Performance Of Hidden Markov Models Research Articles

A Novel Approach for Speech Recognition of Malayalam Using Enhanced MFCC Features

The study Investigates into the realm of speech recognition, particularly focusing on the evaluation of Hidden Markov Models (HMM), leveraging the MalayalamVoice dataset. The research scrutinizes the performance of HMM concerning word error rate (WER) and accuracy across varied word lengths, revealing a consistent trend of increased WER and decreased accuracy with longer utterances. This underscores the challenges inherent in accurately transcribing extended speech segments, accentuating the necessity for algorithmic enhancements. Moreover, analyses across diverse datasets and noisy environments underscore the criticality of comprehending dataset characteristics for refining recognition algorithms. Additionally, comparisons of different feature extraction methods elucidate the efficacy of Enhanced MFCC, particularly for shorter word lengths. However, as the word length extends, the distinctions between extraction methods diminish, highlighting the multifaceted nature of speech recognition. Overall, this study underscores the intricacies involved in speech recognition and the imperative of algorithmic refinements for augmenting accuracy, especially in practical scenarios.

Principal components-based hidden Markov model for automatic detection of whale vocalisations

Over the years, researchers have continued to put forward solutions to lessen the threats faced by whales within their ecosystem. The correct detection of the different species of whale is important in the search for solutions that will lessen the threats. In order to accurately detect and classify whale species, a number of techniques have been proposed over the years, with varying degrees of success. This research seeks to improve the performance of the hidden Markov models (HMMs), which is one of the most consistent methods for the detection and classification of whale vocalisations. The performance of HMMs is affected by the quality of the feature vectors fed into them. Thus, this research proposes feature extraction (FE) techniques based on principal component analysis. The principal components (PC) computed from PCA and kernel PCA were uniquely transformed into feature vector structures suitable for the HMMs. The emerging models, PCA-HMMs and kPCA-HMMs, were experimented with on passive acoustic monitoring (PAM) datasets containing southern right whale and humpback whale vocalisations. The results from the experiments showed that the kPCA-HMMs outperformed PCA-HMMs. This is due to kPCA’s ability to find non-linear subspaces that may exist in whale vocalisations. Furthermore, the results of the developed PC-HMMs were compared with other existing FE techniques used with HMMs in the literature for the detection of whale vocalisations. The proposed PC-HMMs did not only outperform the existing FE-HMMs but also offered lower computational complexity than the existing HMMs for the detection of whale vocalisations.

은닉 마르코프 모델을 이용한 적응형 학습경로 생성에 관한 연구

Due to the global impact of COVID-19, the use of non-face-to-face learning is increasing. For non-face-to-face learning, it is important to create a learning path based on efficiency. This study introduces the hidden Markov model (HMM) as a method of creating a learning path known as a network in which knowledge concepts are arranged in order, that is, the path of experience that students may encounter in class. and it aims to improve the accuracy of path prediction by using a variable selection technique that includes least absolute shrinkage and selection operator (LASSO), and random forest (RF) before performing HMM. In addition, this study aims to show that the learning path based on higher-order concepts made of precedence relationships from HMM is more accurate than other candidate paths. As a result of using data shared by AI-hub (https://aihub.or.kr/), the performance of HMM when selecting relational pairs using LASSO, and RF was improved significantly, and the case of using HMM when evaluating the learning path consisting of higher concepts was excellent in terms of model goodness of fit compared to other models.

Cyst identification in retinal optical coherence tomography images using hidden Markov model

Optical Coherence Tomography (OCT) is a useful imaging modality facilitating the capturing process from retinal layers. In the salient diseases of retina, cysts are formed in retinal layers. Therefore, the identification of cysts in the retinal layers is of great importance. In this paper, a new method is proposed for the rapid detection of cystic OCT B-scans. In the proposed method, a Hidden Markov Model (HMM) is used for mathematically modelling the existence of cyst. In fact, the existence of cyst in the image can be considered as a hidden state. Since the existence of cyst in an OCT B-scan depends on the existence of cyst in the previous B-scans, HMM is an appropriate tool for modelling this process. In the first phase, a number of features are extracted which are Harris, KAZE, HOG, SURF, FAST, Min-Eigen and feature extracted by deep AlexNet. It is shown that the feature with the best discriminating power is the feature extracted by AlexNet. The features extracted in the first phase are used as observation vectors to estimate the HMM parameters. The evaluation results show the improved performance of HMM in terms of accuracy.

