Submitted by Fernanda Rodrigues de Lima (fernanda.rlima@ufpe.br) on 2018-08-02T22:51:47Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Daniel Kudlowiez Franch.pdf: 1140156 bytes, checksum: c02b1b4ca33f8165be5960ba5a212730 (MD5) / Approved for entry into archive by Alice Araujo (alice.caraujo@ufpe.br) on 2018-08-07T21:11:31Z (GMT) No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Daniel Kudlowiez Franch.pdf: 1140156 bytes, checksum: c02b1b4ca33f8165be5960ba5a212730 (MD5) / Made available in DSpace on 2018-08-07T21:11:31Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Daniel Kudlowiez Franch.pdf: 1140156 bytes, checksum: c02b1b4ca33f8165be5960ba5a212730 (MD5) Previous issue date: 2017-03-10 / FACEPE / Discrete dynamical systems are widely used in a variety of scientific and engineering applications, such as electrical circuits, machine learning, meteorology and neurobiology. Modeling these systems involves performing statistical analysis of the system output to estimate the parameters of a model so it can behave similarly to the original system. These models can be used for simulation, performance analysis, fault detection, among other applications. The current work presents two new algorithms to model discrete dynamical systems from two categories (synchronizable and non-synchronizable) using Probabilistic Finite State Automata (PFSA) by analyzing discrete symbolic sequences generated by the original system and applying statistical methods and inference, machine learning algorithms and graph minimization techniques to obtain compact, precise and efficient PFSA models. Their performance and time complexity are compared with other algorithms present in literature that aim to achieve the same goal by applying the algorithms to a series of common examples. / Sistemas dinâmicos discretos são amplamente usados em uma variedade de aplicações cientifícas e de engenharia, por exemplo, circuitos elétricos, aprendizado de máquina, meteorologia e neurobiologia. O modelamento destes sistemas envolve realizar uma análise estatística de sequências de saída do sistema para estimar parâmetros de um modelo para que este se comporte de maneira similar ao sistema original. Esses modelos podem ser usados para simulação, referência ou detecção de falhas. Este trabalho apresenta dois novos algoritmos para modelar sistemas dinâmicos discretos de duas categorias (sincronizáveis e não-sincronizáveis) por meio de Autômatos Finitos Probabilísticos (PFSA, Probabilistic Finite State Automata) analisando sequências geradas pelo sistema original e aplicando métodos estatísticos, algoritmos de aprendizado de máquina e técnicas de minimização de grafos para obter modelos PFSA compactos e eficientes. Sua performance e complexidade temporal são comparadas com algoritmos presentes na literatura que buscam atingir o mesmo objetivo aplicando os algoritmos a uma série de exemplos.

O teste baseado em modelos tem como objetivo auxiliar a atividade de testes, gerando conjuntos de casos de teste a partir de modelos, como Máquinas de Estados Finitos (MEFs). Diversos métodos de geração de conjuntos de caso de teste têm sido propostos ao longo das últimas décadas, com algumas contribuições recentes. Dentre esses trabalhos, há os que geram seqüências de verificação que são conjuntos de caso de teste formados por uma única seqüência e que são capazes de detectar os defeitos de uma implementação cujo comportamento pode ser modelado a partir de uma MEF. Neste trabalho é proposto um algoritmo de geração de seqüências de verificação que tem a finalidade de gerar seqüências menores que as seqüências geradas pelos métodos existentes. O algoritmo, que é baseado na técnica de algoritmos genéticos e nas condições de suficiência para a completude de casos de teste, consiste basicamente em criar novas seqüências a partir de seqüências menores. Por meio de mutações, novas seqüências são geradas pelo algoritmo. As condições de suficiência são utilizadas para determinar quais seqüências geradas são seqüências de verificação. Também são apresentados neste trabalho os estudos experimentais realizados para determinar o comportamento do algoritmo diante de diferentes contextos / Model-based testing aims at aiding the testing activity, generating test cases from models such as Finite State Machines (FSM). Several test cases generation methods have been proposed along the last decades, with some recent contributions. Among these works, there are those that generate checking sequences, which are test cases formed by a single sequence and which are capable of detecting faults in an implementation whose behavior can be modeled as an FSM. This work proposes a checking sequences generation algorithm which aims at generating sequences smaller than the sequences generated by existing methods. The algorithm, which is based on the genetic algorithms technique and sufficient conditions for completeness of test cases, basically consists of creating new sequences from small sequences. Through mutations, new sequences are generated by the algorithm. The suffcient conditions are used to determine which sequences are checking sequences. Experimental studies are presented in this work to determine the behavior of the algorithm on different contexts

Developing real-time strategy game AI is a challenging task due to that an AI-player has to deal with many different decisions and actions in an ever changing complex game world. Humans have little problem when it comes to dealing with the complexity of the game genre while it is a difficult obstacle to overcome for the computer. Adapting to the opponents strategy is one of many things that players typically have to do during the course of a game in the real-time strategy genre. This report presents a finite-state machine based solution to the mentioned problem and implements it with the help of the existing Starcraft: Broodwar AI Opprimobot. The extension is experimentally compared to the original implementation of Opprimobot. The comparison shows that both manages to achieve approximately the same win ratio against the built-in AI of Starcraft: Broodwar, but the modified version provides away to model more complex strategies.

