Papers by Eugénio Oliveira
Artificial intelligence for engineering design, analysis and manufacturing, May 2, 2016
The work described in this paper is part of the development of a framework to support the joint e... more The work described in this paper is part of the development of a framework to support the joint execution of cooperative missions by a group of vehicles, in a simulated, augmented, or real environment. Such a framework brings forward the need for formal languages in which to specify the vehicles that compose a team, the scenario in which they will operate, and the mission to be performed. This paper introduces the Scenario Description Language (SDL) and the Team Description Language (TDL), two Extensible Markup Language based dialects that compose the static components necessary for representing scenario and mission knowledge. SDL provides a specification of physical scenario and global operational constraints, while TDL defines the team of vehicles, as well as team-specific operational restrictions. The dialects were defined using Extensible Markup Language schemas, with all required information being integrated in the definitions. An interface was developed and incorporated into the framework, allowing for the creation and edition of SDL and TDL files. Once the information is specified, it can be used in the framework, thus facilitating environment and team specification and deployment. A survey answered by practitioners and researchers shows that the satisfaction with SDL+TDL is elevated (the overall evaluation of SDL+TDL achieved a score of 4 out of 5, with 81%/78.6% of the answers ≥4); in addition, the usability of the interface was evaluated, achieving a score of 86.7 in the System Usability Scale survey. These results imply that SDL+TDL is flexible enough to represent scenarios and teams, through a user-friendly interface.
We propose a learning mechanism suitable for using in a heterogeneous environment, where intercon... more We propose a learning mechanism suitable for using in a heterogeneous environment, where interconnected systems present simultaneously a competitive and cooperative behaviour. By interconnected systems we understand systems that need to interact with each other and then act either in an individual as well as a global way: each of the systems observes itself and the environment, and acts according to its unique perspective; but a system can also need the help of others to complete its own tasks. In our case, systems are represented by agents, since we are studying environments where autonomy and pro-activity are essential features. The learning mechanism we propose here can be classified as a multi-agent learning mechanism, not only because there are multiple agents learning concurrently in the same environment but also because it allows them to understand how to improve their own performance without degrading the performance of the other agents. We tested our learning mechanism over the Disruption Management in Airline Operations Control Center application domain, where interconnected agents (namely the Aircraft Manager, the Passenger Manager and the Crew Manager) need to combine their expertise. The results show that our learning mechanism provides a good performance to the agents that operate simultaneously in cooperative and competitive situations. 4
Springer eBooks, 2005
One of the main questions concerning learning in Multi-Agent Systems is: "(How) can agents benefi... more One of the main questions concerning learning in Multi-Agent Systems is: "(How) can agents benefit from mutual interaction during the learning process?". This paper describes the study of an interactive advice-exchange mechanism as a possible way to improve agents' learning performance. The advice-exchange technique, discussed here, uses supervised learning (backpropagation), where reinforcement is not directly coming from the environment but is based on advice given by peers with better performance score (higher confidence), to enhance the performance of a heterogeneous group of Learning Agents (LAs). The LAs are facing similar problems, in an environment where only reinforcement information is available. Each LA applies a different, well known, learning technique: Random Walk, Simulated Annealing, Evolutionary Algorithms and Q-Learning. The problem used for evaluation is a simplified traffic-control simulation. In the following text the reader can find a description of the traffic simulation and Learning Agents (focused on the advice-exchange mechanism), a discussion of the first results obtained and suggested techniques to overcome the problems that have been observed. Initial results indicate that advice-exchange can improve learning speed, although "bad advice" and/or blind reliance can disturb the learning performance. The use of supervised learning to incorporate advice given from non-expert peers using different learning algorithms, in problems where no supervision information is available, is, to the best of the authors' knowledge, a new concept in the area of Multi-Agent Systems Learning. This work aims at contributing to answer the question: "(How) can agents benefit from mutual interaction during the learning process, in order to achieve better individual and overall system performances?". This question has been deemed a "challenging issue" by several authors in recently published work (
Springer eBooks, 2018
A travel agency has recently proposed the Traveling Salesman Challenge (TSC), a problem consistin... more A travel agency has recently proposed the Traveling Salesman Challenge (TSC), a problem consisting of finding the best flights to visit a set of cities with the least cost. Our approach to this challenge consists on using a meta-optimized Ant Colony Optimization (ACO) strategy which, at the end of each iteration, generates a new "ant" by running Simulated Annealing or applying a mutation operator to the best "ant" of the iteration. Results are compared to variations of this algorithm, as well as to other meta-heuristic methods. They show that the developed approach is a better alternative than regular ACO for the time-dependent TSP class of problems, and that applying a K-Opt optimization will usually improve the results.
