Search

Audio bandwidth extension aims to expand the spectrum of narrow-band audio signals. Although this topic has been broadly studied during recent years, the particular problem of extending the bandwidth of historical music recordings remains an open challenge. This paper proposes BEHM-GAN, a model based on generative adversarial networks, as a practical solution to this problem. The proposed method works with the complex spectrogram representation of audio and, thanks to a dedicated regularization strategy, can effectively extend the bandwidth of out-of-distribution real historical recordings. The BEHM-GAN is designed to be applied as a second step after denoising the recording to suppress any additive disturbances, such as clicks and background noise. We train and evaluate the method using solo piano classical music. The proposed method outperforms the compared baselines in both objective and subjective experiments. The results of a formal blind listening test show that BEHM-GAN significantly increases the perceptual sound quality in early-20th-century gramophone recordings. For several items, there is a substantial improvement in the mean opinion score after enhancing historical recordings with the proposed bandwidth-extension algorithm. This study represents a relevant step toward data-driven music restoration in real-world scenarios., Comment: Accepted at IEEE Transactions on Audio, Speech, and Language Processing

This paper investigates the interval type-2 (IT2) Takagi-Sugeno (T-S) fuzzy asynchronous controller design problem for nonlinear multiagent systems via a dynamic event-triggered scheme in the discrete-time context. Essentially different from most current literature, considering a realistic situation that the system mode and the anticipant controller mode are hardly to maintain synchronization at any time, the problem is characterized by means of hidden Markov model (HMM). The primary attention is focused on devising a feasible dynamic event-triggered strategy with novel threshold parameters to mitigate the communication burden efficiently. On this occasion, the information renewal of the controller is aperiodic. Furthermore, the nonlinear characteristics are effectually disposed through utilizing an unique IT2 T-S fuzzy model, which is with mismatched membership functions (MFs). As a result, the derived closed-loop fuzzy multiagent systems are accompanied by mismatched MFs and asynchronous modes. Whereafter, via solving the convex optimization problem, the desired controller gains are acquired. Eventually, the validity and practicability of the developed control scheme is illustrated by two examples.

Energy disaggregation, known in the literature as Non-Intrusive Load Monitoring (NILM), is the task of inferring the energy consumption of each appliance given the aggregate signal recorded by a single smart meter. In this paper, we propose a novel two-stage optimization-based approach for energy disaggregation. In the first phase, a small training set consisting of disaggregated power profiles is used to estimate the parameters and the power states by solving a mixed integer programming problem. Once the model parameters are estimated, the energy disaggregation problem is formulated as a constrained binary quadratic optimization problem. We incorporate penalty terms that exploit prior knowledge on how the disaggregated traces are generated, and appliance-specific constraints characterizing the signature of different types of appliances operating simultaneously. Our approach is compared with existing optimization-based algorithms both on a synthetic dataset and on three real-world datasets. The proposed formulation is computationally efficient, able to disambiguate loads with similar consumption patterns, and successfully reconstruct the signatures of known appliances despite the presence of unmetered devices, thus overcoming the main drawbacks of the optimization-based methods available in the literature.

Although speech driven facial animation has been studied extensively in the literature, works focusing on the affective content of the speech are limited. This is mostly due to the scarcity of affective audio-visual data. In this article, we improve the affective facial animation using domain adaptation by partially reducing the data scarcity. We first define a domain adaptation to map affective and neutral speech representations to a common latent space in which cross-domain bias is smaller. Then the domain adaptation is used to augment affective representations for each emotion category, including angry, disgust, fear, happy, sad, surprise, and neutral, so that we can better train emotion-dependent deep audio-to-visual (A2V) mapping models. Based on the emotion-dependent deep A2V models, the proposed affective facial synthesis system is realized in two stages: first, speech emotion recognition extracts soft emotion category likelihoods for the utterances; then a soft fusion of the emotion-dependent A2V mapping outputs form the affective facial synthesis. Experimental evaluations are performed on the SAVEE audio-visual dataset. The proposed models are assessed with objective and subjective evaluations. The proposed affective A2V system achieves significant MSE loss improvements in comparison to the recent literature. Furthermore, the resulting facial animations of the proposed system are preferred over the baseline animations in the subjective evaluations., Scientific and Technological Council of Turkey (TÜBİTAK)

Refereed/Peer-reviewed The hedging of European contingent claims in a continuous-time hidden Markov-regime-switching diffusion model is discussed using stochastic flows of diffeomorphisms and Monte-Carlo simulations. Specifically, the price dynamics of an underlying risky asset are governed by a continuous-time hidden Markov-modulated local-volatility model. Filtering theory is used to estimate the unobservable drift given observable price information and to define a filtered market with complete observations. The delta–hedge ratio of a European option is derived using a martingale representation and stochastic flows of diffeomorphisms. The numerical computation of the delta–hedge ratio is estimated via Monte-Carlo simulations. Numerical results for illustrating the proposed method and the (relative) importance of the impacts of the information risk and the local-volatility parametrizations on the delta–hedge ratio are provided for the case of European call options.

Hidden Markov Models (HMM) are widely used to represent time series with regime switching phenomenon. The most popular inference method of HMM is the Baum-Welch (B-W) algorithm, which is a special case of the E-M algorithm. However, it can be too slow for many real-time applications and is prone to being trapped into local optima. Spectral learning of HMM (SHMM), which is based on the method of moments estimation, has been proposed in the literature to overcome these obstacles. Despite its promises, asymptotic theory for SHMM has been elusive, and the long-run performance of SHMM can degrade due to unchecked propagation of error. In this thesis, for spectral estimation, we studied the theoretical property of the approximation error, improved the spectral learning by adding a 'project-onto-simplex' regularization, and provided the online learning of SHMM. We compare the performance of the proposed method with the state-of-the-art methods on both simulated data and real world applications, and find that the new method not only retains the computational advantages of SHMM, but also provides more robust estimation and forecasting. SHMM is dedicated to the endogenous evolution of HMM. Some HMM's evolution is impacted by cross predictors. We propose a variant of HMM called latent control HMM, which provides a binary latent variable controlling whether each cross predictor has impact on the transition or not. This model is inferred by standard Markov Chain Monte Carlo (MCMC) sampling, which provides the predictability and interpretability. We show the latent control HMM's performance on both simulated and real-world data sets.

