Hidden Markov Model Algorithm Research Articles

New decoding algorithms for Hidden Markov Models using distance measures on labellings

BackgroundExisting hidden Markov model decoding algorithms do not focus on approximately identifying the sequence feature boundaries.ResultsWe give a set of algorithms to compute the conditional probability of all labellings "near" a reference labelling λ for a sequence y for a variety of definitions of "near". In addition, we give optimization algorithms to find the best labelling for a sequence in the robust sense of having all of its feature boundaries nearly correct. Natural problems in this domain are NP-hard to optimize. For membrane proteins, our algorithms find the approximate topology of such proteins with comparable success to existing programs, while being substantially more accurate in estimating the positions of transmembrane helix boundaries.ConclusionMore robust HMM decoding may allow for better analysis of sequence features, in reasonable runtimes.

Disease surveillance using a hidden Markov model

BackgroundRoutine surveillance of disease notification data can enable the early detection of localised disease outbreaks. Although hidden Markov models (HMMs) have been recognised as an appropriate method to model disease surveillance data, they have been rarely applied in public health practice. We aimed to develop and evaluate a simple flexible HMM for disease surveillance which is suitable for use with sparse small area count data and requires little baseline data.MethodsA Bayesian HMM was designed to monitor routinely collected notifiable disease data that are aggregated by residential postcode. Semi-synthetic data were used to evaluate the algorithm and compare outbreak detection performance with the established Early Aberration Reporting System (EARS) algorithms and a negative binomial cusum.ResultsAlgorithm performance varied according to the desired false alarm rate for surveillance. At false alarm rates around 0.05, the cusum-based algorithms provided the best overall outbreak detection performance, having similar sensitivity to the HMMs and a shorter average time to detection. At false alarm rates around 0.01, the HMM algorithms provided the best overall outbreak detection performance, having higher sensitivity than the cusum-based Methods and a generally shorter time to detection for larger outbreaks. Overall, the 14-day HMM had a significantly greater area under the receiver operator characteristic curve than the EARS C3 and 7-day negative binomial cusum algorithms.ConclusionOur findings suggest that the HMM provides an effective method for the surveillance of sparse small area notifiable disease data at low false alarm rates. Further investigations are required to evaluation algorithm performance across other diseases and surveillance contexts.

A TUTORIAL OF TECHNIQUES FOR IMPROVING STANDARD HIDDEN MARKOV MODEL ALGORITHMS

In this tutorial, we discuss two main algorithms for Hidden Markov Models or HMMs: the Viterbi algorithm and the expectation phase of the Baum-Welch algorithm, and we describe ways to improve their naïve implementations. For the Baum-Welch algorithm we first present an implementation of the expectation computations using constant space. We then discuss the classical implementation of this calculation and describe ways to reduce its space usage to logarithmic and O(square root n), with their respective CPU costs. We also note where each respective algorithm can be parallelized. For the Viterbi algorithm, we describe O(square root n) and logarithmic space algorithms which increase CPU use by a factor of two and by a logarithmic factor respectively. We also present two recent heuristics for decreasing space use, which in practice lead to logarithmic space use. Classical version of Viterbi cannot be parallelized by splitting sequence in several subsequences, but we show a parallelization that works if we are willing to pay a significant extra CPU cost. Finally we show a very simple parallelization trick which enables full usage of multiple CPUs/cores under the condition that they share memory.

Extracting kinetics information from single-molecule fluorescence resonance energy transfer data using hidden markov models.

Hidden Markov models (HMM) have been proposed as a method of analysis for noisy single-molecule fluorescence resonance energy transfer (SM FRET) data. However, there are practical and fundamental limits in applying HMM to SM FRET data due to the short photobleaching lifetimes of fluorophores and the limited time resolution of detection devices. The fast photobleaching fluorophores yield short SM FRET time traces, and the limited detection time resolution generates abnormal FRET values, which result in systematic underestimation of kinetic rates. In this work, a HMM algorithm is implemented to optimize one set of HMM parameters with multiple short SM FRET traces. The FRET efficiency distribution function for the HMM optimization was modified to accommodate the abnormal FRET values resulting from limited detection time resolution. Computer simulations reveal that one set of HMM parameters is optimized successfully using multiple short SM FRET traces and that the degree of the kinetic rate underestimation is reduced by using the proposed modified FRET efficiency distribution. In conclusion, it is demonstrated that HMM can be used to reproducibly analyze short SM FRET time traces.

Speeding up HMM algorithms for genetic linkage analysis via chain reductions of the state space.