Adaptive Agents and Multi-Agents Systems, Jul 19, 2004
We consider the problem of learning accurate models from multiple sources of "nearby" data. Given... more We consider the problem of learning accurate models from multiple sources of "nearby" data. Given distinct samples from multiple data sources and estimates of the dissimilarities between these sources, we provide a general theory of which samples should be used to learn models for each source. This theory is applicable in a broad decision-theoretic learning framework, and yields results for classification and regression generally, and for density estimation within the exponential family. A key component of our approach is the development of approximate triangle inequalities for expected loss, which may be of independent interest.
TraSMAPI (Traffic Simulation Manager Application Programming Interface) is designed to provide re... more TraSMAPI (Traffic Simulation Manager Application Programming Interface) is designed to provide real-time interaction with Traffic Simulators, collect relevant metrics and statistics, and offer an integrated framework to develop Multi-Agent Systems. It is presented as a tool for the simulation of dynamic control systems in road networks with special focus on Multi-Agent Systems. The abstraction over the simulator opens up the possibility of running different traffic simulators using the same API (application programming interface) allowing the comparison of results of the same application in different simulators. The proposed approach is, therefore, expected to be a key asset in supporting and enhancing engineers and practitioners to make more effective control decisions and implement more efficient management policies while analyzing and addressing traffic related problems in urban areas.
IGI Global eBooks, Jan 18, 2011
Abstract Trading off between realism and too much abstraction is an important issue to address in... more Abstract Trading off between realism and too much abstraction is an important issue to address in microscopic traffic simulation. In this chapter the authors bring this discussion forward and propose a multi-agent model of the traffic domain where integration is ascribed to the way the environment is represented and agents interoperate. While most approaches still deal with drivers and vehicles indistinguishably, in the proposed framework vehicles are merely moveable objects whereas the driving role is played by agents fully endowed with ...
In B2B contract enactment, cooperation should be taken into account when modeling contractual com... more In B2B contract enactment, cooperation should be taken into account when modeling contractual commitments through obligations. We advocate a directed deadline obligation approach, taking inspiration on international legislation over trade procedures. Our proposal is based on authorizations granted in specific states of an obligation lifecycle model. Flexible deadlines provide an additional level of cooperation between contractual agents. Moreover, agents increase their decision-making options concerning obligations.
Lecture Notes in Computer Science, 2019
In this paper we propose the adoption of Agent Process Modelling, a theoretical-practical framewo... more In this paper we propose the adoption of Agent Process Modelling, a theoretical-practical framework for the orchestration of agent behaviours running process models. The agent model is revisited from a distributed, cloud-native perspective and agents’ capabilities are externalized as microservices. Agent logic is represented in process models that orchestrate services execution by using a standard process notation, BPMN, increasing explainability and verification of agents decisions. As a preliminary work, we demonstrate a possible workflow to design and run agent-processes by instantiating the FIPA ContractNet protocol.