This article focuses on the issue of checking critical observability for labeled Petri nets. Critical observability is a property related to the safety concern of cyber-physical systems. With the aim of checking this property of a net system, it is required to detect whether a set of markings consistent with any observed word of the net system is a subset of a set of critical states representing undesirable operations or a set of noncritical states. In this work, we prove a necessary and sufficient condition to check critical observability when the critical state set is described by an arbitrary subset of reachable markings. Then, the result is extended to the case when a critical state set is modeled by all the reachable markings that satisfy disjunctions of generalized mutual exclusion constraints. The proposed method is derived from the solutions of integer linear programming problems and is applicable to net systems with liveness and boundness. Several case studies show the performance of the presented methodology for discrete-event systems.

From sirens to lane markings, the urban environment is full of sounds that are designed to navigate the attention of the driver towards events that require special care. Microphone-equipped autonomous vehicles can also use these acoustic cues for increasing safety and performance. This article explores auditory perception in the context of autonomous driving and smart vehicles in general, examining the potential of exploiting acoustic cues in driverless vehicle technology. With a journey through the literature, we discuss various applications of auditory perception in driverless vehicles, ranging from the identification and localisation of external acoustic objects to leveraging ego-noise for motion estimation and engine fault detection. In addition to solutions already proposed in the literature, we also point out directions for further investigations, focusing in particular on parallel studies in the areas of acoustics and audio signal processing that demonstrate the potential for improving the performance of driverless cars.

This paper proposes a new design of an event-based dissipative asynchronous controller for the interval type2 (IT2) fuzzy Markov jump systems (MJSs) subject to sensor saturation and actuator nonlinearity. By resorting to a generalized performance index, the $H_{\infty}$, passive, and dissipative fuzzy control problems are solved in a unified framework. The event-based scheme is developed for the IT2 fuzzy MJSs subject to sensor saturation and actuator nonlinearity, and the energy consumption of communication can be reduced. Moreover, the system and controller modes are asynchronous, and a hidden Markov model (HMM) is employed to observe the modes of the original system. The Membership-Function-Dependent (MFD) approach is applied to analyze the stability of the closed-loop system. Finally, two examples are given to demonstrate the effectiveness of the proposed algorithms.

Speaker recognition is the process of recognizing a speaker from his speech. This can be used in many aspects of life, such as taking access remotely to a personal device, securing access to voice control, and doing a forensic investigation. In speaker recognition, extracting features from the speech is the most critical process. The features are used to represent the speech as unique features to distinguish speech samples from one another. In this research, we proposed the use of a combination of Wavelet and Mel Frequency Cepstral Coefficient (MFCC), Wavelet-MFCC, as feature extraction methods, and Hidden Markov Model (HMM) as classification. The speech signal is first extracted using Wavelet into one level of decomposition, then only the sub-band detail coefficient is used as the feature for further extraction using MFCC. The modeled system was applied in 300 speech datasets of 30 speakers uttering “HADIR” in the Indonesian language. K-fold cross-validation is implemented with five folds. As much as 80% of the data were trained for each fold, while the rest was used as testing data. Based on the testing, the system's accuracy using the combination of Wavelet-MFCC obtained is 96.67%. ABSTRAK: Pengecaman penutur adalah proses mengenali penutur dari ucapannya yang dapat digunakan dalam banyak aspek kehidupan, seperti mengambil akses dari jauh ke peranti peribadi, mendapat kawalan ke atas akses suara, dan melakukan penyelidikan forensik. Ciri-ciri khas dari ucapan merupakan proses paling kritikal dalam pengecaman penutur. Ciri-ciri ini digunakan bagi mengenali ciri unik yang terdapat pada sesebuah ucapan dalam membezakan satu sama lain. Penyelidikan ini mencadangkan penggunaan kombinasi Wavelet dan Mel Frekuensi Pekali Cepstral (MFCC), Wavelet-MFCC, sebagai kaedah ekstrak ciri-ciri penutur, dan Model Markov Tersembunyi (HMM) sebagai pengelasan. Isyarat penuturan pada awalnya diekstrak menggunakan Wavelet menjadi satu tahap penguraian, kemudian hanya pekali perincian sub-jalur digunakan bagi pengekstrakan ciri-ciri berikutnya menggunakan MFCC. Model ini diterapkan kepada 300 kumpulan data ucapan daripada 30 penutur yang mengucapkan kata "HADIR" dalam bahasa Indonesia. Pengesahan silang K-lipat dilaksanakan dengan 5 lipatan. Sebanyak 80% data telah dilatih bagi setiap lipatan, sementara selebihnya digunakan sebagai data ujian. Berdasarkan ujian ini, ketepatan sistem yang menggunakan kombinasi Wavelet-MFCC memperolehi 96.67%.

Large scale adoption and public acceptance of Electric Vehicles (EVs) require availability of charging stations. Electrification of transport has been identified as the one of the significant factors that would increase the power demand. Management of charger load has become a matter of concern for the power system engineers. Uncoordinated charging can be detrimental to the smooth operation of the power grid. On the contrary, smart charging gives certain amount of control over the charging process with respect to the power grid. Hence, adaptivity of the charging process of EVs in smart charging assists to meet the needs of power system as well as EV users. A smart charger can adjust the charging power according to the power available from the grid, EV user needs, and also support the grid during emergency. Smart charging enables EVs to act as flexible grid resources thereby providing ancillary services to the grid in case of emergency. Further, EV users can gain significant financial benefits through smart timing of their charging against spot market prices. This work presents a comprehensive overview of smart charging thereby explaining its perception, impact, user acceptance, global status and pilot projects. Also, case studies highlighting the benefits of smart charging are presented. This detailed elucidation of smart charging will assist the researchers, and experts of power industry as well as transport to find research initiatives on smart charging at one platform thereby promoting adoption of smart charging.