The Stock Prices Prediction Performance of Hidden Markov Models in the Luxury Category

The Stock Prices Prediction Performance of Hidden Markov Models in the Luxury Category

Using Accelerometers to Detect Activity Type in a Sport Setting: Challenges with Using Multiple Types of Conventional Machine Learning Approaches

ABSTRACT A universal approach to characterizing sport-related physical activity (PA) types in sport settings does not yet exist. Young adults (n = 30), 19–33 years, engaged in a 15-min activity session, performing warm-ups, 3-on-3 soccer, and 3-on-3 basketball. Videos were recorded and manually coded as criterion PA types (walking, running, jumping, rapid lateral movements). Participants wore an accelerometer on their right hip. Multiple machine learning models were developed and compared for predicting PA type. Most models underestimated time spent completing the activities performed least commonly. Point estimates for percent agreement, sensitivity, specificity, F-scores, and kappa were similar across models, with Hidden Markov Models (HMMs) being best at classifying rare events. Models detected activity type during sport-related movements with modest accuracy (kappas ≤ .40). Given the better performance of HMMs, incorporating the temporal nature of sport-related activities is important for improving sport-related PA classification.

Stock price forecasting using hidden Markov model

We used hidden Markov model (HMM) with single observation to estimate stock prices of selected manufacturing companies from the Nigerian Stock Exchange. Data from 22 November 2013 to 6 July 2018 were partitioned into two datasets for training and testing. Subsequently, the data were differenced, trained, tested and used to forecast closing prices for 60 days for each equity. The HMM was implemented with Matlab. The research revealed closing price prediction accuracy ranging from 3.33% to 96.67% and trade signal precision ranging from 31.67% to 97.67%. Also, the MAE values range from 0.0013 to 34.2867 while the MAPE values are between 0.1498% and 6.0034%. The hypothesis tested revealed that the model is efficient. Similarly, the comparison test conducted revealed the performance of HMM is better than ARIMA and neural network (NN). The research proposes that hidden Markov model be adopted in the exercise of stock price forecasting.

Assessment of the performance of hidden Markov models for imputation in animal breeding

BackgroundIn this paper, we review the performance of various hidden Markov model-based imputation methods in animal breeding populations. Traditionally, pedigree and heuristic-based imputation methods have been used for imputation in large animal populations due to their computational efficiency, scalability, and accuracy. Recent advances in the area of human genetics have increased the ability of probabilistic hidden Markov model methods to perform accurate phasing and imputation in large populations. These advances may enable these methods to be useful for routine use in large animal populations, particularly in populations where pedigree information is not readily available.MethodsTo test the performance of hidden Markov model-based imputation, we evaluated the accuracy and computational cost of several methods in a series of simulated populations and a real animal population without using a pedigree. First, we tested single-step (diploid) imputation, which performs both phasing and imputation. Second, we tested pre-phasing followed by haploid imputation. Overall, we used four available diploid imputation methods (fastPHASE, Beagle v4.0, IMPUTE2, and MaCH), three phasing methods, (SHAPEIT2, HAPI-UR, and Eagle2), and three haploid imputation methods (IMPUTE2, Beagle v4.1, and Minimac3).ResultsWe found that performing pre-phasing and haploid imputation was faster and more accurate than diploid imputation. In particular, among all the methods tested, pre-phasing with Eagle2 or HAPI-UR and imputing with Minimac3 or IMPUTE2 gave the highest accuracies with both simulated and real data.ConclusionsThe results of this study suggest that hidden Markov model-based imputation algorithms are an accurate and computationally feasible approach for performing imputation without a pedigree when pre-phasing and haploid imputation are used. Of the algorithms tested, the combination of Eagle2 and Minimac3 gave the highest accuracy across the simulated and real datasets.