We develop an hidden Markov model (HMM)-based algorithm for computing exact parametric and non-parametric linkage scores in larger pedigrees than was possible before. The algorithm is applicable whenever there are chains of persons in the pedigree with no genetic measurements and with unknown affection status. The algorithm is based on shrinking the state space of the HMM considerably using such chains. In a two g-degree cousins pedigree the reduction drops the state space from being exponential in g to being linear in g. For a Finnish family in which two affected children suffer from a rare cold-inducing sweating syndrome, we were able to reduce the state space by more than five orders of magnitude from 250 to 232. In another pedigree of state-space size of 227, used for a study of pituitary adenoma, the state space reduced by a factor of 8.5 and consequently exact linkage scores can now be computed, rather than approximated.Contact: dang@cs.technion.ac.ilSupplementary information: Supplementary data are available at Bioinformatics online.

Research on Running Time Behavior Analyzing and Trend Predicting of Modern Distributed Software

Interactive behavior trust of modern distributed software systems (MDSS) should be “monitored” and “grasped” at running time. The paper investigates the relationships between behaviors and their effects at running time in MDSS, uses statistical machine learning tools to analyze the laws of behavior traces, and presents a novel behavior analyzing and trend predicting method. We use hierarchical Dirichlet process and infinite hidden Markov model to converge monitored interface data to determine unknown events, and learn behavior patterns from event sequences including unknown events in terms of semisupervised method. As determining unknown events and behavior patterns, Beam sampling has higher efficiency in sampling and inference compared with other method (e.g., Gibbs sampling). When behavior patterns reach a certain scale, MDSS can analyze and predict interactive behaviors in terms of unsupervised method. We adopt Viterbi algorithm of hidden Markov model to analyze optimal sequences of interactive events, which help to determine good and evil of current behaviors. MDSS can send early warning for hostile behaviors, actively predict subsequent trends for non-hostile behaviors. Simulation experiments testify that the novel method has unique predominance in software behavior analyzing and trend predicting.

Application of Improved HMM Algorithm in Slag Detection System

To solve the problems of ladle slag detection system (SDS), such as high cost, short service life, and inconvenient maintenance, a new SDS realization method based on hidden Markov model (HMM) was put forward. The physical process of continuous casting was analyzed, and vibration signal was considered as the main detecting signal according to the difference in shock vibration generated by molten steel and slag because of their difference in density. Automatic control experiment platform oriented to SDS was established, and vibration sensor was installed far away from molten steel, which could solve the problem of easy power consumption by the sensor. The combination of vector quantization technology with learning process parameters of HMM was optimized, and its revaluation formula was revised to enhance its recognition effectiveness. Industrial field experiments proved that this system requires low cost and little rebuilding for current devices, and its slag detection rate can exceed 95%.

Balanced Truncation for a Class of Stochastic Jump Linear Systems and Model Reduction for Hidden Markov Models

This paper develops a generalization of the balanced truncation algorithm applicable to a class of discrete-time stochastic jump linear systems. The approximation error, which is captured by means of the stochastic L 2 gain, is bounded from above by twice the sum of singular numbers associated to the truncated states, similar to the case of linear time-invariant systems. A two step model reduction algorithm for hidden Markov models is also developed. The first step relies on the aforementioned balanced truncation algorithm due to a topological equivalence established between hidden Markov models and a subclass of stochastic jump linear systems. In a second step the positivity constraints, which reflect the hidden Markov model structure, are enforced by solving a low dimensional optimization problem.

Deletion in Chromosome 17p12 and Gains in Chromosome 9q33.3 by Array Comparative Hybridization Are Associated with R-CHOP Treatment Failure in Patients with Diffuse Large B Cell Lymphoma

