Reinforcement Learning by Policy Search

One objective of artificial intelligence is to model the behavior of an intelligent agent interacting with its environment. The environment's transformations can be modeled as a Markov chain, whose state is partially observable to the agent and affected by its actions; such processes are known as partially observable Markov decision processes (POMDPs). While the environment's dynamics are assumed to obey certain rules, the agent does not know them and must learn. In this dissertation we focus on the agent's adaptation as captured by the reinforcement learning framework. This means learning a policy---a mapping of observations into actions---based on feedback from the environment. The learning can be viewed as browsing a set of policies while evaluating them by trial through interaction with the environment. The set of policies is constrained by the architecture of the agent's controller. POMDPs require a controller to have a memory. We investigate controllers with memory, including controllers with external memory, finite state controllers and distributed controllers for multi-agent systems. For these various controllers we work out the details of the algorithms which learn by ascending the gradient of expected cumulative reinforcement. Building on statistical learning theory and experiment design theory, a policy evaluation algorithm is developed for the case of experience re-use. We address the question of sufficient experience for uniform convergence of policy evaluation and obtain sample complexity bounds for various estimators. Finally, we demonstrate the performance of the proposed algorithms on several domains, the most complex of which is simulated adaptive packet routing in a telecommunication network.

Studies of dynamics of physical agent ecosystems

This thesis addresses the problem of learning in physical heterogeneous multi-agent systems (MAS) and the analysis of the benefits of using heterogeneous MAS with respect to homogeneous ones. An algorithm is developed for this task; building on a previous work on stability in distributed systems by Tad Hogg and Bernardo Huberman, and combining two phenomena observed in natural systems, task partition and hierarchical dominance. This algorithm is devised for allowing agents to learn which are the best tasks to perform on the basis of each agent's skills and the contribution to the team global performance. Agents learn by interacting with the environment and other teammates, and get rewards from the result of the actions they perform. This algorithm is specially designed for problems where all robots have to co-operate and work simultaneously towards the same goal. One example of such a problem is role distribution in a team of heterogeneous robots that form a soccer team, where all members take decisions and co-operate simultaneously. Soccer offers the possibility of conducting research in MAS, where co-operation plays a very important role in a dynamical and changing environment. For these reasons and the experience of the University of Girona in this domain, soccer has been selected as the test-bed for this research. In the case of soccer, tasks are grouped by means of roles. One of the most interesting features of this algorithm is that it endows MAS with a high adaptability to changes in the environment. It allows the team to perform their tasks, while adapting to the environment. This is studied in several cases, for changes in the environment and in the robot's body. Other features are also analysed, especially a parameter that defines the fitness (biological concept) of each agent in the system, which contributes to performance and team adaptability. The algorithm is applied later to allow agents to learn in teams of homogeneous and heterogeneous robots which roles they have to select, in order to maximise team performance. The teams are compared and the performance is evaluated in the games against three hand-coded teams and against the different homogeneous and heterogeneous teams built in this thesis. This section focuses on the analysis of performance and task partition, in order to study the benefits of heterogeneity in physical MAS. In order to study heterogeneity from a rigorous point of view, a diversity measure is developed building on the hierarchic social entropy defined by Tucker Balch. This is adapted to quantify physical diversity in robot teams. This tool presents very interesting features, as it can be used in the future to design heterogeneous teams on the basis of the knowledge on other teams.

Landscape of intelligent cores: An autonomic multi-agent approach for space application

Space applications are challenged by the reliability of parallel computing systems (FPGAs) employed in space crafts due to Single-Event Upsets. The work reported in this paper aims to achieve self-managing systems which are reliable for space applications by applying autonomic computing constructs to parallel computing systems. A novel technique, 'Swarm-Array Computing' inspired by swarm robotics, and built on the foundations of autonomic and parallel computing is proposed as a path to achieve autonomy. The constitution of swarm-array computing comprising for constituents, namely the computing system, the problem / task, the swarm and the landscape is considered. Three approaches that bind these constituents together are proposed. The feasibility of one among the three proposed approaches is validated on the SeSAm multi-agent simulator and landscapes representing the computing space and problem are generated using the MATLAB.

Blind channel estimation in multi-rate multicarrier DS-CDMA systems

Multi-rate multicarrier DS-CDMA is a potentially attractive multiple access method for future broadband wireless multimedia networks that must support integrated voice/data traffic. This paper proposes a subspace based channel estimation scheme for multi-rate multicarrier DS-CDMA, which is applicable to both multicode and variable spreading factor systems. The performance of the proposed scheme for these two multi-rate systems is compared via numerical simulations.

