|
Retargetable Analysis of Machine Code
Křoustek, Jakub ; Janoušek, Jan (oponent) ; Návrat,, Pavol (oponent) ; Kolář, Dušan (vedoucí práce)
Program analysis is a computer-science methodology whose task is to analyse the behavior of a given program. The methods of program analysis can also be used in other methodologies such as reverse engineering, re-engineering, code migration, etc. In this thesis, we focus on program analysis of a machine-code and we address the limitations of a nowadays approaches by proposing novel methods of a fast and accurate retargetable analysis (i.e. they are designed to be independent of a particular target platform). We focus on two types of analysis - dynamic analysis (i.e. run-time analysis) and static analysis (i.e. analysing application without its execution). The contribution of this thesis within the dynamic analysis lays in the extension and enhancement of existing methods and their implementation as a retargetable debugger and two types of a retargetable translated simulator. Within the static analysis, we present a concept and implementation of a retargetable decompiler that performs a program transformation from a machine code into a human-readable form of representation. All of these tools are based on several novel methods defined by the author. According to our experimental results and users feed-back, all of the proposed tools are at least fully competitive to existing solutions, while outperforming these solutions in several ways.
|
|
Packet Classification Algorithms
Puš, Viktor ; Lhotka,, Ladislav (oponent) ; Dvořák, Václav (vedoucí práce)
This thesis deals with packet classification in computer networks. Classification is the key task in many networking devices, most notably packet filters - firewalls. This thesis therefore concerns the area of computer security. The thesis is focused on high-speed networks with the bandwidth of 100 Gb/s and beyond. General-purpose processors can not be used in such cases, because their performance is not sufficient. Therefore, specialized hardware is used, mainly ASICs and FPGAs. Many packet classification algorithms designed for hardware implementation were presented, yet these approaches are not ready for very high-speed networks. This thesis addresses the design of new high-speed packet classification algorithms, targeted for the implementation in dedicated hardware. The algorithm that decomposes the problem into several easier sub-problems is proposed. The first subproblem is the longest prefix match (LPM) operation, which is used also in IP packet routing. As the LPM algorithms with sufficient speed have already been published, they can be used in out context. The following subproblem is mapping the prefixes to the rule numbers. This is where the thesis brings innovation by using a specifically constructed hash function. This hash function allows the mapping to be done in constant time and requires only one memory with narrow data bus. The algorithm throughput can be determined analytically and is independent on the number of rules or the network traffic characteristics. With the use of available parts the throughput of 266 million packets per second can be achieved. Additional three algorithms (PFCA, PCCA, MSPCCA) that follow in this thesis are designed to lower the memory requirements of the first one without compromising the speed. The second algorithm lowers the memory size by 11 % to 96 %, depending on the rule set. The disadvantage of low stability is removed by the third algorithm, which reduces the memory requirements by 31 % to 84 %, compared to the first one. The fourth algorithm combines the third one with the older approach and thanks to the use of several techniques lowers the memory requirements by 73 % to 99 %.
|
|
Harnessing Forest Automata for Verification of Heap Manipulating Programs
Šimáček, Jiří ; Abdulla, Parosh (oponent) ; Křetínský, Mojmír (oponent) ; Vojnar, Tomáš (vedoucí práce)
This work addresses verification of infinite-state systems, more specifically, verification of programs manipulating complex dynamic linked data structures. Many different approaches emerged to date, but none of them provides a~sufficiently robust solution which would succeed in all possible scenarios appearing in practice. Therefore, in this work, we propose a new approach which aims at improving the current state of the art in several dimensions. Our approach is based on using tree automata, but it is also partially inspired by some ideas taken from the methods based on separation logic. Apart from that, we also present multiple advancements within the implementation of various tree automata operations, crucial for our verification method to succeed in practice. Namely, we provide an optimised algorithm for computing simulations over labelled transition systems which then translates into more efficient computation of simulations over tree automata. We also give a new algorithm for checking inclusion over tree automata, and we provide experimental evaluation demonstrating
|
|
Optimization of Gaussian Mixture Subspace Models and Related Scoring Algorithms in Speaker Verification
Glembek, Ondřej ; Brummer, Niko (oponent) ; Campbell,, William (oponent) ; Burget, Lukáš (vedoucí práce)
This thesis deals with Gaussian Mixture Subspace Modeling in automatic speaker recognition. The thesis consists of three parts. In the first part, Joint Factor Analysis (JFA) scoring methods are studied. The methods differ mainly in how they deal with the channel of the tested utterance. The general JFA likelihood function is investigated and the methods are compared both in terms of accuracy and speed. It was found that linear approximation of the log-likelihood function gives comparable results to the full log-likelihood evaluation while simplyfing the formula and dramatically reducing the computation speed. In the second part, i-vector extraction is studied and two simplification methods are proposed. The motivation for this part was to allow for using the state-of-the-art technique on small scale devices and to setup a simple discriminative-training system. It is shown that, for long utterances, while sacrificing the accuracy, we can get very fast and compact i-vector systems. On a short-utterance(5-second) task, the results of the simplified systems are comparable to the full i-vector extraction. The third part deals with discriminative training in automatic speaker recognition. Previous work in the field is summarized and---based on the knowledge from the earlier chapters of this work---discriminative training of the i-vector extractor parameters is proposed. It is shown that discriminative re-training of the i-vector extractor can improve the system if the initial estimation is computed using the generative approach.
