Chairi Kiourt

Social Reinforcement Learning in Game Playing

In this work we discuss Social Reinforcement Learning on self-trained agents. We simulate social learning by implementing a tournament on an existing board game that utilizes reinforcement learning for playing and learning. The socially trained agents are compared to self-trained agents and their superior performance is noted. The findings and the infrastructure requirements mandate the development of a Social Reinforcement Learning Multi-Agent-Based Simulation platform.

Building a social multi-agent system simulation management toolbox

The development of a novel Multi-Agent-Based Social Simulation (MABS) platform is undertaken after considering the advantages and disadvantages of existing platforms. We study their adaptability and usage in an existing strategy board game and attempt to model tournaments in social environments. To facilitate this experimentation, we arrive at the need to develop a new platform which features dynamic handling of game objects at runtime.

A platform for large-scale game-playing multi-agent systems on a high performance computing infrastructure

The simulation of societies requires vast amounts of computing resources, which must be managed over distributed or high performance computing infrastructures to provide for cost-effective experimentation. To that end, this paper presents a novel platform for the segmentation and management of social simulation experiments in game-playing multi-agent systems; the platform, also serves as a working proof of concept for similar experiments. The platform is managed through a web-based graphical user interface, to combine the advantages of powerful grid infrastructure middleware and sophisticated workflow systems in a way that some generic functionality is sacrificed for the benefit of obtaining a smooth and brief learning curve, without compromising security. The paper sets out the architecture and implementation details of the platform and demonstrates its use with two sample games, RLGame and Rock Scissors Paper, to underline the scale of the experiments and to indicate the class of social simulation problems that it can help investigate. The platform can be loosely coupled with analytics software for data mining; for our sample problems, this analysis leads to associating the learning mechanism each agent employs with its eventual performance ranking.

Human Rating Methods on Multi-agent Systems

Modern artificial intelligence approaches study game-playing agents in multi-agent social environments, in order to better simulate the real world playing behaviors; these approaches have already produced promising results. In this paper we present the results of applying human rating systems for competitive games with social activity, to evaluate synthetic agents’ performance in multi-agent systems. The widely used Elo and Glicko rating systems are tested in large-scale synthetic multi-agent game-playing social events, and their rating outcome is presented and analyzed.

A dynamic web-based 3D virtual museum framework based on open data

Continuous developments in the web and computer technologies along with an increasing availability of game engines contribute to an expansion of techniques that bridge culture and education with gaming. In addition, open linked data technologies pave the way towards the semantic web of the future by exploiting the abundance in data availability. In this work we present an innovative and content-dynamic web-based framework, which relies and exploits the rich content of distributed web cultural resources and supports the creation of custom virtual exhibitions for cultural and educational purposes based on gaming technologies.

Learning in Multi Agent Social Environments with Opponent Models

We examine how synthetic agents interact in social environments employing a variety of agent training strategies against diverse opponents. Such agent training and playing methods indicate that quality playing relies more on the correct set-up of the learning mechanism than on experience. The experimentation provides valuable insight into the potential of an agent to compete against other agents in its environment and yet manage to also co-operate so that this particular environment allows for the emergence of a competitive champion agent, which will represent its group in further contests. Additionally, by investigating performance while constraining the number of moves we gain interesting insight into competitive learning and playing with resource constraints.

Development of grid-based multi agent systems for social learning

In this paper we present a novel architecture for streamlining human expert management for a multi agent system deployed at the grid with the objective of investigating social learning. The coupling of the communication services offered by the grid, with an administration layer and conventional web server programming, via a data synchronization utility, leads to the straightforward development of a web-based user interface that allows the monitoring and managing of diverse online distributed computing applications. Finally, we present some experimental results to underline the scale of experiments we view as indicative of such a platform. Our implementation has been developed and can be deployed over existing grid infrastructures without compromising security.

ReSkill: Relative Skill-Level Calculation System

The introduction of social dynamics in multi-agent environments with synthetic agents is an effective way to simulate real-life conditions. Nowadays there is a trend towards the integration of social dynamics in multi-agent virtual environments to better assess the performance of synthetic agents in competitive situations. This assessment is usually carried out using human rating methods, such as Elo and Glicko, two of the most widespread methods, primarily used for chess. This paper introduces a web-based system that was developed to provide a way for everyone to be able to use these well-known human rating systems in various multi-agent rating experiments. A large-scale experiment has been conducted and the results have been used to present and prove the functionality of the developed system.

Using opponent models to train inexperienced synthetic agents in social environments

This paper investigates the learning progress of inexperienced agents in competitive game playing social environments. We aim to determine the effect of a knowledgeable opponent on a novice learner. For that purpose, we used synthetic agents whose playing behaviors were developed through diverse reinforcement learning set-ups, such as exploitation-vs-exploration trade-off, learning backup and speed of learning, as opponents, and a self-trained agent. The paper concludes by highlighting the effect of diverse knowledgeable synthetic agents in the learning trajectory of an inexperienced agent in competitive multiagent environments.

Synthetic learning agents in game-playing social environments

This paper investigates the performance of synthetic agents in playing and learning scenarios in a turn-based zero-sum game and highlights the ability of opponent-based learning models to demonstrate competitive playing performances in social environments. Synthetic agents are generated based on a variety of combinations of some key parameters, such as exploitation-vs-exploration trade-off, learning back-up and discount rates, and speed of learning, and interact over a very large number of games on a grid infrastructure; experimental data is then analysed to generate clusters of agents that demonstrate interesting associations between eventual performance ranking and learning parameters’ set-up. The evolution of these clusters indicates that agents with a predisposition to knowledge exploration and slower learning tend to perform better than exploiters, which tend to prefer fast learning. Observing these clusters vis-à-vis the playing behaviours of the agents makes it also possible to investigate how to select opponents best from a group; initial results suggest that good progress and stable evolution arise when an agent faces opponents of increasing capacity, and that an agent with a good learning mechanism set-up progresses better when it faces less favourably set-up agents.