Multiple Hidden Markov Models Research Articles

Stacking models of brain dynamics to improve prediction of subject traits in fMRI

Abstract Beyond structural and time-averaged functional connectivity brain measures, modelling the way brain activity dynamically unfolds can add important information to our understanding and characterisation of individual cognitive traits. One approach to leveraging this information is to extract features from models of brain network dynamics to predict individual traits. However, these predictions are susceptible to variability due to factors such as variation in model estimation induced by the choice of hyperparameters. We suggest that, rather than merely being statistical noise, this variability may be useful in providing complementary information that can be leveraged to improve prediction accuracy. To leverage this variability, we propose the use of stacking, a prediction-driven approach for model selection. Specifically, we combine predictions developed from multiple hidden Markov models—a probabilistic generative model of network dynamics that identifies recurring patterns of brain activity—to demonstrate that stacking can slightly improve the accuracy and robustness of cognitive trait predictions. By comparing analysis from the Human Connectome Project and UK Biobank datasets, we show that stacking is relatively effective at improving prediction accuracy and robustness when there are enough subjects, and that the effectiveness of combining predictions from static and dynamic functional connectivity approaches depends on the length of scan per subject. We also show that the effectiveness of stacking predictions is driven by the accuracy and diversity in the underlying model estimations.

Ensemble of HMMs for Sequence Prediction on Multivariate Biomedical Data

Background: Biomedical data are usually collections of longitudinal data assessed at certain points in time. Clinical observations assess the presences and severity of symptoms, which are the basis for the description and modeling of disease progression. Deciphering potential underlying unknowns from the distinct observation would substantially improve the understanding of pathological cascades. Hidden Markov Models (HMMs) have been successfully applied to the processing of possibly noisy continuous signals. We apply ensembles of HMMs to categorically distributed multivariate time series data, leaving space for expert domain knowledge in the prediction process. Methods: We use an ensemble of HMMs to predict the loss of free walking ability as one major clinical deterioration in the most common autosomal dominantly inherited ataxia disorder worldwide. Results: We present a prediction pipeline that processes data paired with a configuration file, enabling us to train, validate and query an ensemble of HMMs. In particular, we provide a theoretical and practical framework for multivariate time-series inference based on HMMs that includes constructing multiple HMMs, each to predict a particular observable variable. Our analysis is conducted on pseudo-data, but also on biomedical data based on Spinocerebellar ataxia type 3 disease. Conclusions: We find that the model shows promising results for the data we tested. The strength of this approach is that HMMs are well understood, probabilistic and interpretable models, setting it apart from most Deep Learning approaches. We publish all code and evaluation pseudo-data in an open-source repository.

Taxonomic distribution of opsin families inferred from UniProt Reference Proteomes and a suite of opsin-specific hidden Markov models

IntroductionOpsins are a large and sequence-diverse family of light-responsive G-protein coupled receptors involved in vision, circadian rhythm, and other processes. Numerous subfamilies have been defined based on sequence similarity, cell-type localization, signal transduction mechanism, or biological function, but there is no consensus classification system.MethodsWe used multiple hidden Markov models (HMMs) to identify opsins in the UniProt Reference Proteomes database. Opsin-specific HMMs were also used in an annotation procedure that represents sequences as a vector of HMM scores and assess the similarity of these vectors to those of annotated sequences. UniProt Reference Proteomes are built from genome sequences, allowing us to make meaningful comparisons of the number of opsins in each of the 260 species available at the time of the survey in absolute terms and relative to a larger superfamily of which opsins are a member.ResultsMore than 2,000 opsins were retrieved from 262 species (all metazoans).DiscussionMerging opsin counts into higher order taxa paints a broad view of the taxonomic distribution of opsins, and of opsin subfamilies, annotated according to three different schemes.

Integration of Phonotactic Features for Language Identification on Code-Switched Speech

In this paper, phoneme sequences are used as language information to perform code-switched language identification (LID). With the one-pass recognition system, the spoken sounds are converted into phonetically arranged sequences of sounds. The acoustic models are robust enough to handle multiple languages when emulating multiple hidden Markov models (HMMs). To determine the phoneme similarity among our target languages, we reported two methods of phoneme mapping. Statistical phoneme-based bigram language models (LM) are integrated into speech decoding to eliminate possible phone mismatches. The supervised support vector machine (SVM) is used to learn to recognize the phonetic information of mixed-language speech based on recognized phone sequences. As the back-end decision is taken by an SVM, the likelihood scores of segments with monolingual phone occurrence are used to classify language identity. The speech corpus was tested on Sepedi and English languages that are often mixed. Our system is evaluated by measuring both the ASR performance and the LID performance separately. The systems have obtained a promising ASR accuracy with data-driven phone merging approach modelled using 16 Gaussian mixtures per state. In code-switched speech and monolingual speech segments respectively, the proposed systems achieved an acceptable ASR and LID accuracy.

