Search

Dissertação de mestrado integrado em Engenharia Informática, This document presents a Master Thesis in the Integrated Master’s in Informatics Engi neering focused on the automatic identification of vulnerabilities, that was accomplished at Universidade do Minho in Braga, Portugal. This thesis work aims at developing a machine learning based tool for automatic iden tification of vulnerabilities on programs (source, high level code), that uses an abstract syntax11tree representation. It is based on FastScan, using code2seq approach. Fastscan is a recently developed system aimed capable of detecting vulnerabilities in source code using machine learning techniques. Nevertheless, FastScan is not able of identifying the vulnerability type. In the presented work the main goal is to go further and develop a method to identify specific types of vulnerabilities. As will be shown, the goal will be achieved by changing the method of receiving and processing in a different way the input data and developing an architecture that brings together multiple models to predict different specific vulnerabilities. The best f1 metric obtained is 93% resulting in a precision of 90% and accuracy of 85%, according to the performed tests and regarding a trained model to predict vulnerabilities of the injection type. These results were obtained with the contribution given by the optimization of the model’s hyperparameters and also the use of the Search Cluster from University of Minho that greatly diminished the necessary time to perform training and testing. It is important to refer that overfitting was detected in the late stages of the tests, so this results do not represent the true value in real context. Also an interface is presented, it allows to better interact with the models and analyse the scan results., Este documento apresenta uma dissertação do Mestrado Integrado em Engenharia Infor mática, que tem como foco a automação da deteção de vulnerabilidades e foi concluída na Universidade do Minho em Braga, Portugal. O trabalho apresentado nesta tese pretende desenvolver uma ferramenta que utiliza machine learning e que seja capaz de identificar vulnerabilidades em código. Utilizando para isso a representação do mesmo numa abstract syntax tree. Tem como base FastScan que utiliza a abordagem do code2seq. Fastscan é um projeto recentemente desenvolvido que é capaz de detetar vulnerabilidades em código utilizando técnicas de machine learning, sendo que tem algumas lacunas como o facto de não ser capaz de identificar vulnerabilidades específicas. No trabalho apresentado o objetivo é ir mais além e desenvolver um método capaz de identificar qual o tipo específico de vulnerabilidade presente. Como será apresentado ao longo do documento, este objetivo será alcançado pela alteração do método de receção e processamento dos dados recebidos, assim como o desenvolvimento de uma arquitetura que junte os vários modelos de maneira a cooperarem e a ferramenta ser capaz de detetar e prever a presença de vulnerabilidades específicas. A melhor métrica de f1 obtida foi de 93%, com precisão de 90% e accuracy de 85%, de acordo com os testes efetuados sobre um modelo treinado para prever a presença de vulnerabilidades do tipo de injection. Os resultados foram obtidos devido à otimização dos hiper-parâmetros dos modelos e o cluster Search da Universidade do Minho diminuiu consideravelmente o tempo necessário para efetuar o traino e testes dos modelos. É importante referir que foi detetado overfitting na fase final do desenvolvimento deste trabalho, sendo que os resultados apresentados não representam o valor real dos modelos em contexto real. Para além disso é apresentada uma interface que permite interagir e analisar os resultados de um scan feito pelos modelos.

The work described in this paper aims at developing a machine learning based tool for automatic identification of vulnerabilities on programs (source, high level code), that uses an abstract syntax tree representation. It is based on FastScan, using code2seq approach. Fastscan is a recently developed system aimed capable of detecting vulnerabilities in source code using machine learning techniques. Nevertheless, FastScan is not able of identifying the vulnerability type. In the presented work the main goal is to go further and develop a method to identify specific types of vulnerabilities. As will be shown, the goal will be achieved by optimizing the model’s hyperparameters, changing the method of preprocessing the input data and developing an architecture that brings together multiple models to predict different specific vulnerabilities. The preliminary results obtained from the training stage, are very promising. The best f1 metric obtained is 93% resulting in a precision of 90% and accuracy of 85%, according to the performed tests and regarding a trained model to predict vulnerabilities of the injection type., OASIcs, Vol. 94, 10th Symposium on Languages, Applications and Technologies (SLATE 2021), pages 14:1-14:14

