Eventos

Prova de Doutoramento

22 Junho 2018, 16:58 - Fátima Sampaio


Candidate: Fernando Pedro Pascoal dos Santos Nº 64757

Title: Dynamics of Reputation and the Self-organization of Cooperation 

Date: 29/06/2018

Time: 12:15

Location: Sala de Reuniões do Departamento de Engenharia Informática (0.19), Pavilhão de Informática II do IST, Alameda 

Supervisor: Professor Francisco João Duarte Cordeiro Correia dos Santos

Co-Supervisors: Professora Ana Maria Severino de Almeida e Paiva / Professor Jorge Manuel dos Santos Pacheco

Abstract: Indirect Reciprocity (IR) – Alice behaves adequately towards Bob; Carol knows about it and thus helps Alice – is a central mechanism to explain human cooperation. IR involves reputation, status and complex information processing, being the most specifically human of all cooperation mechanisms discovered to date. Understanding evolutionary dynamics under IR can shed light on human social behaviors and morality. Simultaneously, IR can inform about ways of engineering cooperation in new one-shot interaction paradigms where reputations are central – such as web-based platforms or, with the advent of Artificial Intelligence, agent-agent and human-agent interactions.In this thesis we develop a formulation of IR to deal with populations of finite sizeand, in this context, investigate the limits of IR in promoting cooperation along several nonexcluding directions. We start by exploring the stochastic dynamics of IR employing na analytical framework that relies on the assumption that exploration rates – the likelihood of spontaneously adopting another strategy – are low. Subsequently, we develop a computational model to simulate, visualize and investigate the evolutionary dynamics, under IR, of populations of finite size and at arbitrary exploration rates. This allows us to study, originally, the impact of two key features of human social interactions: (i) costly reputation building, whereby individuals may decide to pay a cost to share the outcome of a private interaction; (ii) high order social norms, including the past reputations of individuals (fourth-order under the existing IR hierarchical classification). The complexity of both norms and strategies involved in (ii) called for the development of a new paradigm in assessing information exchange and processing in IR. To this end, we propose a new means of quantifying the complexity of both norms and strategies, resorting to concepts rooted in Boolean algebra.Employing these new methodologies, we conclude that (i) the leading norms capable of promoting cooperation, not only depend sensitively on the population size, but also they may be affected by high exploration rates, which may both improve and harm cooperation, depending on the social norm at work; (ii) under costly reputation building, cooperation depends on the ability of individuals to anticipate accurately the reporting intention of their peers. Finally, we identify a key pattern of fourth-order norms that constitutes a necessary condition for a norm to promote cooperation. We show that, out of 216 possible norms, there exists one very simple norm that complies with this pattern, promoting high levels of cooperation while ignoring the previous reputation of individuals.


Prova de Doutoramento

18 Junho 2018, 12:12 - Fátima Sampaio


Candidate: Luís Filipe Nobre Horta Baptista Garcia Nº 60957

Title: Toward a Context-Aware Augmentative and Alternative Communication System with Vocabulary Prediction for the European Portuguese 

Date: 26/06/2018

Time: 11:00

Location: Anfiteatro PA-3 (Piso -1, Pavilhão de Matemática) IST, Alameda

Advisors: Professor Luís Miguel Veiga Vaz Caldas de Oliveira / Professor David Manuel Martins de Matos 

Abstract: Persons who rely on Augmentative and Alternative Communication (AAC) systems to communicate, face many difficulties when they try to maintain a dialog, in part because of the poor output rate offered by current systems. To minimize this problem, this thesis proposes and evaluates applying the wholeutterance approach to vocabulary prediction. One of the techniques studied was sentence prediction, a technique that may complement word prediction to assist users compose their messages faster. Since prediction of sentences seems to be a very context-dependent process, we also evaluated use of context-awareness to improve vocabulary prediction. For AAC users who can only communicate using pictograms, or images, we also evaluated counterpart techniques of word and sentence prediction, namely single pictogram and pictogram sentences prediction.Our approach to context-awareness consisted of using specific user profiles for different communication contexts (location, time, and speaking partner). Depending on the context, the AAC device can configure itself using data from the associated user profile, to adapt its interface, and language models for vocabulary prediction, to the current communication context. To evaluate this approach we developed, in conjunction with rehabilitation professionals four context-aware AAC solutions for real AAC users. Using these real AAC solutions, we were able to collect corpora representative of this type of communication, train language models for vocabulary prediction, and carry out user tests and software simulations to emulate ideal performance.In user tests, combining word and sentence prediction were obtained statistically significant higher words per minute (WPM) than using only word prediction (18.8 WPM vs 8.3 WPM). In another study, using location-specific language models for word and sentence prediction were achieved mean improvements of 2.4% in words per minute, but differences were not statistically significant. Software simulations showed that location-specific language models tended to perform better than a single language model under high sentence reuse scenarios.Concerning pictogram prediction, in user tests, the condition combining single pictogram and pictogram sentence prediction obtained the best results (6.2 WPM). The condition with no pictogram prediction achieved the worst result (4.5 WPM). However, there were not statistically significant diferences between conditions. There were though, statistically significant increases in keystroke savings. In a subsequent study, simulations a pictogram corpus showed that location-specific language models could outperform a single language model, in a statistically significant way, when the sentence reuse rate was greater than 75%. Very interestingly, analysis of the corpus produced by a real AAC user, we followed-up during one year, showed high sentence reuse rates, ranging from 73.5% to 99.0%, depending on user location. Globally speaking, results obtained show that the proposed techniques can improve the AAC users’ performance.