El presente trabajo describe el diseño y desarrollo de un reconocedor automático de voz disártrica en español y una interfaz gráfica para su uso, el sistema en su totalidad será denominado Sistema de Ayuda al Habla (SAH). Esto para realizar funciones de apoyo en el nivel de comunicación de personas con el trastorno de disartria. Para la función de reconocimiento del sistema se utilizó la biblioteca HTK Toolkit, siendo la técnica de modelado acústico los modelos ocultos de Markov. En las pruebas con un vocabulario de cinco palabras, se tuvo un aumento del nivel de comunicación de un 46,67 %, y una tasa de reconocimiento de voz de 65 %. Estos resultados fueron comparando la precisión del SAH con la precisión del reconocimiento humano.

Dynamic stochastic optimization models provide a powerful tool to represent sequential decision-making processes. Typically, these models use statistical predictive methods to capture the structure of the underlying stochastic process without taking into consideration estimation errors and model misspecification. In this context, we propose a data-driven prescriptive analytics framework aiming to integrate the machine learning and dynamic optimization machinery in a consistent and efficient way to build a bridge from data to decisions. The proposed framework tackles a relevant class of dynamic decision problems comprising many important practical applications. The basic building blocks of our proposed framework are: (1) a Hidden Markov Model as a predictive (machine learning) method to represent uncertainty; and (2) a distributionally robust dynamic optimization model as a prescriptive method that takes into account estimation errors associated with the predictive model and allows for control of the risk associated with decisions. Moreover, we present an evaluation framework to assess out-of-sample performance in rolling horizon schemes. A complete case study on dynamic asset allocation illustrates the proposed framework showing superior out-of-sample performance against selected benchmarks. The numerical results show the practical importance and applicability of the proposed framework since it extracts valuable information from data to obtain robustified decisions with an empirical certificate of out-of-sample performance evaluation.

Financial markets exhibit alternating periods of rising and falling prices. Stock traders seeking to make profitable investment decisions have to account for those trends, where the goal is to accurately predict switches from bullish to bearish markets and vice versa. Popular tools for modelling financial time series are hidden Markov models, where a latent state process is used to explicitly model switches among different market regimes. In their basic form, however, hidden Markov models are not capable of capturing both short- and long-term trends, which can lead to a misinterpretation of short-term price fluctuations as changes in the long-term trend. In this article, we demonstrate how hierarchical hidden Markov models can be used to draw a comprehensive picture of market behaviour, which can contribute to the development of more sophisticated trading strategies. The feasibility of the suggested approach is illustrated in two real-data applications, where we model data from the Deutscher Aktienindex and the Deutsche Bank stock. The proposed methodology is implemented in the R package fHMM, which is available on CRAN.

Animal behavioural responses to the environment ultimately affect their survival. Monitoring animal fine-scale behaviour may improve understanding of animal functional response to the environment and provide an important indicator of the welfare of both wild and domesticated species. In this study, we illustrate the application of collar-attached acceleration sensors for investigating reindeer fine-scale behaviour. Using data from 19 reindeer, we tested the supervised machine learning algorithms Random forests, Support vector machines, and hidden Markov models to classify reindeer behaviour into seven classes: grazing, browsing low from shrubs or browsing high from trees, inactivity, walking, trotting, and other behaviours. We implemented leave-one-subject-out cross-validation to assess generalizable results on new individuals. Our main results illustrated that hidden Markov models were able to classify collar-attached accelerometer data into all our pre-defined behaviours of reindeer with reasonable accuracy while Random forests and Support vector machines were biased towards dominant classes. Random forests using 5-s windows had the highest overall accuracy (85%), while hidden Markov models were able to best predict individual behaviours and handle rare behaviours such as trotting and browsing high. We conclude that hidden Markov models provide a useful tool to remotely monitor reindeer and potentially other large herbivore species behaviour. These methods will allow us to quantify fine-scale behavioural processes in relation to environmental events.

Aim: The aim of this study is to assess the interpretability of a representation of patients’ trajectories, collected from electronic health records (EHR) in the form of multivariate time series, using both time aware machine learning techniques, such as Hidden Markov Model (HMM) or neural networks and fixed frame machine learning techniques. . The quality of this representation is evaluated comparing its value as input variables for the prediction of adverse outcomes in clinical admission episodes (inpatient mortality, 30-day post-discharge readmission and 30-day post-discharge mortality) and clinical diagnosis at admission and discharge. Previous work: An unsupervised HMM model has been trained using expectation maximisation using as input data numeric multivariate time series from laboratory tests results and vital signs from EHR patients’ data. No outcome data was used in the model, or the training of the model or the assessment of the model. As an extra precaution a random subset of the data was held back as fully “unseen” data so far. This generative model has been used to represent each day of admission as a unique state instead of 23 numeric variables. The visual interpretation of this characterization of patients’ trajectories has shown relationships between states and outcomes of admission (e.g., inpatient mortality or length of admission episode) and diagnosis at admission and discharge. Some of these relationships have been confirmed using chi-squared t-tests but a more detailed assessment of the quality and interpretability of this representation will be explored in this research.

The goal of the OMNI project is to develop a computational framework for the prediction and optimization of human memory which is informed by indirect measurements of brain activity in humans. Simply put, we aim to record people's brains while they learn things, predict which things they learned successfully and which they didn't, and then devise better ways to teach them. The project applies state of the art in machine learning and cognitive modeling methods to the design of next-generation intelligent tutoring systems.

Pre-print (and github repo) for a study proposing a method for adaptive sampling in Ecological Momentary Assessment type settings, using continuous-time hidden Markov models (CTHMMs). Part of the EPSRC-funded Wearable Clinic project at the Centre for Health Informatics, University of Manchester.