Studies in computational intelligence, Oct 18, 2015
Rework Management in software development is a challenging and complex issue. Defined as the effo... more Rework Management in software development is a challenging and complex issue. Defined as the effort spent to redo some work, rework implies big costs given the fact that the time spent on rework does not count to the improvement of the project. Predicting and controlling rework causes is a valuable asset for companies, which maintain closed policies on choosing team members and assigning activities to developers. However, a trending growth in development consists in Open Source Software (OSS) projects. This is a totally new and diverse environment, in the sense that not only the projects but also their resources, e.g., developers change dynamically. There is no guarantee that developers will follow the same methodologies and quality policies as in a traditional and closed project. In such world, identifying rework causes is a necessary step to reduce project costs and to help project managers to better define their strategies. We observed that in real OSS projects there are no fixed team, but instead, developers assume some kind of auction in which the activities are assigned to the most interested and less-cost developer. This lead us to think that a more complex auctioning mechanism should not only model the task allocation problem, but also consider some other factors related to rework causes. By doing this, we could optimise the task allocation, improving the development of the project and reducing rework. In this paper we presented MAESTROS, a Multi-Agent System that implements an auction mechanism for simulating task allocation in OSS. Experiments were conducted to measure costs and rework with different project characteristics. We analysed the impact of introducing a Q-learning reinforcement algorithm on reducing costs and rework. Our findings correspond to
Lecture Notes in Computer Science, 2017
The use of general descriptive names, registered names, trademarks, service marks, etc. in this p... more The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Developments and Advances in Intelligent Systems and Applications, 2017
Elevation data is important for precise electric vehicle simulation. However, traffic simulators ... more Elevation data is important for precise electric vehicle simulation. However, traffic simulators are often strictly two-dimensional and do not offer the capability of modelling urban networks taking elevation into account. In particular, SUMO—Simulation of Urban Mobility, a popular microscopic traffic simulation platform, relies on urban networks previously modelled with elevation data in order to use this information during simulations. This work tackles the problem of how to add this elevation data to urban network models—in particular for the case of the Porto urban network, in Portugal. With this goal in mind, a comparison between altitude information retrieval approaches is made and a tool to annotate network models with altitude data is proposed. This paper starts by describing the methodological approach followed to develop the work, then describing and analysing its main findings, including an in-depth explanation of the proposed tool.
2016 11th Iberian Conference on Information Systems and Technologies (CISTI), 2016
Elevation data is very important for precise electric vehicle simulation. However, traffic simula... more Elevation data is very important for precise electric vehicle simulation. However, traffic simulators are often strictly two-dimensional and do not offer the capability of modelling urban networks taking elevation into account. In particular, SUMO - Simulation of Urban Mobility, an often used microscopic traffic simulation platform, relies on urban networks previously modelled with elevation data in order to provide access to this information during simulations. This work intends on tackling the problem of how to add this elevation data to urban network models - in particular for the case of the Porto urban network, in Portugal. With this goal in mind, a comparison between different altitude information retrieval approaches is made and a simple tool to annotate network models with altitude data is proposed. This paper starts by describing the methodological approach followed to develop the work, then describing and analysing its main findings. This description includes an in-depth explanation of the proposed tool. Lastly, this paper reviews some related work to this subject.
The autonomous agents and multi-agent systems domain was very active and fruitful along the last ... more The autonomous agents and multi-agent systems domain was very active and fruitful along the last decade, mainly due to the community research efforts, with the organization of more than 50 workshops and a yearly major international conference (AAMAS). Moreover, the domain has reached the 3rd. place in IJCAI 2009 and the 2nd. place in ECAI 2010 in the number of accepted full papers, thus revealing its high relevance within the mainstream current research in the major field of Artificial Intelligence. In this paper, we try to cover its five key elements (agents, environments, interactions, organizations and users), after presenting a brief sketch of its historical milestones. We conclude by pointing out the future aims of research and the right place of negotiation and argumentation within the context of the domain.
This paper reports on a novel method to explore and map an entirely unknown network using a coope... more This paper reports on a novel method to explore and map an entirely unknown network using a cooperative Multi-Agent System (MAS) to extract knowledge or information from nodes and connections. We consider the likely presence of obstacles, eventually making the network disconnected. The MAS architecture is applicable to a vast range of scenarios. Our main goal is to discover the entire network as quickly as possible, characterizing its nodes' meta-structures. In this paper, we propose a novel method that relies on agents that can communicate to each other through simple messages, ensuring that there is no resource sharing. The proposed method is compared to other two non-cooperative methods through simulation, in order to establish a basis for comparison. Preliminary results show that our cooperative approach produces better results than the other two implemented and guarantees that the entire network is explored at the end.
