Speaker or voice recognition is the task of automatically recognizing people from their speech signals. This technique makes it possible to use uttered speech to verify the speaker's identity and control access to secured services. Surveillance, counter-terrorism and homeland security department can collect voice data from telephone conversation without having to access to any other biometric dataset. In this type of scenario it would be beneficial if the confidence level of authentication is high. Other applicable areas include online transactions,database access services, information services, security control for confidential information areas, and remote access to computers. Speaker recognition systems, even though they have been around for four decades, have not been widely considered as standalone systems for biometric security because of their unacceptably low performance, i.e., high false acceptance and true rejection. This thesis focuses on the enhancement of speaker recognition through a combination of intra-modal fusion and accent modeling. Initial enhancement of speaker recognition was achieved through intra-modal hybrid fusion (HF) of likelihood scores generated by Arithmetic Harmonic Sphericity (AHS) and Hidden Markov Model (HMM) techniques. Due to the Contrastive nature of AHS and HMM, we have observed a significant performance improvement of 22% , 6% and 23% true acceptance rate (TAR) at 5% false acceptance rate (FAR), when this fusion technique was evaluated on three different datasets -- YOHO, USF multi-modal biometric and Speech Accent Archive (SAA), respectively. Performance enhancement has been achieved on both the datasets; however performance on YOHO was comparatively higher than that on USF dataset, owing to the fact that USF dataset is a noisy outdoor dataset whereas YOHO is an indoor dataset. In order to further increase the speaker recognition rate at lower FARs, we combined accent information from an accent classification (AC) system with our earlier HF system. Also, in homeland security applications, speaker accent will play a critical role in the evaluation of biometric systems since users will be international in nature. So incorporating accent information into the speaker recognition/verification system is a key component that our study focused on. The proposed system achieved further performance improvements of 17% and 15% TAR at an FAR of 3% when evaluated on SAA and USF multi-modal biometric datasets. The accent incorporation method and the hybrid fusion techniques discussed in this work can also be applied to any other speaker recognition systems.

ITC/USA 2012 Conference Proceedings / The Forty-Eighth Annual International Telemetering Conference and Technical Exhibition / October 22-25, 2012 / Town and Country Resort & Convention Center, San Diego, California / Telemetry systems are migrating from links to networks. Security solutions that simply encrypt radio links no longer protect the network of Test Articles or the networks that support them. The use of network telemetry is dramatically expanding and new risks and vulnerabilities are challenging issues for telemetry networks. Most of these vulnerabilities are silent in nature and cannot be detected with simple tools such as traffic monitoring. The Intrusion Detection System (IDS) is a security mechanism suited to telemetry networks that can help detect abnormal behavior in the network. Our previous research in Network Intrusion Detection Systems focused on "Password" attacks and "Syn" attacks. This paper presents a generalized method that can detect both "Password" attack and "Syn" attack. In this paper, a K-means Clustering algorithm is used for vector quantization of network traffic. This reduces the scope of the problem by reducing the entropy of the network data. In addition, a Hidden-Markov Model (HMM) is then employed to help to further characterize and analyze the behavior of the network into states that can be labeled as normal, attack, or anomaly. Our experiments show that IDS can discover and expose telemetry network vulnerabilities using Vector Quantization and the Hidden Markov Model providing a more secure telemetry environment. Our paper shows how these can be generalized into a Network Intrusion system that can be deployed on telemetry networks.

Epilepsy is a neurological disorder characterized by recurrent seizures. Sleep problems can cooccur with epilepsy, and adversely affect seizure diagnosis and treatment. In fact, the relationship between sleep and seizures in individuals with epilepsy is a complex one. Seizures disturb sleep and sleep deprivation aggravates seizures. Antiepileptic drugs may also impair sleep quality at the cost of controlling seizures. In general, particular vigilance states may inhibit or facilitate seizure generation, and changes in vigilance state can affect the predictability of seizures. A clear understanding of sleep-seizure interactions will therefore benefit epilepsy care providers and improve quality of life in patients. Notable progress in neuroscience research—and particularly sleep and epilepsy—has been achieved through experimentation on animals. Experimental models of epilepsy provide us with the opportunity to explore or even manipulate the sleep-seizure relationship in order to decipher different aspects of their interactions. Important in this process is the development of techniques for modeling and tracking sleep dynamics using electrophysiological measurements. In this dissertation experimental and computational approaches are proposed for modeling vigilance dynamics and their utility demonstrated in nonepileptic control mice. The general framework of hidden Markov models is used to automatically model and track sleep state and dynamics from electrophysiological as well as novel motion measurements. In addition, a closed-loop sensory stimulation technique is proposed that, in conjunction with this model, provides the means to concurrently track and modulate 3 vigilance dynamics in animals. The feasibility of the proposed techniques for modeling and altering sleep are demonstrated for experimental applications related to epilepsy. Finally, preliminary data from a mouse model of temporal lobe epilepsy are employed to suggest applications of these techniques and directions for future research. The methodologies developed here have clear implications the design of intelligent neuromodulation strategies for clinical epilepsy therapy.