International audience

Prior studies suggested that SAGA and TFIID are alternative factors that promote RNA polymerase II (RNA Pol II) transcription with about 10% of genes in S. cerevisiae dependent on SAGA. The remainder 90% of the genome would be regulated by TFIID. We reassessed the role of SAGA by mapping its genome-wide location and role in global transcription in budding yeast. We observed that SAGA maps to regulatory elements of most genes, irrespective of previous designations of SAGA- or TFIID-dominated genes. Additionally, disruption of either SAGA or TFIID through mutation or rapid subunit depletion reduces transcription from nearly all genes, measured by newly-synthesized RNA or RNA Pol II chromatin immunoprecipitation. We also found that the acetyltransferase Gcn5 synergizes with Spt3 to promote global transcription and that Spt3 functions to stimulate TBP recruitment at all tested genes. Our data demonstrate that both SAGA and TFIID act as general cofactors required for essentially all RNA Pol II transcription and is not consistent with the previous classification of SAGA- and TFIID-dominated genes., Des études antérieures suggèrent que les complexes SAGA et TFIID sont des facteurs jouant un rôle complémentaire dans la transcription par l’ARN polymérase II. Chez S. cerevisiae, environ 10% des gènes semblent dépendants de SAGA alors que TFIID aurait un rôle dominant sur 90% du génome. De nouvelles approches m’ont permis de montrer que SAGA est recruté sur les régions régulatrices d’une majorité de gènes, indépendamment de leur classification. Des analyses d’ARN nouvellement-synthétisés ont également démontré que l’inactivation des complexes SAGA ou TFIID, par mutation ou déplétion inductible de leurs sous-unités, altèrent la transcription de pratiquement tous les gènes par l’ARN polymérase II. L’acétyltransférase Gcn5 et la sous-unité Spt3 agissent de façon synergique au sein du complexe SAGA pour stimuler le recrutement de TBP et la transcription par l’ARN polymérase II. Ces données indiquent que les complexes SAGA et TFIID agissent comme des cofacteurs généraux, étant nécessaires pour la synthèse de quasiment tous les ARNm et ayant des effets équivalents sur les gènes précédemment décrits comme dominés par SAGA ou par TFIID.

Submitted by Biblioteca Escola Superior de Saúde (biblioteca@ess.ipp.pt) on 2019-01-30T17:52:29Z No. of bitstreams: 1 ART_VieiraAna_2013.pdf: 1521656 bytes, checksum: 968082c63d858c3513ceec645a27968e (MD5) Approved for entry into archive by Biblioteca Escola Superior de Saúde (biblioteca@ess.ipp.pt) on 2019-01-30T17:55:14Z (GMT) No. of bitstreams: 1 ART_VieiraAna_2013.pdf: 1521656 bytes, checksum: 968082c63d858c3513ceec645a27968e (MD5) Approved for entry into archive by Biblioteca Escola Superior de Saúde (biblioteca@ess.ipp.pt) on 2019-01-30T18:06:23Z (GMT) No. of bitstreams: 1 ART_VieiraAna_2013.pdf: 1521656 bytes, checksum: 968082c63d858c3513ceec645a27968e (MD5) Made available in DSpace on 2019-01-30T18:06:23Z (GMT). No. of bitstreams: 1 ART_VieiraAna_2013.pdf: 1521656 bytes, checksum: 968082c63d858c3513ceec645a27968e (MD5) Previous issue date: 2015 info:eu-repo/semantics/publishedVersion

Learning objectives Background Findings and procedure details Conclusion Personal information References, Learning objectives: The objective of this study is to simulate, by using computational fluid dynamics, the haemodynamic conditions in pre-treatment aneurysms and correlate that features with the follow-up observations after treatment with flow-diverter...