Lecture Notes in Computer Science, 2019
The fast pace in blockchain development introduced the concept of Smart Contract as a way to depl... more The fast pace in blockchain development introduced the concept of Smart Contract as a way to deploy digital contracts as pieces of code on the blockchain network. Smart contracts differ from the traditional ones in many aspects related to their automated execution. Still, there are many open issues, such as legal frameworks and regulatory aspects. The study of agreement technologies and agent-based market systems have already dealt with some of these issues and can help us discuss solutions for the future. In this paper, we present an overview of the agreement pipeline, considering regulation-enabled systems and compare the implementation of auction agreements purely on the blockchain with a hybrid approach where markets are active agents that can ensure regulation outside the contractual phase and contract execution runs on blockchain. We have compared the cost of failing negotiations, that is always charged in the blockchain case.
Advances in distributed computing and artificial intelligence journal, Jul 1, 2013
Multi-agent systems Computational Trust Normative Control Betrayal Despite relevant insights from... more Multi-agent systems Computational Trust Normative Control Betrayal Despite relevant insights from socio-economics, little research in multi-agent systems has addressed the interconnections between trust and normative notions such as contracts and sanctions. Focusing our attention on scenarios of betrayal, in this paper we combine the use of trust and sanctions in a negotiation process. We describe a scenario of dyadic relationships between truster agents, which make use of trust and/or sanctions, and trustee agents, characterized by their ability and integrity, which may influence their attitude toward betrayal. Both agent behavior models are inspired in socio-economics literature. Through simulation, we show the virtues and shortcomings of using trust, sanctions, and a combination of both in processes of selection of partners.
The multi-agent system (MAS) paradigm has become a prominent approach in distributed artificial i... more The multi-agent system (MAS) paradigm has become a prominent approach in distributed artificial intelligence. Many real-world applications of MAS require ensuring cooperative outcomes in scenarios populated with self-interested agents. Following this concern, a strong research emphasis has been given recently to normative MAS. A major application area of MAS technology is e-business automation, including the establishment and operation of business relationships and the formation of virtual organizations (VOs). One of the key factors influencing the adoption of agent-based approaches in real-world business scenarios is trust. The concept of an electronic institution (EI) has been proposed as a means to provide a regulated and trustable environment, by enforcing norms of behavior and by providing specific services for smooth inter-operability. This chapter exposes our work towards the development of an agent-based EI providing a virtual normative environment that assists and regulates the creation and operation of VOs through contract-related services. It includes a presentation of the EI framework, knowledge representation structures for norms in contracts, and a description of two main institutional services, namely negotiation mediation and contract monitoring.
Advances in intelligent and soft computing, 2012
Despite relevant insights from socio-economics, little research in multiagent systems has address... more Despite relevant insights from socio-economics, little research in multiagent systems has addressed the interconnections between trust and normative notions such as contracts and sanctions. Focusing our attention on scenarios of betrayal, in this paper we combine the use of trust and sanctions in a negotiation process. We describe a scenario of dyadic relationships between truster agents, which make use of trust and/or sanctions, and trustees characterized by their ability and integrity, which may influence their attitude toward betrayal. Both agent behavior models are inspired in socio-economics literature. Through simulation, we show the virtues and shortcomings of exploiting trust, sanctions and a combination of both.
2018 Fifth International Conference on Social Networks Analysis, Management and Security (SNAMS), 2018
There is currently few research in using deep learning (DL) applied to Named Entities Recognition... more There is currently few research in using deep learning (DL) applied to Named Entities Recognition (NER) in Portuguese texts. This work exposes some challenges and limitations but also the benefits of applying DL architectures to NER in Portuguese. Four different DL architectures are applied to Portuguese datasets. All architectures are heavily influenced by previous published work in NER applied to English. Annotated data is used to train and test NER models, while non-annotated data is used to train word embeddings, as well as being a key part of a bootstrapping approach, where raw textual data is used to create NER models.
Uploads
Papers by Eugénio Oliveira
Moreover these disruptive events usually affect at least three different dimensions of the situation: the aircraft assigned to the flight, the crew assignment and, often forgotten, the passengers’ journey and satisfaction.
This book includes answers to this challenge and proposes the use of the Multi-agent System paradigm to rapidly compose a multi-faceted solution to the disruptive event taking into consideration possible preferences of those three key aspects of the problem.
Negotiation protocols taking place between agents that are experts in solving the different problem dimensions, combination of different utility functions and, not less important, the inclusion of the human in the automatic decision-making loop make MASDIMA, the system described in this book, well suited for real-life plan-disruption management applications