The decentralized nature of the Internet forms its very foundation, yet it is this very nature that has opened networks and individual machines to a host of threats and attacks from malicious agents. Consequently, forensic specialists - tasked with the investigation of crimes commissioned through the use of computer systems, where evidence is digital in nature - are often unable to adequately reach convincing conclusions pertaining to their investigations. Some of the challenges within reliable forensic investigations include the lack of a global view of the investigation landscape and the complexity and obfuscated nature of the digital world. A perpetual challenge within the evidence analysis process is the reliability and integrity associated with digital evidence, particularly from disparate sources. Given the ease with which digital evidence (such as metadata) can be created, altered, or destroyed, the integrity attributed to digital evidence is of paramount importance. This dissertation focuses on the challenges relating to the integrity of digital evidence within reliable forensic investigations. These challenges are addressed through the proposal of a model for the construction of a Forensic Evidence Management System (FEMS) to preserve the integrity of digital evidence within forensic investigations. The Biba Integrity Model is utilized to maintain the integrity of digital evidence within the FEMS. Casey's Certainty Scale is then employed as the integrity classifcation scheme for assigning integrity labels to digital evidence within the system. The FEMS model consists of a client layer, a logic layer and a data layer, with eight system components distributed amongst these layers. In addition to describing the FEMS system components, a fnite state automata is utilized to describe the system component interactions. In so doing, we reason about the FEMS's behaviour and demonstrate how rules within the FEMS can be developed to recognize and pro le various cyber crimes. Furthermore, we design fundamental algorithms for processing of information by the FEMS's core system components; this provides further insight into the system component interdependencies and the input and output parameters for the system transitions and decision-points infuencing the value of inferences derived within the FEMS. Lastly, the completeness of the FEMS is assessed by comparing the constructs and operation of the FEMS against the published work of Brian D Carrier. This approach provides a mechanism for critically analyzing the FEMS model, to identify similarities or impactful considerations within the solution approach, and more importantly, to identify shortcomings within the model. Ultimately, the greatest value in the FEMS is in its ability to serve as a decision support or enhancement system for digital forensic investigators. Copyright / Dissertation (MSc)--University of Pretoria, 2010. / Computer Science / unrestricted

O teste baseado em modelos visa à derivação de casos de teste a partir de modelos produzidos ao longo do desenvolvimento de software. Nesse contexto, as Máquinas de Estados Finitos têm sido amplamente pesquisadas e utilizadas para derivação de seqüências de teste. Para isso, vários métodos de geração de seqüências de teste têm sido desenvolvidos há várias décadas. O objetivo desses métodos é a obtenção de um conjunto de teste que seja capaz de revelar os defeitos de uma implementação. Entretanto, muitas vezes os conjuntos gerados são muito grandes, o que torna sua aplicação inviável. Trabalhos recentes definiram condições que podem ser utilizadas para investigar mecanismos de redução de casos de teste. Este trabalho apresenta uma estratégia para a redução de conjuntos de seqüências de teste a partir de Máquinas de Estados Finitos com base em condições de suficiência. A estratégia baseia-se na combinação de seqüências de um conjunto de teste, de forma a reduzir o número de seqüências e o tamanho delas, mantendo a completude do conjunto. São apresentadas seis abordagens de redução baseadas na estratégia proposta, as quais foram implementadas em uma ferramenta. Para avaliar as abordagens foram conduzidos estudos experimentais, os quais também serviram para inferir sobre as características e propriedades de cada abordagem. Além disso, um estudo de caso com MEFs reais também foi realizado / Model-based testing aims at generating test cases from models produced along the software development process. In this context, Finite State Machines (FSM) have been largely investigated and used for generating test sequences. In the past decades, several test generation methods have been proposed to obtain test suites that are able to reveal implementation faults. Nevertheless, most of the generated test suites are huge, thus hindering their application in practice. Recent research has defined new sufficient conditions that can be employed in mechanisms for reducing the length of test sequences. This work presents a strategy based on sufficient conditions for reducing the length of test cases derived from FSMs. Our strategy is based on sequence combination of a test suite, aiming to reduce the number of sequences and their length, however still keeping full fault coverage. Six reduction approaches are presented based on the proposed strategy and implemented in a tool. In order to evaluate the strategy, we conducted experimental studies that identified characteristics and properties for each of the six proposed approaches. Moreover, a case study with real-world FSMs was performed

This master’s thesis deals with analysis of state machines for embedded applications. The issue of finite-state machine is described theoretically. The document also contains a proposal for funding for modeling finite state machines in Matlab/Simulink. It is designed data representation of finite automaton. Over this data representation algorithm of minimization is applied. Finally, the algorithm is implemented to generate code in C language.