Monthly streamflow forecasting based on hidden Markov model and Gaussian Mixture Regression

Reliable streamflow forecasts can be highly valuable for water resources planning and management. In this study, we combined a hidden Markov model (HMM) and Gaussian Mixture Regression (GMR) for probabilistic monthly streamflow forecasting. The HMM is initialized using a kernelized K-medoids clustering method, and the Baum–Welch algorithm is then executed to learn the model parameters. GMR derives a conditional probability distribution for the predictand given covariate information, including the antecedent flow at a local station and two surrounding stations. The performance of HMM–GMR was verified based on the mean square error and continuous ranked probability score skill scores. The reliability of the forecasts was assessed by examining the uniformity of the probability integral transform values. The results show that HMM–GMR obtained reasonably high skill scores and the uncertainty spread was appropriate. Different HMM states were assumed to be different climate conditions, which would lead to different types of observed values. We demonstrated that the HMM–GMR approach can handle multimodal and heteroscedastic data.

Inferring Object Properties with a Tactile-Sensing Array Given Varying Joint Stiffness and Velocity

Whole-arm tactile sensing enables a robot to sense contact and infer contact properties across its entire arm. Within this paper, we demonstrate that using data-driven methods, a humanoid robot can infer mechanical properties of objects from contact with its forearm during a simple reaching motion. A key issue is the extent to which the performance of data-driven methods can generalize to robot actions that differ from those used during training. To investigate this, we developed an idealized physics-based lumped element model of a robot with a compliant joint making contact with an object. Using this physics-based model, we performed experiments with varied robot, object and environment parameters. We also collected data from a tactile-sensing forearm on a real robot as it made contact with various objects during a simple reaching motion with varied arm velocities and joint stiffnesses. The robot used 1-nearest-neighbor (1-NN) classifiers, hidden Markov models (HMMs), and long short-term memory (LSTM) networks to infer two object properties (hard versus soft and moved versus unmoved) based on features of time-varying tactile sensor data (maximum force, contact area, and contact motion). We found that, in contrast to 1-NN, the performance of LSTMs (with sufficient data availability) and multivariate HMMs successfully generalized to new robot motions with distinct velocities and joint stiffnesses. Compared to single features, using multiple features gave the best results for both experiments with physics-based models and a real-robot.

HUMAN IDENTIFICATION BASED ON HEART SOUND AUSCULTATION POINT

The application of human identification and verification has widely been used for over the past few decades. Drawbacks of such system however, are inevitable as forgery sophisticatedly developed alongside the technology advancement. Thus, this study proposed a research on the possibility of using heart sound as biometric. The main aim is to find an optimal auscultation point of heart sounds from either aortic, pulmonic, tricuspid or mitral that will most suitable to be used as the sound pattern for personal identification. In this study, the heart sound was recorded from 92 participants using a Welch Allyn Meditron electronic stethoscope whereas Meditron Analyzer software was used to capture the signal of heart sounds and ECG simultaneously for duration of 1 minute. The system is developed by a combination Mel Frequency Cepstrum Coefficients (MFCC) and Hidden Markov Model (HMM). The highest recognition rate is obtained at aortic area with 98.7% when HMM has 1 state and 32 mixtures, the lowest Equal Error Rate (EER) achieved was 0.9% which is also at aortic area. In contrast, the best average performance of HMM for every location is obtained at mitral area with 99.1% accuracy and 17.7% accuracy of EER at tricuspid area.

Experimental analysis of hidden Markov model based secure misuse intrusion trace classification and hacking detection

In this paper, we consider hidden Markov model (HMM) based sequence classification to misuse based intrusion detection. HMM is one of statistical Markov models that regards the system as a group of observable states and hidden states. We apply HMM for detecting intrusive program traces in some public benchmark datasets including University of New Mexico (UNM) and Massachusetts Institute of Technology: Lincoln Laboratory (MIT LL) datasets. We compare the performance of HMM with that of Naive Bayes (NB) classification algorithm, support vector machines (SVM), and other basic machine learning algorithms. Our experiments and their results on the UNM and MIT LL datasets show that HMM shows comparable performance to previous methods.

Performance of Hidden Markov Models in Recovering the Standard Classification of Glycoside Hydrolases.