Wearable and mobile technology provides new opportunities to manage health conditions remotely and unobtrusively. For example, healthcare providers can repeatedly sample a person's condition to monitor progression of symptoms and intervene if necessary. There is usually a utility-tolerability trade-off between collecting information at sufficient frequencies and quantities to be useful, and over-burdening the user or the underlying technology, particularly when active input is required from the user. Selecting the next sampling time adaptively using previous responses, so that people are only sampled at high frequency when necessary, can help to manage this trade-off. We present a novel approach to adaptive sampling using clustered continuous-time hidden Markov models. The model predicts, at any given sampling time, the probability of moving to an "alert" state, and the next sample time is scheduled when this probability has exceeded a given threshold. The clusters, each representing a distinct sub-model, allow heterogeneity in states and state transitions. The work is illustrated using longitudinal mentalhealth symptom data in 49 people collected using ClinTouch, a mobile app designed to monitor people with a diagnosis of schizophrenia. Using these data, we show how the adaptive sampling scheme behaves under different model parameters and risk thresholds, and how the average sampling can be substantially reduced whilst maintaining a high sampling frequency during high-risk periods.

Using hidden Markov models (HMM), the current study looked at how learners’ metacognitive monitoring is related to their physiological reactivity in the context of collaborative learning. The participants (N = 12, age 16–17 years, three females and nine males) in the study were high school students enrolled in an advanced physics course. The results show that during collaborative learning, the students engaged in monitoring in each self-regulated learning phase such as task understanding, planning and goal setting, task enactment, adaptation and reflection. The results of the HMM indicated that the learners’ physiological reactivity was low when monitoring occurred. The associations between the states based on the HMM provide insights not only into how learners engage in metacognitive monitoring but also about their level of physiological reactivity in each state. In conclusion, exploring aspects of metacognitive monitoring in collaborative learning can be done with the help of physiological reactions.

In this paper, we propose a novel unsupervised approach for sequence matching by explicitly accounting for the locality properties in the sequences. In contrast to conventional approaches that rely on frame-to-frame matching, we conduct matching using sequencelet or seqlet , a sub-sequence wherein the frames share strong similarities and are thus grouped together. The optimal seqlets and matching between them are learned jointly, without any supervision from users. The learned seqlets preserve the locality information at the scale of interest and resolve the ambiguities during matching, which are omitted by frame-based matching methods. We show that our proposed approach outperforms the state-of-the-art ones on datasets of different domains including human actions, facial expressions, speech, and character strokes.

Digital image capturing technologies and the internet are widely used today. These technologies make it very easy and fast to capture and share personal images in daily life. This causes difficulties in ensuring the confidentiality of private data and risks such as third persons getting hold of these data. The main goal of this study is to develop a user-friendly, powerful and effective method to encrypt digital images. For this aim, we propose a new block encryption method based on the Profile Hidden Markov Model. The method we propose consists of three main components. These are probability vector (PV), initialization vector (IV) and substitution-box (S-box). Encryption is in 24-bit blocks for color images and 8-bit blocks for grayscale images. The encryption rate in the proposed block encryption method is 0.7747 Mbit/s for color images and 1.0535 Mbit/s for the grayscale images. Theoretical analysis and experimental results confirm that the proposed encryption algorithm can provide high security both for color and grayscale images.

Human preferences and lifestyles significantly impact buildings’ energy consumption. Consequently, a better understanding of occupants’ behavior is crucial to decrease energy consumption and maintain occupants’ comfort. Occupant-centric control (OCC) strategies are effective approaches to fulfil such a purpose. As such, occupancy detection and prediction are of prime importance, particularly to manage Electric Space Heating (ESH) systems, due to the relatively slow dynamics of the temperature in dwellings. This paper proposes an Explicit Duration Hidden Markov Model (EDHMM) for unsupervised online presence detection and a hazard-based approach for occupancy prediction. Moreover, a control strategy using a cost function, weighted by occupancy predictions, and a load-shifting strategy based on time-varying electricity price are put forward. This work initially validates the consistency of the proposed approach by using synthetic data generated by a Monte Carlo simulation. Subsequently, the performance of our framework is compared with previous methods presented in the literature through experimental validation. Results demonstrate that the proposed EDHMM approach is efficient in detecting occupancy states. Besides, the results of the field implementation show the potential of the proposed control strategy to preserve occupants’ thermal comfort while decreasing the heating energy consumption.

Automatic speech recognition (ASR) technologies enable humans to communicate with computers. Isolated word recognition (IWR) is an important part of many known ASR systems. Minimizing the word error rate in cases of incremental learning is a unique challenge for developing an on-line ASR system. This paper focuses on on-line IWR using a recursive hidden Markov model (HMM) multivariate parameter estimation algorithm. The maximum likelihood method was used to estimate the unknown parameters of the model, and an algorithm for the adapted recursive EM algorithm for HMMs parameter estimation was derived. The resulting recursive EM algorithm is unique among its counterparts because of state transition probabilities calculation. It obtains more accurate parameter estimates compared to other algorithms of this type. In our experiment, the algorithm was implemented and adapted to several datasets for IWR. Thus, the recognition rate and algorithm convergence results are discussed in this work.

Supporting teaching and learning programming with learning analytics is an active area of inquiry. Most data used for learning analytics research comes from learning management systems. However, such systems were not developed to support learning programming. Therefore, educators have to resort to other systems that support the programming process, which can pose a challenge when it comes to understanding students’ learning since it takes place in different contexts. Methods that support the combination of different data sources are needed. Such methods would ideally account for the time-ordered sequence of students’ learning actions. In this article, we use a novel method (multi-channel sequence mining with Hidden Markov Models, HMMs) that allows the combination of multiple data sources, accounts for the temporal nature of students’ learning actions, and maps the transitions between different learning tactics. Our study included 291 students enrolled in a higher education programming course. Students’ trace-log data were collected from the learning management system and from a programming automated assessment tool. Data were analyzed using multi-channel sequence mining and HMM. The results reveal different patterns of students’ approaches to learning programming. High achievers start earlier to work on the programming assignments, use more independent strategies and consume learning resources more frequently, while the low achievers procrastinate early in the course and rely on help forums. Our findings demonstrate the potentials of multi-channel sequence mining and how this method can be analyzed using HMM. Furthermore, the results obtained can be of use for educators to understand students’ strategies when learning programming.