|
|
Network-wide Security Analysis
de Silva, Hidda Marakkala Gayan Ruchika ; Šafařík,, Jiří (oponent) ; Šlapal, Josef (oponent) ; Švéda, Miroslav (vedoucí práce)
The objective of the research is to model and analyze the effects of dynamic routing protocols. The thesis addresses the analysis of service reachability, configurations, routing and security filters on dynamic networks in the event of device or link failures. The research contains two main sections, namely, modeling and analysis. First section consists of modeling of network topology, protocol behaviors, device configurations and filters. In the modeling, graph algorithms, routing redistribution theory, relational algebra and temporal logics were used. For the analysis of reachability, a modified topology table was introduced. This is a unique centralized table for a given network and invariant for network states. For the analysis of configurations, a constraint-based analysis was developed by using XSD Prolog. Routing and redistribution were analyzed by using routing information bases and for analyzing the filtering rules, a SAT-based decision procedure was incorporated. A part of the analysis was integrated to a simulation tool at OMNeT++ environment. There are several innovations introduced in this thesis. Filtering network graph, modified topology table, general state to reduce the state space, modeling devices as filtering nodes and constraint-based analysis are the key innovations. Abstract network graph, forwarding device model and redistribution with routing information are extensions of the existing research. Finally, it can be concluded that this thesis discusses novel approaches, modeling methods and analysis techniques in the area of dynamic networks. Integration of these methods into a simulation tool will be a very demanding product for the network designers and the administrators.
|
|
Acceleration Methods for Evolutionary Design of Digital Circuits
Vašíček, Zdeněk ; Miller, Julian (oponent) ; Zelinka,, Ivan (oponent) ; Sekanina, Lukáš (vedoucí práce)
Although many examples showing the merits of evolutionary design over conventional design techniques utilized in the field of digital circuits design have been published, the evolutionary approaches are usually hardly applicable in practice due to the various so-called scalability problems. The scalability problem represents a general problem that refers to a situation in which the evolutionary algorithm is able to provide a solution to a small problem instances only. For example, the scalability of evaluation of a candidate digital circuit represents a serious issue because the time needed to evaluate a candidate solution grows exponentially with the increasing number of primary inputs. In this thesis, the scalability problem of evaluation of a candidate digital circuit is addressed. Three different approaches to overcoming this problem are proposed. Our goal is to demonstrate that the evolutionary design approach can produce interesting and human competitive solutions when the problem of scalability is reduced and thus a sufficient number of generations can be utilized. In order to increase the performance of the evolutionary design of image filters, a domain specific FPGA-based accelerator has been designed. The evolutionary design of image filters is a kind of regression problem which requires to evaluate a large number of training vectors as well as generations in order to find a satisfactory solution. By means of the proposed FPGA accelerator, very efficient nonlinear image filters have been discovered. One of the discovered implementations of an impulse noise filter consisting of four evolutionary designed filters is protected by the Czech utility model. A different approach has been introduced in the area of logic synthesis. A method combining formal verification techniques with evolutionary design that allows a significant acceleration of the fitness evaluation procedure was proposed. The proposed system can produce complex and simultaneously innovative designs, overcoming thus the major bottleneck of the evolutionary synthesis at gate level. The proposed method has been evaluated using a set of benchmark circuits and compared with conventional academia as well as commercial synthesis tools. In comparison with the conventional synthesis tools, the average improvement in terms of the number of gates provided by our system is approximately 25%. Finally, the problem of the multiple constant multiplier design, which belongs to the class of problems where a candidate solution can be perfectly evaluated in a short time, has been investigated. We have demonstrated that there exists a class of circuits that can be evaluated efficiently if a domain knowledge is utilized (in this case the linearity of components).