Glycoside hydrolases (GHs) are carbohydrate-active enzymes that assist the hydrolysis of glycoside bonds of complex sugars into carbohydrates. The current standard GH family classification is available in the CAZy database, which is based on the similarities of amino acid sequences and curated semi-automatically. However, with the exponential increase in data availability from genome sequences, automated classification methods are required for the fast annotation of coding sequences. Currently, the dbCAN database offers automatic annotations of signature domains from CAZy-defined classifications using a statistical approach, the hidden Markov models (HMMs). However, dbCAN does not contain the entire set of CAZy GH families. Moreover, no evaluation has been conducted so far of the viability of using HMM profiles as a means of automatically assigning GH amino acid sequences to the standard CAZy GH family classification itself. In this work, we performed a meta-analysis in which amino acid sequences from CAZy-defined GH families were used to build HMM family-specific profiles. We then queried a set with ~300 000 GH sequences against our database of HMM profiles estimated from CAZy families. We conducted the same evaluation against the available dbCAN HMM profiles. Our analyses recovered 65% of matches with the standard CAZy classification, whereas dbCAN HMMs resulted in 61% of matches. We also provided an analysis of the types of errors commonly found when HMMs are used to recover CAZy-based classifications. Although the performance of HMM was good, further developments are necessary for a fully automated classification of GH, allowing the standardization of GH classification among protein databases.

Statistical analysis of features and classification of alphasyllabary sounds in Kannada language

Automatic speech recognition (ASR) for a given audio file is a challenging task due to the variations in the type of speech input. Variations may be the environment, language spoken, emotions of the speaker, age/gender of speaker etc. The two main steps in ASR are converting the audio file into features and classifying it appropriately. Basic unit of speech sound is phoneme and the list of such phoneme is language dependent. In Indian languages, basic unit of language is known as Akshara i.e the alphabet. It is known to be an alphasyllabary unit. In our work, we have analyzed the behavior of the acoustic features like, Mel frequency cepstral coefficients and linear predictive coding for various aksharas using techniques like, visualization, probability density function (pdf), Q---Q plot and F-ratio. The classifiers, support vector machine (SVM) and hidden Markov model (HMM) are used for classifying the recorded audio into corresponding aksharas. We have also compared the classification performance of HMM and SVM.

Hidden Markov model and support vector machine based decoding of finger movements using electrocorticography

Objective. Support vector machines (SVM) have developed into a gold standard for accurate classification in brain–computer interfaces (BCI). The choice of the most appropriate classifier for a particular application depends on several characteristics in addition to decoding accuracy. Here we investigate the implementation of hidden Markov models (HMM) for online BCIs and discuss strategies to improve their performance. Approach. We compare the SVM, serving as a reference, and HMMs for classifying discrete finger movements obtained from electrocorticograms of four subjects performing a finger tapping experiment. The classifier decisions are based on a subset of low-frequency time domain and high gamma oscillation features. Main results. We show that decoding optimization between the two approaches is due to the way features are extracted and selected and less dependent on the classifier. An additional gain in HMM performance of up to 6% was obtained by introducing model constraints. Comparable accuracies of up to 90% were achieved with both SVM and HMM with the high gamma cortical response providing the most important decoding information for both techniques. Significance. We discuss technical HMM characteristics and adaptations in the context of the presented data as well as for general BCI applications. Our findings suggest that HMMs and their characteristics are promising for efficient online BCIs.

A Comparative Study of Hidden Markov Model and Support Vector Machine in Anomaly Intrusion Detection

This paper aims to analyse the performance of Hidden Markov Model (HMM) and Support Vector Machine (SVM) for anomaly intrusion detection. These techniques discriminate between normal and abnormal behaviour of network traffic. The specific focus of this study is to investigate and identify distinguishable TCP services that comprise of both normal and abnormal types of TCP packets, using J48 decision tree algorithm. The publicly available KDD Cup 1999 dataset has been used in training and evaluation of such techniques. Experimental results demonstrate that the HMM is able to classify network traffic with approximately 76% to 99% accuracy while SVM classifies it with approximately 80% to 99% accuracy.