To prevent work-related musculoskeletal disorders (WMSD) the ergonomists apply manual heuristic methods to determine when the worker is exposed to risk factors. However, these methods require an observer and the results can be subjective. This paper proposes a method to automatically evaluate the ergonomic risk factors when performing a set of postures from the ergonomic assessment worksheet (EAWS). Joint angle motion data have been recorded with a full-body motion capture system. These data modeled the motion patterns of four different risk factors, with the use of hidden Markov models (HMMs). Based on the EAWS, automated scores were assigned by the HMMs and were compared to the scores calculated manually. Because the method proposed here is intrusive and requires expensive equipment, kinematic data from a reduced set of two sensors was also evaluated.

Cooperative Adaptive Cruise Control (CACC), a promising Vehicular Ad-hoc Network (VANET) application, automates transportation and improves efficiency. Vehicles form a platoon, following a leader, with their controllers automatically adjusting velocity, based on messages by other vehicles, to keep appropriate distances for safety. Towards deploying secure CACC, several proposals in academia and standardization leave significant questions unanswered. Thwarting adversaries is hard: cryptographic protection ensures access control (authentication and authorization) but falsified kinematic information by faulty insiders (platoon members with credentials, even the platoon leader) can cause platoon instability or vehicle crashes. Filtering out such adversarial data is challenging (computational cost and high false positive rates) but, most important, state-of-the-art misbehavior detection algorithms completely fail during platoon maneuvering. In this paper, we systematically investigate how and to what extent controllers for existing platooning applications are vulnerable, mounting a gamut of attacks, ranging from falsification attacks to jamming and collusion; including two novel attacks during maneuvering. We show how the existing middle-join and leave processes are vulnerable to falsification or 'privilege escalation' attacks. We mitigate such vulnerabilities and enable vehicles joining and exiting from any position (middle-join and middle-exit). We propose a misbehavior detection system that achieves an F1 score of ≈87 on identifying attacks throughout the lifetime of the platoon formation, including maneuvers. Our cyberphysical simulation framework can be extended to assess any other driving automation functionality in the presence of attackers. Part of proceedings ISBN 978-1-4503-9216-7QC 20220629

How do instantaneous rate models of insertion-deletion processes relate to distributions over pairwise sequence alignments? The only exactly-solved model is the 1991 Thorne....., We introduce a systematic method of approximating finite-time transition probabilities for continuous-time insertion-deletion models on sequences. The method uses automata theory to describe the action of an infinitesimal evolutionary generator on a probability distribution over alignments, where both the generator and the alignment distribution can be represented by pair hidden Markov models (HMMs). In general, combining HMMs in this way induces a multiplication of their state spaces; to control this, we introduce a coarse-graining operation to keep the state space at a constant size. This leads naturally to ordinary differential equations for the evolution of the transition probabilities of the approximating pair HMM. The TKF91 model emerges as an exact solution to these equations for the special case of single-residue indels. For the more general case of multiple-residue indels, the equations can be solved by numerical integration. Using simulated data, we show that the resulting distribution over alignments, when compared to previous approximations, is a better fit over a broader range of parameters. We also propose a related approach to develop differential equations for sufficient statistics to estimate the underlying instantaneous indel rates by expectation maximization. Our code and data are available at https://github.com/ihh/trajectory-likelihood.

In this digital world integrating smart city concepts, there is a tremendous scope and need for e-governance applications. Now people analyze the opinion of others before purchasing any product, hotel booking, stepping onto restaurants etc. and the respective user share their experience as a feedback towards the service. But there is no e-governance platform to obtain public opinion grievances towards covid19, government new laws, policies etc. With the growing availability and emergence of opinion rich information's, new opportunities and challenges might arise in developing a technology for mining the huge set of public messages, opinions and alert the respective departments to take necessary actions and also nearby ambulances if its related to covid-19. To overcome this pandemic situation a natural language processing based efficient e-governance platform is demandful to detect the corona positive patients and provide transparency on the covid count and also alert the respective health ministry and nearby ambulance based on the user voice inputs. To convert the public voice messages into text, we used Hidden Markov Models (HMMs). To identify respective government department responsible for the respective user voice input, we perform pre-processing, part of speech, unigram, bigram, trigram analysis and fuzzy logic (machine learning technique). After identifying the responsible department, we perform 2 methods, (1) Automatic alert e-mail and message to the government departmental officials and nearby ambulance or covid camp if the user input is related to covis19. (2) Ticketing system for public and government officials monitoring. For experimental results, we used Java based web and mobile application to execute the proposed methodology. Integration of HMM, Fuzzy logic provides promising results.