Background: The genetic alterations associated with survival in patients with diffuse large B cell lymphomas (DLBCL) treated with combined rituximab-CHOP (R-CHOP) chemo-immunotherapy are not well understood.Methods: 92 patients with DLBCL treated with R-CHOP who had a biopsy available at the time of diagnosis were included in the study. 31 patients were classified as treatment failures defined as progression < 6 months of completing R-CHOP and 61 patients were classified as treatment successes defined as a maintained remission >2 years after diagnosis. We used genome-wide BAC array comparative genomic hybridization (aCGH) to determine the genomic copy number imbalances. The presence of genetic gains and losses were determined using the intersection between visual annotations and a Hidden Markov model algorithm. DLBCL cell of origin (COO) subtype distinctions (GCB vs ABC) were determined using a Bayesian predictor model on gene expression derived from custom Affymetrix arrays (Dave, N Engl J Med , 2006; 354:2431). A permutation test was used to identify genetic regions that were significantly different between treatment failures and treatment successes. Functional pathway analysis was performed using Ingenuity software. A novel model based clustering algorithm was applied to the normalized data to determine if any association with outcome correlated with the observed genetic alterations.Results: Lymphoma progressed in 31/92 (34%) patients < 6 months after R-CHOP (median follow-up = 4 y). All 92 patients had successful aCGH and 81 had COO available for this analysis. The International Prognostic Index (IPI) and COO were predictive of outcome (p=0.04, p=0.02, respectively). 451 regions containing 338 genes were associated with treatment failure with a p-value of <10−6. Gains in 9q33.3 were found in 13 patients (14%) and were significantly associated with treatment failure p<10−8. This region contains genes such as HSPA5, a negative regulator of apoptosis and PPP6C, a positive regulator of the cell cycle by targeting IKBe thereby removing inhibition of the NFkB pathway. Deletions in 17p12 were detected in 24 (26%) and were the most statistically significantly associated with treatment failure p<10−9. This region contains tumor necrosis factor (TNF) receptor superfamily member (TNFRSF13B or TACI) which, when deleted or mutated, has been previously shown to lead to activation of the noncanonical NFkB pathway and B cell proliferation. 21 of these 24 patients also had deletion of 17p13 at the TP53 locus (p<10−6). Neither 9p33.3 nor 17p12 deletion was associated with COO distinctions. Using Ingenuity, pathways involving apoptosis and cellular proliferation, specifically those involving P53, MYC and HSPA5 genes, were over-represented in treatment failures (p=2.04 × 10−4). Unsupervised clustering of the aCGH data demonstrated that 60% of cases could be stratified into 4 genetic sub-groups based on the presence of 1q+, 6q−, +7 and the concurrent presence of +3 and +18. Supervised analysis demonstrated that the +3/+18 group and the 6q− group were associated with ABC subtype of DLBCL whereas the +7 and 1q+ groups were associated with the GCB subtype. However, these genetic groups did not correlate with treatment outcome.Conclusions: Some genetic alterations cluster together and can distinguish COO subtypes of DLBCL. Gains on 9q33.3 and deletions of 17p12 are common alterations detected by high resolution aCGH in DLBCL. Most importantly, these alterations involve genes known to be critical in B cell proliferation and apoptosis and alterations at these sites are strongly associated with treatment failure (p values <10−8) in patients treated with R-CHOP.

Reversible jump and the label switching problem in hidden Markov models

Reversible jump Markov chain Monte Carlo (RJMCMC) algorithms can be efficiently applied in Bayesian inference for hidden Markov models (HMMs), when the number of latent regimes is unknown. As for finite mixture models, when priors are invariant to the relabelling of the regimes, HMMs are unidentifiable in data fitting, because multiple ways to label the regimes can alternate during the MCMC iterations; this is the so-called label switching problem. HMMs with an unknown number of regimes are considered here and the goal of this paper is the comparison, both applied and theoretical, of five methods used for tackling label switching within a RJMCMC algorithm; they are: post-processing, partial reordering, permutation sampling, sampling from a Markov prior and rejection sampling. The five strategies we compare have been proposed mostly in the literature of finite mixture models and only two of them, i.e. rejection sampling and partial reordering, have been presented in RJMCMC algorithms for HMMs. We consider RJMCMC algorithms in which the parameters are updated by Gibbs sampling and the dimension of the model changes in split-and-merge and birth-and-death moves. Finally, an example illustrates and compares the five different methodologies.

Recurrent CNVs Disrupt Three Candidate Genes in Schizophrenia Patients

Schizophrenia is a severe psychiatric disease with complex etiology, affecting approximately 1% of the general population. Most genetics studies so far have focused on disease association with common genetic variation, such as single-nucleotide polymorphisms (SNPs), but it has recently become apparent that large-scale genomic copy-number variants (CNVs) are involved in disease development as well. To assess the role of rare CNVs in schizophrenia, we screened 54 patients with deficit schizophrenia using Affymetrix's GeneChip 250K SNP arrays. We identified 90 CNVs in total, 77 of which have been reported previously in unaffected control cohorts. Among the genes disrupted by the remaining rare CNVs are MYT1L, CTNND2, NRXN1, and ASTN2, genes that play an important role in neuronal functioning but--except for NRXN1--have not been associated with schizophrenia before. We studied the occurrence of CNVs at these four loci in an additional cohort of 752 patients and 706 normal controls from The Netherlands. We identified eight additional CNVs, of which the four that affect coding sequences were found only in the patient cohort. Our study supports a role for rare CNVs in schizophrenia susceptibility and identifies at least three candidate genes for this complex disorder.

Online Learning with Hidden Markov Models

We present an online version of the expectation-maximization (EM) algorithm for hidden Markov models (HMMs). The sufficient statistics required for parameters estimation is computed recursively with time, that is, in an online way instead of using the batch forward-backward procedure. This computational scheme is generalized to the case where the model parameters can change with time by introducing a discount factor into the recurrence relations. The resulting algorithm is equivalent to the batch EM algorithm, for appropriate discount factor and scheduling of parameters update. On the other hand, the online algorithm is able to deal with dynamic environments, i.e., when the statistics of the observed data is changing with time. The implications of the online algorithm for probabilistic modeling in neuroscience are briefly discussed.