Prova de Dissertação METI

18 Junho 2018, 11:46 - Elisabete Maria Santos Madeira Ribeiro da Fonseca

Semana de 18 a 22 de junho de 2018

 

Candidato: Pedro dos Santos Duarte Dias (73848)

Título da Dissertação Dinâmicas de punição em populações estruturadas

Data: 20 de junho de 2018

Hora:
11:15h

Local: Sala de reuniões 2.6 - IST - TagusPark

Orientação:  Prof. Francisco João Duarte Cordeiro Correia dos Santos/ Dr. Fernando Pedro Pascoal dos Santos

_____________________________________


Prova de Dissertação MEIC-A

18 Junho 2018, 11:44 - Elisabete Maria Santos Madeira Ribeiro da Fonseca

Semana de 25 a 29 de junho de 2018

 

Candidato: Filipe Pedro Guerra Magalhães (79118)

Título da Dissertação: Distributed Algorithm for the Analysis of Properties of Complex Networks

Data: 26 de junho de 2018

Hora:
09:00h

Local: Sala de reuniões, 0.17, Pavilhão de Informática II - IST - Alameda

Orientação:  Prof. Juan António Acebron/Prof. José Carlos Alves Pereira Monteiro

_____________________________________


Prova de CAT

14 Junho 2018, 17:28 - Fátima Sampaio


Candidate: Pradeeban Kathiravelu N.º 77177

Title: Software-Defined Systems for Network-Aware Service Composition and Workflow Placement

Date: 18/06/2018

Time: 13h00

Location: Anfiteatro QA1.1 (Piso 1, Torre Sul) Alameda

Advisors: Professor Luís Manuel Antunes Veiga / Professor Peter Van Roy

Abstract: Network Softwarization revolutionizes the networking landscape from building, incrementally deploying, and maintaining the environment, regarding performance and management. Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) are two core tenets of network softwarization. SDN abstracts away the control of the network devices in the data plane such as switches and routers and offers a logically centralized control plane. Through its unified view and control of the data plane devices, SDN facilitates programmability and configurability in the data center networks. On the other hand, NFV virtualizes dedicated hardware middleboxes and deploys them on top of servers and data centers as network functions. Thus, network softwarization enables efficient management of the network.The scale of data and the demand for high data rate in applications are rapidly increasing over the years. By moving the computing resources closer to the users, edge computing minimizes the latency caused by otherwise pulling the data between the user and a centralized platform. Thus, edge nodes continue to overtake or replace traditional centralized cloud platforms in composing eScience workflows and Internet of Things (IoT) applications. Due to the inherent bandwidth cost associated with the distance between the network services and the end user, more and more third-party service providers choose to deploy their services at the edge. However, increasing variety and volume of network services and devices have caused challenges in exploiting the edge environments to their full potential.In this work, we aim at exploiting network softwarization to address the deployment and management challenges of network services in heterogeneous execution environments, ranging from the data centers to the edge. This work proposes three significant contributions. First, we extend SDN in data center environments to unify various phases of development and deploy the workloads seamlessly, from simulations and emulations to physical deployment environments. We further extend this work to support multiple Service Level Agreements (SLAs) across different network flows in the data centers, by leveraging a differentiated redundancy. Finally, we design a cloud-assisted network, to offer a latency-aware and cost-efficient virtual connectivity provider in the Internet-scale.Second, we propose a scalable architecture for the deployment of our edge services, leveraging and extending SDN and Message-Oriented Middleware (MOM) for a logically centralized execution of service workflows in edge environments. We present virtual network allocation strategies for the execution of multi-tenant network flows at the edge while sharing the same physical infrastructure. We propose a framework to construct service chains of various services, including, web services, data services, and network services, in multi-tenant edge environments. We thus propose a Software-Defined Service Composition architecture for web service compositions and Network Service Chains (NSCs).Third, we investigate how the proposed software-defined systems can be applied to big data applications. Mainly, we design Software-Defined Data Services, interoperable and network-aware big data executions. Finally, we conclude this document with a summary of the on-going work. Despite the proliferating network services and cloud providers, the users are often locked to a particular vendor due to incompatible interfaces and limitations in a seamless migration. Furthermore, the cloud and service providers typically control the user network services, thus offering limited control to the users. We see our work as the first step in addressing those aforementioned issues in bringing the control back to the user, despite using several third-party network and cloud services for the execution of their services.