Modelling phosphorus dynamics in multi-branch river systems: A study of the Black River, Lake Simcoe, Ontario, Canada

High rates of nutrient loading from agricultural and urban development have resulted in surface water eutrophication and groundwater contamination in regions of Ontario. In Lake Simcoe (Ontario, Canada), anthropogenic nutrient contributions have contributed to increased algal growth, low hypolimnetic oxygen concentrations, and impaired fish reproduction. An ambitious programme has been initiated to reduce phosphorus loads to the lake, aiming to achieve at least a 40% reduction in phosphorus loads by 2045. Achievement of this target necessitates effective remediation strategies, which will rely upon an improved understanding of controls on nutrient export from tributaries of Lake Simcoe as well as improved understanding of the importance of phosphorus cycling within the lake. In this paper, we describe a new model structure for the integrated dynamic and process-based model INCA-P, which allows fully-distributed applications, suited to branched river networks. We demonstrate application of this model to the Black River, a tributary of Lake Simcoe, and use INCA-P to simulate the fluxes of P entering the lake system, apportion phosphorus among different sources in the catchment, and explore future scenarios of land-use change and nutrient management to identify high priority sites for implementation of watershed best management practises.

Multi-agent based modeling and simulation of complex system in hospital

This study puts forward a method to model and simulate the complex system of hospital on the basis of multi-agent technology. The formation of the agents of hospitals with intelligent and coordinative characteristics was designed, the message object was defined, and the model operating mechanism of autonomous activities and coordination mechanism was also designed. In addition, the Ontology library and Norm library etc. were introduced using semiotic method and theory, to enlarge the method of system modelling. Swarm was used to develop the multi-agent based simulation system, which is favorable for making guidelines for hospital's improving it's organization and management, optimizing the working procedure, improving the quality of medical care as well as reducing medical charge costs.

Is the unfolding of the group discussion off-pattern? Improving coordination support in educational forums using mobile devices

A forum is a valuable tool to foster reflection in an in-depth discussion; however, it forces the course mediator to continually pay close attention in order to coordinate learners` activities. Moreover, monitoring a forum is time consuming given that it is impossible to know in advance when new messages are going to be posted. Additionally, a forum may be inactive for a long period and suddenly receive a burst of messages forcing forum mediators to frequently log on in order to know how the discussion is unfolding to intervene whenever it is necessary. Mediators also need to deal with a large amount of messages to identify off-pattern situations. This work presents a piece of action research that investigates how to improve coordination support in a forum using mobile devices for mitigating mediator`s difficulties in following the status of a forum. Based on summarized information extracted from message meta-data, mediators consult visual information summaries on PDAs and receive textual notifications in their mobile phone. This investigation revealed that mediators used the mobile-based coordination support to keep informed on what is taking place within the forum without the need to log on their desktop computer. (C) 2009 Elsevier Ltd. All rights reserved.

Agent-Oriented Data Curation in Bioinformatics

The practitioners of bioinformatics require increasing sophistication from their software tools to take into account the particular characteristics that make their domain complex. For example, there is a great variation of experience of researchers, from novices who would like guidance from experts in the best resources to use to experts that wish to take greater management control of the tools used in their experiments. Also, the range of available, and conflicting, data formats is growing and there is a desire to automate the many trivial manual stages of in-silico experiments. Agent-oriented software development is one approach to tackling the design of complex applications. In this paper, we argue that, in fact, agent-oriented development is a particularly well-suited approach to developing bioinformatics tools that take into account the wider domain characteristics. To illustrate this, we design a data curation tool, which manages the format of experimental data, extend it to better account for the extra requirements placed by the domain characteristics, and show how the characteristics lead to a system well suited to an agent-oriented view.

Multi-Agent Simulation of Urban Social Dynamics for Spatial Load Forecasting

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

A framework for learning in humanoid simulated robots

One of the most important characteristics of intelligent activity is the ability to change behaviour according to many forms of feedback. Through learning an agent can interact with its environment to improve its performance over time. However, most of the techniques known that involves learning are time expensive, i.e., once the agent is supposed to learn over time by experimentation, the task has to be executed many times. Hence, high fidelity simulators can save a lot of time. In this context, this paper describes the framework designed to allow a team of real RoboNova-I humanoids robots to be simulated under USARSim environment. Details about the complete process of modeling and programming the robot are given, as well as the learning methodology proposed to improve robot's performance. Due to the use of a high fidelity model, the learning algorithms can be widely explored in simulation before adapted to real robots. © 2008 Springer-Verlag Berlin Heidelberg.