Submitted by Agostinho Macau (arm@ucp.pt) on 2021-09-15T14:47:14Z No. of bitstreams: 1 58_Texto_Artigo_171_2_10_20140211.pdf: 1035426 bytes, checksum: f45fe97e604f86ccc491a9fe6d38e522 (MD5) Made available in DSpace on 2021-09-15T14:47:14Z (GMT). No. of bitstreams: 1 58_Texto_Artigo_171_2_10_20140211.pdf: 1035426 bytes, checksum: f45fe97e604f86ccc491a9fe6d38e522 (MD5) Previous issue date: 2014-01-28 info:eu-repo/semantics/publishedVersion

This article discusses the production and reception of Portuguese and foreign “war movies” in Portugal during 1914-1918. It focuses on non-fiction cinema but includes a brief description of the only Portuguese fiction film about the war, ''João Ratão'' [Jorge Brum do Canto, 1940]. Special attention is given to the Portuguese Army’s official film service and its inability to make any films about the Portuguese troops fighting on the European front. The article ends by considering the role of war movies in Portuguese film culture and their role in the process of institutionalization of cinema in this country during the 1910s., 1914-1918-Online International Encyclopedia of the First World War

Tese de doutoramento em Ciências (Engenharia Informática), apresentada à Faculdade de Ciências e Tecnologia da Universidade de Coimbra Throughout the last decades, Darwin’s theory of natural selection has fuelled a vast amount of research in the field of computer science, and more specifically in artificial intelligence. The majority of this work has focussed on artificial selection, rather than on natural selection. In parallel, a growing interest in complexity science brought new modelling paradigms into the scene, with a focus on bottom-up approaches. By combining ideas from complex systems research and artificial life, we present a multi-agent simulation model for open-ended evolution, and a software framework (BitBang) that implements it. We also present a rule list based algorithm implemented for the brain component of the agents. Genetic variation operators were created to drive the evolution of the rule list brains. Several simulation environments were created using the BitBang framework. Experimental results are presented and analysed, validating our model. The results presented show that the model is capable of evolving agents’ controllers in an open-ended evolution simulation. We see that populations evolve sustainable reproduction behaviours, without hard-coding the reproduction conditions into the simulations. By providing evolutionary pressure through the modelling of the environment, we see that on increasingly complex environments, agents evolve increasingly complex behaviours. The rule list brain is shown to provide an important analysis advantage by having readability into the agents’ evolved behaviours. This feature proved to be especially important when unexpected behaviours emerged. Financial support by Fundação para a Ciência e a Tecnologia (FCT) through PhD grant SFRH/BD/18401/2004

Learning objectives Background Imaging findings OR Procedure details Conclusion Personal Information References, Learning objectives: To demonstrate the different locations of focal brainstem tumors in children. To explain the utility of MRI and MR-spectroscopy in the diagnosis and follow-up of these tumors. To confirm that in children, focal brainstem tumors have a better...

International audience; The relationship the firm has with its clients and suppliers largely determines the amount of value it is able to capture in the chain. Yet little is known about the impact of market uncertainty on the choice of the best buyer-supplier strategy in terms of value appropriation. The paper proposes to investigate which among different types of strategies bring more value to the buyer when market uncertainty increases. It relies on a very innovative method of multi-agents simulation based on an in-depth ethnographic observation phase of a real market in France. It shows that pure loyalty is not necessarily the best strategy when market uncertainty increases and that the buyer may face a dilemma between maximizing the margin rate and ensuring regularity of supply.

International audience; In this paper, we present our first results obtained with a multi-agent based simulation of the fruit and vegetable Rungis market. Our goal is to simulate negotiation and price formation based on observed strategies on the real market. We aim to understand negotiation and price formation mechanisms, and explain their impact on macro variables (like the difference between the official price and the average price used on the market).

Ethnographic film is often associated with many European countries’ past as colonial powers and the way these countries used film to depict African, American and Asian territories and populations they once ruled. However, ethnographic film also has a European tradition of its own, closely interlaced with the history of ethnography and anthropology as autonomous sciences and with the desire of scholars to represent local, regional and national cultural identities. This paper presents a Portuguese attempt of this sort dating from 1938, when the authoritarian regime organized a national contest to determine which would be Portugal’s most “authentic” village – something other European countries also did. As part of this metonymic contribution to the construction of Portugal’s national identity as an agrarian utopia, a short documentary was shot, sponsored by the same official propaganda office that had organized the contest. In this film, the viewer’s gaze is made to coincide with the one of the national jury visiting the final selection of 12 villages and to whose benefit local scholars had organized all sorts of colourful peasant traditions hoping to cause the strongest impression. The film makes a strong case for the importance of ethnographic film as a relevant instance not only of the iteration of existing European national cultures, but also of the construction of so many of Europe’s national identities and traditions. Suffice to say that even today the village of “Monsanto”, which won the 1938 contest, is still referred to as “Portugal’s most Portuguese village”.