A Hidden Markov Model Based Detecting Solution for Detecting the Situation of Balance During Unsupported Standing Using the Electromyography of Ankle Muscles

Background: In this study, three detecting approaches have been proposed and evaluated for online detection of balance situations during quiet standing. The applied methods were based on electromyography of the gastrocnemius muscles adopting the hidden Markov models. Methods: The levels of postural stability during quiet standing were regarded as the hidden states of the Markov models while the zones in which the center of pressure lies within determines the level of stability. The Markov models were trained by using the well-known Baum-Welch algorithm. The performance of a single hidden Markov model, the multiple hidden Markov model, and the multiple hidden Markov model alongside an adaptive neuro-fuzzy inference system (ANFIS), were compared as three different detecting methods. Results: The obtained results show the better and more promising performance of the method designed based on a combination of the hidden Markov models and optimized neuro-fuzzy system. Conclusion: According to the results, using the combined detecting method yielded promising results.

Identifying Intrusion Behaviour using Enhanced Hidden Markov Model

Data Mining is a method for detecting network intrusion detection in networks. It brings ideas from variety of areas including statistics, machine learning and database processes. Decreasing price of digital networking is now economically viable for network intrusion detection. This analysis chiefly examines the system intrusion detection with machine learning and DM methods. To improve the accuracy and efficiency of SHMM, we are collecting multiple observation in SHMM that will be called as Multiple Hidden Markov Model (MHMM). It is used to improve better Detection accuracy compare with SHMM. In the standard Hidden Markov Model, we have observed three fundamental problems are Evaluation and decoding another one is learning problem. The Evaluation problem can be used for word recognition. And the Decoding problem is related to constant attention and also the segmentation. In this Proposed Research, the primary purpose is to model the sequence of observation in Network log and credit card log transactions process using Enhanced Hidden Markov Model (EHMM). And show how it can be used for intrusion detection in Network. In this procedure, an EHMM is primarily trained with the conventional manners of a intruders. If the trained EHMM does not recognize an incoming Intruder transaction with adequately high probability, it is thought to be fraudulent.

Real-Time Mode Diagnosis for Processes With Multiple Operating Conditions Using Switching Conditional Random Fields

Over the past few decades, hidden Markov models (HMMs) have been widely employed for real-time diagnosis of processes with multiple modes. Recently, the conditional random field (CRF) model has also been applied for process monitoring and proven to perform superior to the HMMs. In this paper, a new framework, termed as a switching CRF (SCRF), is designed to diagnose the modes of processes that have multiple operating conditions. In the proposed framework, multiple linear-chain CRF models are allowed to switch between each other according to a scheduling variable that reflects the real-time operating conditions. The expectation-maximization algorithm is employed for the parameter estimation of SCRF. For validation, two case studies, namely a continuous stirred tank reactor system and an experimental hybrid tank system, are employed. The results demonstrate that the proposed SCRF approach shows superior diagnosis performances to the linear-chain CRF model and the multiple HMMs for the processes with multiple operating conditions.

Multiple Hidden Markov Model for Pathological Vessel Segmentation.

One of the obstacles that prevent the accurate delineation of vessel boundaries is the presence of pathologies, which results in obscure boundaries and vessel-like structures. Targeting this limitation, we present a novel segmentation method based on multiple Hidden Markov Models. This method works with a vessel axis + cross-section model, which constrains the classifier around the vessel. The vessel axis constraint gives our method the potential to be both physiologically accurate and computationally effective. Focusing on pathological vessels, we reap the benefits of the redundant information embedded in multiple vessel-specific features and the good statistical properties coming with Hidden Markov Model, to cover the widest possible spectrum of complex situations. The performance of our method is evaluated on synthetic complex-structured datasets, where we achieve a 91% high overlap ratio. We also validate the proposed method on a real challenging case, segmentation of pathological abdominal arteries. The performance of our method is promising, since our method yields better results than two state-of-the-art methods on both synthetic datasets and real clinical datasets.