Das Ziel dieser Arbeit ist die Formulierung eines Ansatzes zum maschinellen Lernen von Sprachmorphologie, in dem letztere als Zeichenkettentransformationen auf ganzen Wörtern, und nicht als Zerlegung von Wörtern in kleinere stukturelle Einheiten, modelliert wird. Der Beitrag besteht aus zwei wesentlichen Teilen: zum einen wird ein Rechenmodell formuliert, in dem morphologische Regeln als Funktionen auf Zeichenketten definiert sind. Solche Funktionen lassen sich leicht zu endlichen Transduktoren übersetzen, was eine solide algorithmische Grundlage für den Ansatz liefert. Zum anderen wird ein statistisches Modell für Graphen von Wortab\-leitungen eingeführt. Die Inferenz in diesem Modell erfolgt mithilfe des Monte Carlo Expectation Maximization-Algorithmus und die Erwartungswerte über Graphen werden durch einen Metropolis-Hastings-Sampler approximiert. Das Modell wird auf einer Reihe von praktischen Aufgaben evaluiert: Clustering flektierter Formen, Lernen von Lemmatisierung, Vorhersage von Wortart für unbekannte Wörter, sowie Generierung neuer Wörter.

Fuzzing is a test method used to automatically generate test case inputs and to executea system under test (SUT) with those inputs. The method is traditionally used to discovercrash-inducing bugs in software. Fuzzing can generate thousands of inputs per secondand many implementations use smart techniques to reach deeply into the code. Fewfuzz testing implementations, however, have the ability to explore and retain informationof state in stateful applications. We develop an extension of the fuzzer American Fuzzy Lop (AFL), building on the workof the Ijon project, and utilize its fuzzing capabilities to discover states in SUT; inparticular, applications built as finite state machines. The extension successfullyharnesses AFL’s input generation to explore the SUT’s state space. We then implement functionality that allows for the SUT to return state information tothe fuzzer, including the state path and path length. Furthermore, functionality is addedthat allows the test operator to specify the expected number of states in the SUT, andGUI extensions that provide real-time information of state discovery during fuzzing. The state information retained after a completed fuzzing session is automaticallysummarized in a structured format. We further demonstrate that the summarizedinformation can be used to generate test cases for a test operator to verify the SUT.

Les travaux de cette thèse s'inscrivent dans la théorie des automates et des langages formels. ils peuvent se diviser en deux parties qui donnent également deux visions différentes de manipuler les langages dans la théorie des automates. La première partie s'intéresse à la notion de dérivation des expressions qui permet de faire passer le formalisme des quotients de langages au niveau des expressions rationnelles. en particulier cette thèse étudie les termes dérivés cassés d'une expression rationnelle. ces termes dérivés cassés permettent, sous certaines circonstances, et à l'aide d'autres opérations, une réversibilité de la transformation d'un automate en une expression rationnelle. Dans la seconde partie, la théorie des automates est utilisée pour traiter des problèmes sur les systèmes de numération. les systèmes de numération représentent des nombres par des mots. il est possible d'utiliser des automates et des transducteurs afin d'être capable de 'compter' sur un langage rationnel représentant les entiers. plus précisément ces automates sont étudiés pour le cas des systèmes de numération abstraits qui associent à chaque entier un mot d'un langage rationnel, ordonné par l'ordre radiciel. dans un tel système, la fonction qui permet de calculer le mot suivant est une fonction co-séquentielle par morceaux, c'est-à-dire qu'il suffit de lire deux fois le mot d'entrée de la droite vers la gauche pour qu'une machine calcule son image. / The works in this thesis lies in the automata and formal languages theory. in the first part, the notion of derivation of rational expressions is studied. more precisely the broken derived terms of a rational expressions. Theses broken derived terms allow, under certain circumstances, with some other operations on automata, to have the reversibility of the transformation of an automaton into a rational expression. In the second part, automata and tranducers allow to 'count' on a numeration system, where integers are represented by words on a rational language. more precisely, this part adress the problem of counting in an abstract numeration systems, which maps to any word of a rational language, ordored by radix order, the integer corresponding to the order of the word. in such a numeration system, the function which computes the successor of a word is a piecewise co-sequential function: it can be realised by a machine which reads the input two times to give the output.

No description available.