Purpose: Banking business faces a major challenge with defaults. This may not be practical, as there is no control over the borrower’s financial situation or their intents to repay. However, if the banks get to know the possible defaults ahead of some actionable time frame, with a certain degree of accuracy in such prediction, Banks may apply any a possible risk mitigation strategy to remediate possible defaults. Willingness to repay the debt and the capability to repay the debt are two primary reasons for the loan default. The subject of this paper is to closely monitor the Facebook activities and check if we can predict if the borrower may become a defaulter any soon, by applying the sentiment analysis on Facebook data and use Hidden Markov model to compute the probabilities of the possible default. Approach/Methodology/Design: The loan dataset was used for the borrower details and the Facebook data for all those borrowers were gathered. The data from Facebook posts, likes and shares on a borrower were subjected to sentiment analysis, considering income-related information of spend related information on neutral. Hidden Markov model was applied to the polarized data based on the sequence of the sentiment analysis. Findings: Hidden Markov Model gives the transition probability of state, default or regular, for the observed polarized sentiments from Facebook data for borrowers. Practical Implications: This mechanism can be integrated into the bank's credit risk management system and could help predict the possibility of a borrower becoming a defaulter. This is very much useful where the tenure of the loan is longer. This research paper fills the gap of active monitoring of the credit risk for long term loans, where the financial status of the borrower could change but the lender doesn’t get to know until the borrower stops the repayment., peer-reviewed

Prostate cancer (PCa) is a common, serious form of cancer in men that is still prevalent despite ongoing developments in diagnostic oncology. Current detection methods lead to high rates of inaccurate diagnosis. We present a method to directly model and exploit temporal aspects of temporal enhanced ultrasound (TeUS) for tissue characterization, which improves malignancy prediction. We employ a probabilistic-temporal framework, namely, hidden Markov models (HMMs), for modeling TeUS data obtained from PCa patients. We distinguish malignant from benign tissue by comparing the respective log-likelihood estimates generated by the HMMs. We analyze 1100 TeUS signals acquired from 12 patients. Our results show improved malignancy identification compared to previous results, demonstrating over 85% accuracy and AUC of 0.95. Incorporating temporal information directly into the models leads to improved tissue differentiation in PCa. We expect our method to generalize and be applied to other types of cancer in which temporal-ultrasound can be recorded.

Genetic factors have been pointed out as the primary root associated with the risk of autism. Recent works indicate that approximately 80% of autistic people have inherited the condition from their parents. However, there are no estimates that indicate the likelihood of an autistic parent having an autistic child. Using Hidden Markov Models, together with the data of autism heritability, we developed a model to investigate the likelihood of autistic parents generating autistic children. Hidden Markov Models are a double-layered stochastic process, and it consists of a nonvisible stochastic process (not observable) that can be predicted through a visible one. Our model was built and validated using statistical data from the association of gender with recurrence of autism among siblings, as well as statistical data from the association of genetic factors with autism. Our results suggest that autistic parents may generate autistic children with probabilities of ≈ 33% for female children and ≈ 80% for male children. Such estimates could assist parents in some decision making processes according to the estimated risk of autism in their descendants.

In this paper we developed a modified Hidden Markov Model (HMM) to analyze the raw nanopore experimental data. Traditionally, prior to further analysis the measured nanopore data must be pre-filtered, but the filtering usually distorts the waveform of the blockage current, especially for rapid translocations and bumping blockages. The HMM is known to be robust with respect to strong noise and thus suitable for processing the raw nanopore data, but its performance is susceptible to the setting of initial parameters. To overcome this problem, we use the Fuzzy c-Means (FCM) algorithm to initialize the HMM parameters in this work. Then we use the Viterbi training algorithm to optimize the HMM. Finally, both the simulated and experimental data analysis results are presented to show the effectiveness of the proposed method for detection of the nanopore current blockage events in analytical chemistry.

Despite the close relationship between speech perception and production, research in automatic speech recognition (ASR) and text-to-speech synthesis (TTS) has progressed more or less independently without exerting much mutual influence. In human communication, on the other hand, a closed-loop speech chain mechanism with auditory feedback from the speaker's mouth to her ear is crucial. In this paper, we take a step further and develop a closed-loop machine speech chain model based on deep learning. The sequence-to-sequence model in closed-loop architecture allows us to train our model on the concatenation of both labeled and unlabeled data. While ASR transcribes the unlabeled speech features, TTS attempts to reconstruct the original speech waveform based on the text from ASR. In the opposite direction, ASR also attempts to reconstruct the original text transcription given the synthesized speech. To the best of our knowledge, this is the first deep learning framework that integrates human speech perception and production behaviors. Our experimental results show that the proposed approach significantly improved performance over that from separate systems that were only trained with labeled data.

Accurate confidence measures for predictions from machine learning techniques play a critical role in the deployment and training of many speech and language processing applications. For example, confidence scores are important when making use of automatically generated transcriptions in training automatic speech recognition (ASR) systems, as well as down-stream applications, such as information retrieval and conversational assistants. Previous work on improving confidence scores for these systems has focused on two main directions: designing features correlated with improved confidence prediction; and employing sequence models to account for the importance of contextual information. Few studies, however, have explored incorporating contextual information more broadly, such as from the future, in addition to the past, or making use of alternative multiple hypotheses in addition to the most likely one. This article introduces two general approaches for encapsulating contextual information from lattices. Experimental results illustrating the importance of increasing contextual information for estimating confidence scores are presented on a range of limited resource languages where word error rates range between 30% and 60%. The results show that the novel approaches provide significant gains in the accuracy of confidence estimation.

Many pathways for hydrocarbon degradation have been discovered, yet there are no dedicated tools to identify and predict the hydrocarbon degradation potential of microbial genomes and metagenomes. Here we present the Calgary approach to ANnoTating HYDrocarbon degradation genes (CANT-HYD), a database of 37 HMMs of marker genes involved in anaerobic and aerobic degradation pathways of aliphatic and aromatic hydrocarbons. Using this database, we identify understudied or overlooked hydrocarbon degradation potential in many phyla. We also demonstrate its application in analyzing high-throughput sequence data by predicting hydrocarbon utilization in large metagenomic datasets from diverse environments. CANT-HYD is available at https://github.com/dgittins/CANT-HYD-HydrocarbonBiodegradation.

Background RIFINs and STEVORs are variant surface antigens expressed by P. falciparum that play roles in severe malaria pathogenesis and immune evasion. These two highly diverse multigene families feature multiple paralogs, making their classification challenging using traditional bioinformatic methods. Results STRIDE (STevor and RIfin iDEntifier) is an HMM-based, command-line program that automates the identification and classification of RIFIN and STEVOR protein sequences in the malaria parasite Plasmodium falciparum. STRIDE is more sensitive in detecting RIFINs and STEVORs than available PFAM and TIGRFAM tools and reports RIFIN subtypes and the number of sequences with a FHEYDER amino acid motif, which has been associated with severe malaria pathogenesis. Conclusions STRIDE will be beneficial to malaria research groups analyzing genome sequences and transcripts of clinical field isolates, providing insight into parasite biology and virulence.