A filter model for intrusion detection system in Vehicle Ad Hoc Networks: A hidden Markov methodology

Although Vehicle Ad Hoc Network (VANETs) as a new technology is being used in wide range of applications to improve the driving experience as well as safety, it is vulnerable to various type of network attacks. Literature studies have revealed several reliable approaches based on intrusion detection system (IDS), to protect VANETs against attacks. However, by those solutions, the overheads of IDSs are serious which cause too long detection time, especially when the number of vehicles increases. In this paper, we propose a novel filter model based hidden Markov model (HMM) (FM-HMM) for IDS to reduce the overhead and time for detection without impairing detection rate. To the best of our knowledge, this is the first work in the literature to model the state pattern of each vehicle in VANETs as a HMM to quickly filter the messages from the vehicles instead of detecting these messages. The FM-HMM consists of three modules, i.e., schedule, filter and update. In the schedule module, Baum–Welch algorithm is used to produce a HMM and its parameters for each neighbor vehicle. In the filter module, multiple HMMs are used with their parameters to forecast the future states of neighbor vehicles with which the messages from them are filtered. In the update module, a timeliness method is used to update HMMs and their parameters. Experiments show that the IDS with FM-HMM has a better performance in terms of detection rate, detection time and overhead.

Multiple Hidden Markov Model Post Processed with Support Vector Machine to Recognize English Handwritten Numerals

This paper presents rotation and size invariant English numerals recognition system with, competitive recognition rate. The novelty of this paper is the introduction of two unique methods of feature extraction namely Pixel Moment of Inertia (PMI) and Delta Distance Coding (DDC). The proposed Multiple Hidden Markov Model (MHMM) is a two tier model to neutralize the effect of two very frequent writing styles of numerals ‘4’ and ‘7’ on their recognition rates. The novelty of PMI is that it finds moment of all the pixels of a specified zone about the central pixel and not about geometrical centroid of image area. In this paper, PMI has been observed to have an upper hand over centroidal MI. DDC is a new technique of curvature coding, based on distance from a reference level and is similar to the logic behind Delta modulation scheme in Digital Communications. Thus, the current paper correlates two digital domains namely, Digital Image Processing and Digital Communications. Support Vector Machine differentiates two close output classes obtained from classification with MHMM. The overall recognition accuracy rate of 99.17% has been achieved based on MNIST database.

Find who to look at: Turning from action to saliency.

The past decade has witnessed the use of highlevel features in saliency prediction for both videos and images. Unfortunately, the existing saliency prediction methods only handle high-level static features, such as face. In fact, high-level dynamic features (also called actions), such as speaking or head turning, are also extremely attractive to visual attention in videos. Thus, in this paper, we propose a data-driven method for learning to predict the saliency of multiple-face videos, by leveraging both static and dynamic features at high-level. Specifically, we introduce an eye-tracking database, collecting the fixations of 39 subjects viewing 65 multiple-face videos. Through analysis on our database, we find a set of high-level features that cause a face to receive extensive visual attention. These high-level features include the static features of face size, center-bias and head pose, as well as the dynamic features of speaking and head turning. Then, we present the techniques for extracting these high-level features. Afterwards, a novel model, namely multiple hidden Markov model (M-HMM), is developed in our method to enable the transition of saliency among faces. In our MHMM, the saliency transition takes into account both the state of saliency at previous frames and the observed high-level features at the current frame. The experimental results show that the proposed method is superior to other state-of-the-art methods in predicting visual attention on multiple-face videos. Finally, we shed light on a promising implementation of our saliency prediction method in locating the region-of-interest (ROI), for video conference compression with high efficiency video coding (HEVC).

A weighted hidden Markov model approach for continuous-state tool wear monitoring and tool life prediction

Tool wear is one of the important indicators to reflect the health status of a machining system. In order to obtain tool’s wear status, tool condition monitoring (TCM) utilizes advanced sensor techniques, hoping to find out the wear status through those sensor signals. In this paper, a novel weighted hidden Markov model (HMM)-based approach is proposed for tool wear monitoring and tool life prediction, using the signals provided by TCM techniques. To describe the dynamic nature of wear evolution, a weighted HMM is first developed, which takes wear rate as the hidden state and formulates multiple HMMs in a weighted manner to include sufficient historical information. Explicit formulas to estimate the model parameters are also provided. Then, a particular probabilistic approach using the weighted HMM is proposed to estimate tool wear and predict tool’s remaining useful life during tool operation. The proposed weighted HMM-based approach is tested on a real dataset of a high-speed CNC milling machine cutters. The experimental results show that this approach is effective in estimating tool wear and predicting tool life, and it outperforms the conventional HMM approach.