Part-of-speech (POS) tagging is one of the most fundamental and crucial tasks in Natural Language Processing. Chinese POS tagging is challenging because it also involves word segmentation. In this report, research will be focused on how to improve unsupervised Part-of-Speech (POS) tagging using Hidden Markov Models and the Expectation Maximization parameter estimation approach (EM-HMM). The traditional EM-HMM system uses a dictionary, which is used to constrain possible tag sequences and initialize the model parameters. This is a very crude initialization: the emission parameters are set uniformly in accordance with the tag dictionary. To improve this, word alignments can be used. Word alignments are the word-level translation correspondent pairs generated from parallel text between two languages. In this report, Chinese-English word alignment is used. The performance is expected to be better, as these two tasks are complementary to each other. The dictionary provides information on word types, while word alignment provides information on word tokens. However, it is found to be of limited benefit. In this report, another method is proposed. To improve the dictionary coverage and get better POS distribution, Modified Adsorption, a label propagation algorithm is used. We construct a graph connecting word tokens to feature types (such as word unigrams and bigrams) and connecting those tokens to information from knowledge sources, such as a small tag dictionary, Wiktionary, and word alignments. The core idea is to use a small amount of supervision, in the form of a tag dictionary and acquire POS distributions for each word (both known and unknown) and provide this as an improved initialization for EM learning for HMM. We find this strategy to work very well, especially when we have a small tag dictionary. Label propagation provides a better initialization for the EM-HMM method, because it greatly increases the coverage of the dictionary. In addition, label propagation is quite flexible to incorporate many kinds of knowledge. However, results also show that some resources, such as the word alignments, are not easily exploited with label propagation. / text

Στη παρούσα διπλωματική εργασία αναπτύσσεται ένα πρόγραμμα σε Java που εναρμονίζει μία μονοφωνική ή πολυφωνική μελωδία, η οποία θα εισάγεται στο σύστημα με τη μορφή MIDI αρχείων. Η βασική τεχνική που χρησιμοποιείται είναι τα κρυμμένα μοντέλα Markov. Εισάγονται διάφορες βελτιώσεις που ενσωματώνουν γνώσεις θεωρίας μουσική στα κρυμμένα μοντέλα Μαρκόφ. Τα πειραματικά αποτελέσματα έδειξαν ότι μπορούν να βελτιώσουν την συνολική απόδοση. / In this diploma dissertation an automatic melody harmonization program is developed, written in Java. It can harmonize either a monophonic or a polyphonic melody contained in a MIDI file using the power of hidden Markov Models. We introduce several methods which incorporate musical knowledge into hidden markov models. Experiment results show higher performance in chord recognition than the initial approach.

Date stream mining techniques can be used in tracking user behaviors as they attempt to achieve their goals. Quality metrics over stream-mined models identify potential changes in user goal attainment. When the quality of some data mined models varies significantly from nearby models—as defined by quality metrics—then the user’s behavior is automatically flagged as a potentially significant behavioral change. Decision tree, sequence pattern and Hidden Markov modeling being used in this study. These three types of modeling can expose different aspect of user’s behavior. In case of decision tree modeling, the specific changes in user behavior can automatically characterized by differencing the data-mined decision-tree models. The sequence pattern modeling can shed light on how the user changes his sequence of actions and Hidden Markov modeling can identifies the learning transition points. This research describes how model-quality monitoring and these three types of modeling as a generic framework can aid recognition and diagnoses of behavioral changes in a case study of cognitive rehabilitation via emailing. The date stream mining techniques mentioned are used to monitor patient goals as part of a clinical plan to aid cognitive rehabilitation. In this context, real time data mining aids clinicians in tracking user behaviors as they attempt to achieve their goals. This generic framework can be widely applicable to other real-time data-intensive analysis problems. In order to illustrate this fact, the similar Hidden Markov modeling is being used for analyzing the transactional behavior of a telecommunication company for fraud detection. Fraud similarly can be considered as a potentially significant transaction behavioral change.

The multitude of services delivered over the Internet would have been difficult to fathom 40 years ago when much of the initial design was being undertaken. As a consequence, the resulting architecture did not make provisions for differentiating between, and managing the potentially conflicting requirements of different types of services such as real-time voice communication and peer-to-peer file sharing. This shortcoming has resulted in a situation whereby services with conflicting requirements often interfere with each other and ultimately decrease the effectiveness of the Internet as an enabler of new and transformative services. The ability to passively identify different types of Internet traffic then would address this shortcoming and enable effective management of conflicting types of services, in addition to facilitating a better understanding of how the Internet is used in general. Recent attempts at developing such techniques have shown promising results in simulation environments but perform considerably worse when deployed into real-world scenarios. One possible reason for this descrepancy can be attributed to the implicit assumption shared by recent approaches regarding the degree of similarity between the many networks which comprise the Internet. This thesis quantifies the degradation in performance which can be expected when such an assumption is violated as well as demonstrating alternative classification techniques which are less sensitive to such violations.