Given the patchiness and long-term predictability of marine resources, memory of high-quality foraging grounds is expected to provide fitness advantages for central place foragers. However, it remains challenging to characterise how marine predators integrate memory with recent prey encounters to adjust fine-scale movement and use of foraging patches. Here, we used two months of movement data from harbour seals (Phoca vitulina) to quantify the repeatability in foraging patches as a proxy for memory. We then integrated these data into analyses of fine-scale movement and underwater behaviour to test how both spatial memory and prey encounter rates influenced the seals’ Area Restricted Search (ARS) behaviour. Specifically, we used one month’s GPS data from 29 individuals to build spatial memory maps of searched areas, and archived accelerometry data from a subset of five individuals to detect prey catch attempts, a proxy for prey encounters. Individuals were highly consistent in the areas they visited over two consecutive months. Hidden Markov Models showed that both spatial memory and prey encounters increased the probability of seals initiating ARS. These results provide evidence that predators use memory to adjust their fine scale movement and this ability should be accounted for in movement models.

Since Internet of Things (IoT) has been widely used in our daily life nowadays, it is regarded as a promising and popular application of the Internet, and has attracted more and more attention. However, IoT is also suffered by some security problems which seriously affect the implementation of IoT system. Similar to traditional software, IoT software is always threated by many vulnerabilities, thus how to evaluate the security situation of IoT software chain becomes a basic requirement. In this paper, A framework of security situation awareness for IoT software chain is proposed, which mainly includes two processes: IoT security situation classification based on support vector machine and security situation awareness based on Markov game model. The proposed method firstly constructs a classification model using support vector machine (IoT) to automatically evaluates the security situation of IoT software chain. Based on the situation classification, we further proposed to adopt Markov model to simulate and predict the next behaviors of participants that involved in IoT system. Additionally, we have designed and developed a security situation awareness system for IoT software chain, the developed system supports the detection of typical IoT vulnerabilities and inherits more than 20 vulnerability detection methods, which shows great potential in IoT system protection.

In this work we formulate and treat an extension of the Imitation from Observations problem. Imitation from Observations is a generalisation of the well-known Imitation Learning problem where state-only demonstrations are considered. In our treatment we extend the scope of Imitation from Observations to feature-only demonstrations which could arguably be described as partial observations. Therewith we mean that the full state of the decision makers is unknown and imitation must take place on the basis of a limited set of features. We set out for methods that extract an executable policy directly from those features which, in the literature, would be referred to as Behavioural Cloning methods. Our treatment combines elements from probability and information theory and draws connections with entropy regularized Markov Decision Processes.

The learning process depends on the nature of the learning environment, particularly in the case of open-ended learning environments, where the learning process is considered to be non-linear. In this paper, we report on the findings of employing a multimodal Hidden Markov Model (HMM) based methodology to investigate the temporal learning processes of two types of learners that have learning gains and a type that does not have learning gains in an open-ended collaborative learning activity. Considering log data, speech behavior, affective states and gaze patterns, we find that all learners start from a similar state of non-productivity, but once out of it they are unlikely to fall back into that state, especially in the case of the learners that have learning gains. Those who have learning gains shift between two problem solving strategies, each characterized by both exploratory and reflective actions, as well as demonstrate speech and gaze patterns associated with these strategies, that differ from those who don't have learning gains. Further, the teams that have learning gains also differ between themselves in the manner in which they employ the problem solving strategies over the interaction, as well as in the manner they express negative emotions while exhibiting a particular strategy. These outcomes contribute to understanding the multiple pathways of learning in an open-ended collaborative learning environment, and provide actionable insights for designing effective interventions.

The maintenance paradigm has evolved over the last few years and companies that want to remain competitive in the market need to provide condition-based maintenance (CBM). The diagnosis and prognosis of the health status of equipment, predictive maintenance (PdM), are fundamental strategies to perform informed maintenance, increasing the company’s profit. This article aims to present a diagnosis and prognosis methodology using a hidden Markov model (HMM) classifier to recognise the equipment status in real time and a deep neural network (DNN), specifically a gated recurrent unit (GRU), to determine this same status in a future of one week. The data collected by the sensors go through several phases, starting by cleaning them. After that, temporal windows are created in order to generate statistical features of the time domain to better understand the equipment’s behaviour. These features go through a normalisation to produce inputs for a feature extraction process, via a principal component analysis (PCA). After the dimensional reduction and obtaining new features with more information, a clustering is performed by the K-means algorithm, in order to group similar data. These clusters enter the HMM classifier as observable states. After training using the Baum–Welch algorithm, the Viterbi algorithm is used to find the best path of hidden states that represent the diagnosis of the equipment, containing three states: state 1—“State of Good Operation”; state 2—“Warning State”; state 3—“Failure State”. Once the equipment diagnosis is complete, the GRU model is used to predict the future, both of the observable states as well as the hidden states coming out from the HMM. Thus, through this network, it is possible to directly obtain the health states 7 days ahead, without the necessity to run the whole methodology from scratch.

G quadruplexes are short secondary DNA structures located throughout genomic DNA and transcribed RNA. though G4 structures have been shown to formin vivo, no current search tools are known to exist to examine these structures based on previously identified G quadruplexes, much less filter them based on similar sequence, structure, and thermodynamic properties. We present a framework for clustering G quadruplex sequences into families using theCD-HIT, MeShClust and DNACLUSTmethods along with a combination ofStarcodeandBLAST. Utilizing this framework to filter and annotate clusters, 95 families of G quadruplex sequences were identified within the human genome. Profiles for each family were created using hidden Markov models to allow for identification of additional family members and generate homology probability scores. The thermodynamic folding energy properties, functional annotation of genes associated with the sequences, scores from different prediction algorithms and transcription factor binding and motif to the G4 region for the sequences within a family were used to annotate and compare the diversity within and across clusters. The resulting set of G quadruplex families can be used to further understand how different regions of the genome are regulated by factors targeting specific structures common to members of a specific cluster.

