juntamente com esta rede, foi utilizado um mapa de grade que permite associar as experiências sensoriais com sua localização espacial. Enquanto a rede GTSOM é o ponto central de um sistema de classificação de estados, o algoritmo Q-Learning de aprendizado por reforço foi utilizado para a realização da tarefa. Utilizando a representação compacta de estado criada pela rede auto-organizável, o agente aprende as ações que devem ser executadas em cada ponto, para atingimento de seus objetivos. O modelo foi testado com um experimento que consiste em encontrar um objeto em um labirinto. Os resultados obtidos nos testes mostraram que o modelo consegue segmentar adequadamente o espaço de estados, e realiza o aprendizado da tarefa. O agente consegue aprender a evitar colisões e memorizar a localização do alvo, podendo chegar até ele independentemente de sua posição inicial. Além disso, é capaz de expandir sua representação sempre que se depara com situações não conhecidas, ao mesmo tempo que gradualmente remove da memória estados associados a experiências que não se repetem. / This work describes an architecture for an autonomous robotic agent that is capable of creating a state representation of its environment and learning how to execute simple tasks using this representation. The GTSOM Neural Network was chosen as the method for state clustering. It is used to transform the multidimensional and continuous state signal into a discrete representation, allowing the use of conventional reinforcement learning techniques. Some modifications on the algorithm were necessary so that it could be used in this project. This network is used together with a grid map algorithm that allows the model to associate the sensor readings with the places where they ocurred. While the GTSOM network is the main component of a state clustering system, the Q-Learning reinforcement learning method was chosen for the task execution. Using the compact state representation created by the self-organizing network, the agent learns which actions to execute at each state in order to achieve its objectives. The model was tested in an experiment that consists in finding the path in a maze. The results show that it can divide the state space in an useful way, and is capable of executing the task. It learns to avoid collisions and remembers the location of the target, even when the robot’s initial position is changed. Furthermore, the representation is expanded when the agent faces an unknown situation, and at the same time, states associated with old experiences are forgotten.

An n-armed bandit task was used to investigate the trade-off between exploratory (choosing lesser-known options) and exploitive (choosing options with the greatest probability of reinforcement) human choice in a trial-and-error learning problem. In Experiment 1 a different probability of reinforcement was assigned to each of 8 response options using random-ratios (RRs), and participants chose by clicking buttons in a circular display on a computer screen using a computer mouse. Relative frequency thresholds (ranging from .10 to 1.0) were randomly assigned to each participant and acted as task constraints limiting the proportion of total responses that could be attributed to any response option. Preference for the richer keys was shown, and those with greater constraints explored more and earned less reinforcement. Those with the highest constraints showed no preference, distributing their responses among the options with equal probability. In Experiment 2 the payoff probabilities changed partway through, for some the leanest options increased to richest, and for others the richest became leanest. When the RRs changed, the decrease participants with moderate and low constraints showed immediate increases in exploration and change in preference to the new richest keys, while increase participants showed no increase in exploration, and more gradual changes in preference. For Experiment 3 the constraint was held constant at .85, and the two richest options were decreased midway through the task by varying amounts (0 to .60). Decreases were detected early for participants in all but the smallest decrease conditions, and exploration increased.

This dissertation describes the development and optimisation of a set of touchscreen tasks for the assessment of motivation and reinforcement processes in mice. Changes in motivational state and reinforcement sensitivity are characteristic of depressive symptoms in a number of neuropsychiatric and neurodegenerative conditions. However, the efficacy of current therapeutic strategies is limited. Thus, the development of methods for studying these processes is important both to facilitate investigation of underlying mechanisms and subsequently develop targeted therapeutics. Part 1 describes the optimisation and application of fixed and progressive ratio tasks for the assessment of motivational state in mice. First, a comparison of multiple reinforcers under fixed and progressive ratio schedules is described. Given the limitations of performance measures derived from performance across the entire session, these experiments are complemented by the use of response rate analysis, which reveals dissociations between standard measures and rate of responding. Next, these tasks are applied to a model of maternal high fat diet (hfd), revealing selective alterations in motivated behaviour. Finally, the motivational state of the TgR406W mouse model of frontotemporal dementia, a condition frequently characterised by depressive symptoms, is investigated. It is shown that the motivational state of TgR406W mice is age dependent, with apparent early apathy-like behaviour and a late disinhibition in responding. This is accompanied by deficits in attentional performance as measured by a 3-choice serial reaction time task. Thus, part 1 demonstrates the applicability of the battery of touchscreen tasks for assessment of motivation to diverse mouse models. Part 2 describes the development and validation of touchscreen tasks for the assessment of reinforcement-related choice in mice. First, it is demonstrated that C57BL/6 mice systematically discount a preferred reinforcer in a delay discounting procedure. Subsequently, choice behaviour is modified by systemic treatment of dopaminergic drugs and a reinforcer pre-feeding procedure. Next, procedures for the assessment of learning and choice, dependent on the integration of positive and negative feedback, are described. Specifically, the valence-probe visual-discrimination (VPVD) task is an adapted version of the standard visual discrimination with the addition of a probabilistic stimulus that is reinforced 50% of the time. This modification allows for assessment of the impact of positive and negative feedback on discrimination learning. In addition, a spatial within-session probabilistic reversal learning procedure was adapted for the touchscreen apparatus. The effect of serotonin 2C receptor manipulation on these tasks is presented. Serotonin 2C receptor antagonism is shown to disrupt discrimination learning and result in diminished sensitivity to positive feedback. Conversely, 2C receptor agonism results in changes in feedback sensitivity consistent with heightened sensitivity to positive feedback and diminished sensitivity to negative feedback. Taken together, these results show that these tasks can be used to interrogate reinforcement-related behavioural profiles in experimental mice. Finally, the implications of these tasks with respect to investigation of motivation, reinforcement sensitivity and adaptive decision-making in rodent models, and their relevance to MDD in humans is discussed.