Ce travail de thèse poursuit une perspective double dans l'usage conjoint des méthodes de Monte Carlo séquentielles (MMS) et de l'algorithme Espérance-Maximisation (EM) dans le cadre des modèles de Markov cachés présentant une structure de dépendance markovienne d'ordre supérieur à 1 au niveau de la composante inobservée. Tout d'abord, nous commençons par un exposé succinct de l'assise théorique des deux concepts statistiques à travers les chapitres 1 et 2 qui leurs sont consacrés. Dans un second temps, nous nous intéressons à la mise en pratique simultanée des deux concepts au chapitre 3 et ce dans le cadre usuel où la structure de dépendance est d'ordre 1. L'apport des méthodes MMS dans ce travail réside dans leur capacité à approximer efficacement des fonctionnelles conditionnelles bornées, notamment des quantités de filtrage et de lissage dans un cadre non linéaire et non gaussien. Quant à l'algorithme EM, il est motivé par la présence à la fois de variables observables et inobservables (ou partiellement observées) dans les modèles de Markov Cachés et singulièrement les mdèles de volatilité stochastique étudié. Après avoir présenté aussi bien l'algorithme EM que les méthodes MCs ainsi que quelques unes de leurs propriétés dans les chapitres 1 et 2 respectivement, nous illustrons ces deux outils statistiques au travers de la calibration d'un modèle de volatilité stochastique. Cette application est effectuée pour des taux change ainsi que pour quelques indices boursiers au chapitre 3. Nous concluons ce chapitre sur un léger écart du modèle de volatilité stochastique canonique utilisé ainsi que des simulations de Monte Carlo portant sur le modèle résultant. Enfin, nous nous efforçons dans les chapitres 4 et 5 à fournir les assises théoriques et pratiques de l'extension des méthodes Monte Carlo séquentielles notamment le filtrage et le lissage particulaire lorsque la structure markovienne est plus prononcée. En guise d'illustration, nous donnons l'exemple d'un modèle de volatilité stochastique dégénéré dont une approximation présente une telle propriété de dépendance.

The fast clearing of faults is essential for preventing equipment damage and preserving the stability of the power transmission systems with smaller operating margins. This thesis examined the application of fault generated transients for fast detection and isolation of faults in a transmission system. The basis of the transient based protection scheme developed and implemented in this thesis is the fault current directions identified by a set of relays located at different nodes of the system. The direction of the fault currents relative to a relay location is determined by comparing the signs of the wavelet coefficients of the currents measured in all branches connected to the node. The faulted segment can be identified by combining the fault directions identified at different locations in the system. In order to facilitate this, each relay is linked with the relays located at the adjacent nodes through a telecommunication network. In order to prevent possible malfunctioning of relays due to transients originating from non-fault related events, a transient recognition system to supervise the relays is proposed. The applicability of different classification methods to develop a reliable transient recognition system was examined. A Hidden Markov Model classifier that utilizes the energies associated with the wavelet coefficients of the measured currents as input features was selected as the most suitable solution. Performance of the protection scheme was evaluated using a high voltage transmission system simulated in PSCAD/EMTDC simulation software. The custom models required to simulate the complete protection scheme were implemented in PSCAD/EMTDC. The effects of various factors such as fault impedance, signal noise, fault inception angle and current transformer saturation were investigated. The performance of the protection scheme was also tested with the field recorded signals. Hardware prototypes of the fault direction identification scheme and the transient classification system were implemented and tested under different practical scenarios using input signals generated with a real-time waveform playback instrument. The test results presented in this thesis successfully demonstrate the potential of using transient signals embedded in currents for detection, localization and recognition of faults in transmission networks in a fast and reliable manner.

An ergodic hidden Markov model (EHMM) can be useful in extracting underlying structure embedded in connected speech without the need for a time-aligned transcribed corpus. In this research, we present a query-by-example (QbE) spoken term detection system based on an ergodic hidden Markov model of speech. An EHMM-based representation of speech is not invariant to speaker-dependent variations due to the unsupervised nature of the training. Consequently, a single phoneme may be mapped to a number of EHMM states. The effects of speaker-dependent and context-induced variation in speech on its EHMM-based representation have been studied and used to devise schemes to minimize these variations. Speaker-invariance can be introduced into the system by identifying states with similar perceptual characteristics. In this research, two unsupervised clustering schemes have been proposed to identify perceptually similar states in an EHMM. A search framework, consisting of a graphical keyword modeling scheme and a modified Viterbi algorithm, has also been implemented. An EHMM-based QbE system has been compared to the state-of-the-art and has been demonstrated to have higher precisions than those based on static clustering schemes.