This thesis examines the contemporary practice of the digital audiovisual essay, which is defined as a material form of thinking at the crossroads of academic textual analysis, personal cinephilia, and popular online fandom practices, to suggest that it allows rich epistemological discoveries not only about individual films and viewing experiences, but also about how cinema is perceived in the context of digitally mediated audiovisual culture. Chapter one advances five key defining tensions of the digital audiovisual essay: its object is the investigation of specific films and cinephiliac experiences; it uses a performative research methodology based on the affordances of digital viewing and editing technologies; it exists primarily in Web 2.0 and takes advantage of its collaborative and dialogical modes of production; it is a “rich text object” that continuously tests the different contributions of both verbal and audiovisual forms of communication to the production of knowledge about the cinema; and finally, the digital audiovisual essay has an important pedagogical potential, not only for those who watch it, but especially for those who practice it. Chapter two presents the theoretical framework of the dissertation, challenges the ‘newness’ of the digital audiovisual essay, and suggests that any investigation of this cultural practice must address its ideological implications and its role in the context of contemporary audiovisual culture. Accordingly, it relates the editing and compositional techniques of the digital audiovisual essay with modernist montage and suggests that the audiovisual essay has not only inherited, but has also updated and enhanced the dialectical interdependency between critical and consumerism drives that shaped modernism’s ambiguous relation to mass culture. The final chapter examines four case studies (David Bordwell, Catherine Grant, ::kogonada, and Kevin B. Lee) that showcase the contradictory tensions of this cultural practice and broad our understanding of the politics of the digital audiovisual essay.

This thesis aims to contribute for reviewing Safety film preservation guidelines currently in practice, concerning film identification, film assessment and storage conditions in the archive, Arquivo Nacional das Imagens em Movimento (ANIM – National Archive of Moving Images), conservation center of Cinemateca Portuguesa. Since the 1980s, stability issues and degradation factors concerning the oldest type of safety film, cellulose triacetate (CTA), have been studied. Based on the results accomplished with those studies, preservation guidelines were established highlighting the advantages of low temperature and low relative humidity for film stability. Despite the guideline’s disclosure, there are practical issues, such as lack of staff, financial restraints and lack of studies regarding polyethylene terephthalate (PET) films decay, that ultimately jeopardize the preservation of motion film collections. In order to have an overall comprehension of the identification and assessment methods used, a review of relevant literature about national and international guidelines for film preservation was performed. This review endowed a broader understanding of the identification, assessment and monitorization carried out in ANIM archive. Aiming to understand preservation practices, an investigation was undertaken at ANIM. Current identification and assessment methods were reexamined, tested and reevaluated. Based on the results accomplished, new identification and assessment tools, as well as, recommendations concerning organization and storage are proposed. For the experimental part of this work, identification and assessment of film (image and sound) were performed. For this internship and study, it was decided firstly to perform the identification and assessment of 100 films and film cans according to the macro identification and macro assessment methods described in the literature. For macro assessment, Acid Detection (A-D) strips and macro signs of degradation were used to evaluate the condition of the films studied. Moreover, the macro assessment results obtained in 2006 and 2014 were also analyzed. In order to identify possible degradation trends and contribute to the review of the current preservation practice, past and present macro assessment results were compared. Additionally, new assessment methods, such as hardness measurement, hardness (Shore A), were tested. The results obtained confirmed that free acidity tests are effective methods for monitoring safety film decay. In this study it was also possible to identify degradation trends, specifically concerning magnetic tapes. Moreover, besides CTA film decay, during the macro assessment of ANIM’s collection, PET films degraded were found. Based on these results it was possible to propose a new methodology for film preservation.