Orthogonal nonnegative matrix factorization based local hidden Markov model for multimode process monitoring

Traditional data driven fault detection methods assume that the process operates in a single mode so that they cannot perform well in processes with multiple operating modes. To monitor multimode processes effectively, this paper proposes a novel process monitoring scheme based on orthogonal nonnegative matrix factorization (ONMF) and hidden Markov model (HMM). The new clustering technique ONMF is employed to separate data from different process modes. The multiple HMMs for various operating modes lead to higher modeling accuracy. The proposed approach does not presume the distribution of data in each mode because the process uncertainty and dynamics can be well interpreted through the hidden Markov estimation. The HMM-based monitoring indication named negative log likelihood probability is utilized for fault detection. In order to assess the proposed monitoring strategy, a numerical example and the Tennessee Eastman process are used. The results demonstrate that this method provides efficient fault detection performance.

A Semantic Deduplication of Temporal Dynamic Records from Multiple Web Databases

Objective: The main objective of this paper is to improve the true positive level of record deduplication using Ontology based MHMM-Fuzzy clustering approach. Methods/Statistical Analysis: Most of the record deduplication system in literature used genetic programming based record deduplication which combined different pieces of evidence extracted from the data content. However the accuracy of the system is low. To overcome this problem we propose a Multiple Hidden Markov Model (MHMM) which is used to increase the accuracy and also to identify joint duplicate records. In this model, if the database has multiple columns, it performs the deduplication for the all columns which will degrade the performance of the system. So to solve this problem, MHMM-Fuzzy Clustering based record deduplication is introduced. In this system Fuzzy clustering is performed through multiple observations from the Hidden Markov Model. Then duplicate data are grouped into one cluster according to their fuzzy logic and it can be eliminated easily. However the true positive level of the system is low. To improve the true positive level Fuzzy Ontology based semantic similarity is incorporated in MHMM-Fuzzy Clustering approach. This implies the improvement of the true positive level of the model. Thus it increases the efficiency of deduplication function that identifies the records of replica and duplications. Findings: Multiple Hidden Markov Model (MHMM) based record deduplication, MHMM-Fuzzy clustering based record deduplication and Ontology based MHMM-Fuzzy clustering approach are applied on Cora Bibliographic dataset and Restaurants dataset. The performance measures are evaluated in terms of precision, recall, f-measure, Execution time and accuracy results. Applications/Improvements: Thus the current research achieves improved result on record deduplication is better than previous works in terms of precision, recall, f-measure, Execution time and accuracy results.

Multiple hidden Markov models for categorical time series

We introduce multiple hidden Markov models (MHMMs) where a multivariate categorical time series depends on a latent multivariate Markov chain. MHMMs provide an elegant framework for specifying various independence relationships between multiple discrete time processes. These independencies are interpreted as Markov properties of a mixed graph and a chain graph associated respectively to the latent and observation components of the MHMM. These Markov properties are also translated into zero restrictions on the parameters of marginal models for the transition probabilities and the distributions of observable variables given the latent states.

Characterizing and Differentiating Brain State Dynamics via Hidden Markov Models.

Functional connectivity measured from resting state fMRI (R-fMRI) data has been widely used to examine the brain's functional activities and has been recently used to characterize and differentiate brain conditions. However, the dynamical transition patterns of the brain's functional states have been less explored. In this work, we propose a novel computational framework to quantitatively characterize the brain state dynamics via hidden Markov models (HMMs) learned from the observations of temporally dynamic functional connectomics, denoted as functional connectome states. The framework has been applied to the R-fMRI dataset including 44 post-traumatic stress disorder (PTSD) patients and 51 normal control (NC) subjects. Experimental results show that both PTSD and NC brains were undergoing remarkable changes in resting state and mainly transiting amongst a few brain states. Interestingly, further prediction with the best-matched HMM demonstrates that PTSD would enter into, but could not disengage from, a negative mood state. Importantly, 84% of PTSD patients and 86% of NC subjects are successfully classified via multiple HMMs using majority voting.