Performance of Hidden Markov Model in Landslide Prediction

Performance of Hidden Markov Model in Landslide Prediction

Prediction of 24-hour-average PM2.5 concentrations using a hidden Markov model with different emission distributions in Northern California

Prediction of air pollutant levels plays an important role in the regulatory plans aimed at the control and reduction of airborne pollutants such as fine particulate matter (PM). Deterministic photochemical air quality models, which are commonly used for regulatory management and planning, are computationally intensive and also expensive for routine predictions. Compared to deterministic photochemical air quality models, data-driven statistical models are simpler and may be more accurate. In this paper, hidden Markov models (HMM) are used to forecast daily average PM2.5 concentrations 24h ahead. In conventional HMM applications, observation distributions emitted from certain hidden states are assumed as having Gaussian distributions. However, certain key meteorological factors and most PM2.5 precursors exhibit a non-Gaussian distribution in reality, which would degrade the HMM performance significantly. In order to address this problem, in this paper, HMMs with log-normal, Gamma and generalized extreme value (GEV) distributions are developed to predict PM2.5 concentration at Concord and Sacramento monitors in Northern California. Results show that HMM with non-Gaussian emission distributions is able to predict PM2.5 exceedance days correctly and reduces false alarms dramatically. Compared to HMM with Gaussian distributions, HMM with log-normal distributions can improve the true prediction rate (TPR) by 37.5% and reduce the false alarms by 78% at Concord. And HMM with GEV distribution can improve TPR by 150% and reduce false alarms by 63.62% at Sacramento Del Paso Manor. Comparisons between different distributions used in HMM show that the closer the distribution employed in HMM is to the observation sequence, the better the model prediction performance.

Ozone Prediction Performance of HMM Based on the Data Compressed by Different Wavelet Basis Functions

Surface ozone in the air boundary layer is one of the most harmful air pollutants produced by photochemical reaction between nitrogen oxides and volatile hydrocarbons, which causes great damage to human beings and environment. The prediction of surface ozone levels plays an important role in the control and the reduction of air pollutants. As model-driven statistical prediction models, hidden Markov Models (HMMs) are rich in mathematical structure and work well in many important applications. Due to the complex structure of HMM, long observation sequences would increase computational load by geometric ratio. In order to reduce training time, wavelet decomposition is used to compress the original observations into shorter ones. During compression step, observation sequences compressed by different wavelet basis functions keep different information content. This may have impact on prediction results. In this paper, ozone prediction performance of HMM based on different wavelet basis functions are discussed. Shannon entropy is employed to measure how much information content is kept in the new sequence compared to the original one. Data from Houston Metropolitan Area, TX are used in this paper. Results show that wavelet basis functions used in data compression step can affect the HMM model performance significantly. The new sequence with the maximum Shannon entropy generates the best prediction result.

Prediction of breed composition in an admixed cattle population

Swiss Fleckvieh was established in 1970 as a composite of Simmental (SI) and Red Holstein Friesian (RHF) cattle. Breed composition is currently reported based on pedigree information. Information on a large number of molecular markers potentially provides more accurate information. For the analysis, we used Illumina BovineSNP50 Genotyping Beadchip data for 90 pure SI, 100 pure RHF and 305 admixed bulls. The scope of the study was to compare the performance of hidden Markov models, as implemented in structure software, with methods conventionally used in genomic selection [BayesB, partial least squares regression (PLSR), least absolute shrinkage and selection operator (LASSO) variable selection)] for predicting breed composition. We checked the performance of algorithms for a set of 40492 single nucleotide polymorphisms (SNPs), subsets of evenly distributed SNPs and subsets with different allele frequencies in the pure populations, using F(ST) as an indicator. Key results are correlations of admixture levels estimated with the various algorithms with admixture based on pedigree information. For the full set, PLSR, BayesB and structure performed in a very similar manner (correlations of 0.97), whereas the correlation of LASSO and pedigree admixture was lower (0.93). With decreasing number of SNPs, correlations decreased substantially only for 5% or 1% of all SNPs. With SNPs chosen according to F(ST) , results were similar to results obtained with the full set. Only when using 96 and 48 SNPs with the highest F(ST) , correlations dropped to 0.92 and 0.90 respectively. Reducing the number of pure animals in training sets to 50, 20 and 10 each did not cause a drop in the correlation with pedigree admixture.