Negative social media usage during the COVID-19 pandemic has highlighted the importance of understanding the spread of misinformation and toxicity in public online discussions. In this paper, we propose a novel unsupervised method to discover the structure of online COVID-19-related conversations. Our method trains a nine-state Hidden Markov Model (HMM) initialized from a biclustering of 23 features extracted from online messages. We apply our method to 16,000 conversations (1.5 million messages) that took place on the Facebook pages of 15 Canadian newspapers following COVID-19 news items, and show that it can effectively extract the conversation structure and discover the main themes of the messages. Furthermore, we demonstrate how the PageRank algorithm and the conversation graph discovered can be used to simulate the impact of five different moderation strategies, which makes it possible to easily develop and test new strategies to limit the spread of harmful messages. Although our work in this paper focuses on the COVID-19 pandemic, the methodology is general enough to be applied to handle communications during future pandemics and other crises, or to develop better practices for online community moderation in general.

In this paper, to jointly study the energy dynamic behavior of humans and the corresponding physical dynamics of the microgrid, we bridge two disciplines: systems & control and environmental psychology. Firstly, we develop second order motivation-behavior mathematical models inspired by opinion dynamics models for describing and predicting human activities related to the use of energy, where psychological variables and social interactions are considered. Secondly, based on these models, we develop a human-cyber-physical system framework consisting of three layers: (i) human, (ii) cyber and (iii) physical. The first one describes human behavior influenced by behavioral intervention and motivation, which in turn depend on contextual factors, personal values and social norms. The cyber layer solves an optimization problem and embeds load controllers, which are designed to automatically mimic human behavior. Finally, the physical layer represents an AC microgrid. Thus, we formulate a social-physical welfare optimization problem and solve it by designing a distributed primal-dual control scheme, which generates the optimal behavioral intervention (with respect to a given reference) and the control inputs to the microgrid.

Chromosomal instability characterises several cancer types, in which large-scale structural alterations of the genome accumulate at an increased rate. An important class of structural alterations are somatic copy number alterations (SCNAs). SCNAs have been shown to be major drivers of oncogenesis and are associated with prognosis and response to therapies. Current sequencing and array-based methods that are used to infer SCNAs are cost-prohibitive for widespread clinical use. A low-cost, simple and more clinically applicable method to amplify and sequence more than 10,000 repeat regions across the genome was recently developed, called FAST-SeqS. However, current computational methods do not make effective use of this low-cost assay. This limits its application to clinical medicine and to biomedical research. In this thesis, I develop conliga; a probabilistic generative model and associated inference algorithms to infer relative copy number from FAST-SeqS data at the amplicon level. I implement this method in R and C++ and provide the software as an open-source tool. By applying conliga and FAST-SeqS to oesophageal adenocarcinoma and related conditions, I show that it has similar performance to QDNAseq applied to low-coverage whole-genome sequencing, which is a more expensive and laborious alternative for SCNA profiling. I explore several aspects of FAST-SeqS data and show that sample-specific biases can affect SCNA inferences. By extending the conliga model, I demonstrate that these biases can be jointly inferred with SCNA profiles. I validate these extensions by comparing the results to inferences obtained from whole genome sequencing in prostate cancer samples. I show that the variants present in FAST-SeqS data can be used to infer tumour purity, ploidy and allele-specific copy number. This has potential application in large-scale cancer genome studies to identify samples with sufficient purity before performing high-coverage whole-genome sequencing. Finally, I describe preliminary data showing that the FAST-SeqS protocol can be applied to single cells, enabling further extensions of the conliga model which could lead to the inference of SCNAs in single cells.

Segmentation and genome annotation (SAGA) algorithms are widely used to understand genome activity and gene regulation. These algorithms take as input epigenomic datasets, such as chromatin immunoprecipitation-sequencing (ChIP-seq) measurements of histone modifications or transcription factor binding. They partition the genome and assign a label to each segment such that positions with the same label exhibit similar patterns of input data. SAGA algorithms discover categories of activity such as promoters, enhancers, or parts of genes without prior knowledge of known genomic elements. In this sense, they generally act in an unsupervised fashion like clustering algorithms, but with the additional simultaneous function of segmenting the genome. Here, we review the common methodological framework that underlies these methods, review variants of and improvements upon this basic framework, and discuss the outlook for future work. This review is intended for those interested in applying SAGA methods and for computational researchers interested in improving upon them.

Background Partially methylated domains (PMDs) are a hallmark of epigenomes in reproducible and specific biological contexts, including cancer cells, the placenta, and cultured cell lines. Existing methods for deciding whether PMDs exist in a sample, as well as their identification, are few, often tailored to specific biological questions, and require high coverage samples for accurate identification. Results In this study, we outline a set of axioms that take a step towards a functional definition for PMDs, describe an improved method for comparable PMD detection across samples with substantially differing sequencing depths, and refine the decision criteria for whether a sample contains PMDs using a data-driven approach. Applying our method to 267 methylomes from 7 species, we corroborated recent results regarding the general association between replication timing and PMD state, and report identification of several reproducibly “escapee” genes within late-replicating domains that escape the reduced expression and hypomethylation of their immediate genomic neighborhood. We also explored the discordant PMD state of orthologous genes between human and mouse, and observed a directional association of PMD state with gene expression and local gene density. Conclusions Our improved method makes low sequencing depth, population-level studies of PMD variation possible and our results further refine the model of PMD formation as one where sequence context and regional epigenomic features both play a role in gradual genome-wide hypomethylation.