Hidden Markov models for malware classification

Previous research has shown that hidden Markov model (HMM) analysis is useful for detecting certain challenging classes of malware. In this research, we consider the related problem of malware classification based on HMMs. We train multiple HMMs on a variety of compilers and malware generators. More than 8,000 malware samples are then scored against these models and separated into clusters based on the resulting scores. We observe that the clustering results could be used to classify the malware samples into their appropriate families with good accuracy. Since none of the malware families in the test set were used to generate the HMMs, these results indicate that our approach can effective classify previously unknown malware, at least in some cases. Thus, such a clustering strategy could serve as a useful tool in malware analysis and classification.

Joint class identification and target classification using multiple HMMs

Target classification has received significant attention in the tracking literature. Algorithms for joint tracking and classification that are capable of improving tracking performance by exploiting the interdependency between target class and target kinematic behavior have already been proposed. In these works, target identification relies on the a priori information about target classes, but, in practice, the prior class information may not always be available or not accurate. This motivates the design of a new estimation method that can jointly build target classes and classify targets even when a priori information is not available. Based on the generic expectation-maximization framework, a novel joint multitarget class estimation and target identification algorithm that requires only target feature measurements is proposed in this paper to achieve this goal. In this approach, multitarget classes are characterized by multiple hidden Markov models. Besides theoretical derivations, simulations are presented to verify the effectiveness of the proposed algorithm.

Compensation of SNR and noise type mismatch using an environmental sniffing based speech recognition solution

Multiple-model based speech recognition (MMSR) has been shown to be quite successful in noisy speech recognition. Since it employs multiple hidden Markov model (HMM) sets that correspond to various noise types and signal-to-noise ratio (SNR) values, the selected acoustic model can be closely matched with the test noisy speech, which leads to improved performance when compared with other state-of-the-art speech recognition systems that employ a single HMM set. However, as the number of HMM sets is usually limited due to practical considerations as well as effective model selection, acoustic mismatch can still be a problem in MMSR. In this study, we proposed methods to improve recognition performance by mitigating the mismatch in SNR and noise type for an MMSR solution. For the SNR mismatch, an optimal SNR mapping between the test noisy speech and the HMM was determined by experimental investigation. Improved performance was demonstrated by employing the SNR mapping instead of using the estimated SNR of the test noisy speech directly. We also proposed a novel method to reduce the effect of noise type mismatch by compensating the test noisy speech in the log-spectrum domain. We first derive the relation between the log-spectrum vectors in the test and training noisy speech. Since the relation is a non-linear function of the speech and noise parameters, the statistical information regarding the testing log-spectrum vectors was obtained by approximation using vector Taylor series (VTS) algorithm. Finally, the minimum mean square error estimation of the training log-spectrum vectors was used to reduce the mismatch between the training and test noisy speech. By employing the proposed methods in the MMSR framework, relative word error rate reduction of 18.7% and 21.3% was achieved on the Aurora 2 task when compared to a conventional MMSR and multi-condition training (MTR) method, respectively.

Effective technique for the recognition of offline Arabic handwritten words using hidden Markov models

In this paper, we present a novel segmentation-free Arabic handwriting recognition system based on hidden Markov model (HMM). Two main contributions are introduced: a new technique for dividing the image into nonuniform horizontal segments to extract the features and a new technique for solving the problems of the skewing of characters by fusing multiple HMMs. Moreover, two enhancements are introduced: the pre-processing method and feature extraction using concavity space. The proposed system first pre-processes the input image by setting the thickness of the input word to three pixels and fixing the spacing between the different parts of the word. The input image is divided into constant number of nonuniform horizontal segments depending on the distribution of the foreground pixels. A set of robust features representing the gradient of the foreground pixels is extracted using sliding windows. The input image is decomposed into several images representing the vertical, horizontal, left diagonal and right diagonal edges in the image. A set of robust features representing the densities of the foreground pixels in the various edge images is extracted using sliding windows. The proposed system builds character HMM models and learns word HMM models using embedded training. Besides the vertical sliding window, two slanted sliding windows are used to extract the features. Three different HMMs are used: one for the vertical sliding window and two for the slanted windows. A fusion scheme is used to combine the three HMMs. The proposed system is very promising and outperforms all the other Arabic handwriting recognition systems reported in the literature.