|
|
Code Characterization for Automated User Interface Generation
Kadlec, Jaroslav ; Slavík,, Pavel (oponent) ; Sochor, Jiří (oponent) ; Zemčík, Pavel (vedoucí práce)
This work presents novel approach to automation of user interface creation. A taxonomy based on data characterization was adopted and new taxonomy for code characterization is proposed. Taxonomy points most important aspects of data and code so that it can be used in a process of automated user interface creation. The taxonomy is platform independent and can be stored as a metadata in the application file or in a separate external file. A process of automated user interface creation based on the taxonomy is proposed and individual parts of the process are described in more detail. Presented taxonomy and process of user interface generation are demonstrated on examples.
|
|
Extensions to Probabilistic Linear Discriminant Analysis for Speaker Recognition
Plchot, Oldřich ; Fousek, Petr (oponent) ; McCree,, Alan (oponent) ; Burget, Lukáš (vedoucí práce)
This thesis deals with probabilistic models for automatic speaker verification. In particular, the Probabilistic Linear Discriminant Analysis (PLDA) model, which models i--vector representation of speech utterances, is analyzed in detail. The thesis proposes extensions to the standard state-of-the-art PLDA model. The newly proposed Full Posterior Distribution PLDA models the uncertainty associated with the i--vector generation process. A new discriminative approach to training the speaker verification system based on the~PLDA model is also proposed. When comparing the original PLDA with the model extended by considering the i--vector uncertainty, results obtained with the extended model show up to 20% relative improvement on tests with short segments of speech. As the test segments get longer (more than one minute), the performance gain of the extended model is lower, but it is never worse than the baseline. Training data are, however, usually available in the form of segments which are sufficiently long and therefore, in such cases, there is no gain from using the extended model for training. Instead, the training can be performed with the original PLDA model and the extended model can be used if the task is to test on the short segments. The discriminative classifier is based on classifying pairs of i--vectors into two classes representing target and non-target trials. The functional form for obtaining the score for every i--vector pair is derived from the PLDA model and training is based on the logistic regression minimizing the cross-entropy error function between the correct labeling of all trials and the probabilistic labeling proposed by the system. The results obtained with discriminatively trained system are similar to those obtained with generative baseline, but the discriminative approach shows the ability to output better calibrated scores. This property leads to a better actual verification performance on an unseen evaluation set, which is an important feature for real use scenarios.
|
|
Query-by-Example Spoken Term Detection
Fapšo, Michal ; Matoušek, Jindřich (oponent) ; Metze, Florian (oponent) ; Černocký, Jan (vedoucí práce)
This thesis investigates query-by-example (QbE) spoken term detection (STD). Queries are entered in their spoken form and searched for in a pool of recorded spoken utterances, providing a list of detections with their scores and timing. We describe, analyze and compare three different approaches to QbE STD, in various language-dependent and language-independent setups with diverse audio conditions, searching for a single example and five examples per query. For our experiments we used Czech, Hungarian, English and Levantine data and for each of the languages we trained a 3-state phone posterior estimator. This gave us 16 possible combinations of the evaluation language and the language of the posterior estimator, out of which 4 combinations were language-dependent and 12 were language-independent. All QbE systems were evaluated on the same data and the same features, using the metrics: non-pooled Figure-of-Merit and our proposed utterrance-normalized non-pooled Figure-of-Merit, which provided us with relevant data for the comparison of these QbE approaches and for gaining a better insight into their behavior. QbE approaches presented in this work are: sequential statistical modeling (GMM/HMM), template matching of features (DTW) and matching of phone lattices (WFST). To compare the performance of QbE approaches with the common query-by-text STD systems, for language-dependent setups we also evaluated an acoustic keyword spotting system (AKWS) and a system searching for phone strings in lattices (WFSTlat). The core of this thesis is the development, analysis and improvement of the WFST QbE STD system, which after the improvements, achieved similar performance to the DTW system in language-dependent setups.
|
|
Intrusion Detection in Network Traffic
Homoliak, Ivan ; Čeleda, Pavel (oponent) ; Ochoa,, Martín (oponent) ; Hanáček, Petr (vedoucí práce)
The thesis deals with anomaly based network intrusion detection which utilize machine learning approaches. First, state-of-the-art datasets intended for evaluation of intrusion detection systems are described as well as the related works employing statistical analysis and machine learning techniques for network intrusion detection. In the next part, original feature set, Advanced Security Network Metrics (ASNM) is presented, which is part of conceptual automated network intrusion detection system, AIPS. Then, tunneling obfuscation techniques as well as non-payload-based ones are proposed to apply as modifications of network attack execution. Experiments reveal that utilized obfuscations are able to avoid attack detection by supervised classifier using ASNM features, and their utilization can strengthen the detection performance of the classifier by including them into the training process of the classifier. The work also presents an alternative view on the non-payload-based obfuscation techniques, and demonstrates how they may be employed as a training data driven approximation of network traffic normalizer.
|
host ::