Inference from genome‐wide association studies using a novel Markov model

In this paper we propose a Bayesian modeling approach to the analysis of genome-wide association studies based on single nucleotide polymorphism (SNP) data. Our latent seed model combines various aspects of k-means clustering, hidden Markov models (HMMs) and logistic regression into a fully Bayesian model. It is fitted using the Markov chain Monte Carlo stochastic simulation method, with Metropolis-Hastings update steps. The approach is flexible, both in allowing different types of genetic models, and because it can be easily extended while remaining computationally feasible due to the use of fast algorithms for HMMs. It allows for inference primarily on the location of the causal locus and also on other parameters of interest. The latent seed model is used here to analyze three data sets, using both synthetic and real disease phenotypes with real SNP data, and shows promising results. Our method is able to correctly identify the causal locus in examples where single SNP analysis is both successful and unsuccessful at identifying the causal SNP.

은닉 마르코프 모델의 다목적함수 최적화를 통한 자동 독순의 성능 향상

This paper proposes a new multiobjective optimization method for discriminative training of hidden Markov models (HMMs) used as the recognizer for automatic lipreading. While the conventional Baum-Welch algorithm for training HMMs aims at maximizing the probability of the data of a class from the corresponding HMM, we define a new training criterion composed of two minimization objectives and develop a global optimization method of the criterion based on simulated annealing. The result of a speaker-dependent recognition experiment shows that the proposed method improves performance by the relative error reduction rate of about 8% in comparison to the Baum-Welch algorithm.

Hidden Markov models for the complex stimulus-response relationships of multi-state neurons

Event Abstract Back to Event Hidden Markov models for the complex stimulus-response relationships of multi-state neurons Sean Escola1* and Liam Paninski1 1 Columbia University, United States Recent experimental results suggest that neurons are associated with multiple firing regimes, or states. We develop a general framework for estimating state-dependent neural receptive fields from paired spike-train and stimulus data assuming that neurons transition between several discrete hidden states. We modify both the discrete-time and continuous-time traditional hidden Markov model (HMM) frameworks to permit point-process observables such as spike-trains. For maximal flexibility in our model, we allow the external, time-varying stimulus and the neuron's own spike-history to drive both the spiking behavior in each state and the transitioning behavior between states. We employ an appropriately modified expectation-maximization algorithm to learn the model parameters. The expectation step is solved by the standard forward-backward algorithm for HMMs. The maximization step, though it is not analytically tractable, reduces to a simple concave optimization problem if the model is restricted slightly. We test our algorithm on two stimulated data sets and are able both to fully recover the parameters used to generate the data and to accurately recapitulate the sequence of hidden states. As a demonstration of the flexibility of our model, we apply our framework to a recently-published data set of purportedly multi-state neurons and show that we can implement a hybrid half multi-state/half histogram model which captures more of the neuronal variability than either a simple HMM or a simple histogram-based model alone Conference: Bernstein Symposium 2008, Munich, Germany, 8 Oct - 10 Oct, 2008. Presentation Type: Poster Presentation Topic: All Abstracts Citation: Escola S and Paninski L (2008). Hidden Markov models for the complex stimulus-response relationships of multi-state neurons. Front. Comput. Neurosci. Conference Abstract: Bernstein Symposium 2008. doi: 10.3389/conf.neuro.10.2008.01.043 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 13 Nov 2008; Published Online: 13 Nov 2008. * Correspondence: Sean Escola, Columbia University, New York, United States, gse3@columbia.edu Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Sean Escola Liam Paninski Google Sean Escola Liam Paninski Google Scholar Sean Escola Liam Paninski PubMed Sean Escola Liam Paninski Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.

은닉 마르코프 모델의 확률적 최적화를 통한 자동 독순의 성능 향상

본 논문에서는 자동 독순(automatic lipreading)의 인식기로 쓰이는 은닉 마르코프 모델(HMM: hidden Markov model)의 새로운 확률적 최적화 기법을 제안한다. 제안하는 기법은 전역 최적화가 가능한 확률적 기법인 모의 담금질과 지역 최적화 기법을 결합하는 것으로써, 알고리즘의 빠른 수렴과 좋은 해로의 수렴을 가능하게 한다. 제안하는 알고리즘이 전역 최적해로 수렴함을 수학적으로 보인다. 제안하는 기법을 통해 HMM을 학습함으로써 기존의 알고리즘이 지역해만을 찾는 단점을 개선함으로써 향상된 독순 성능을 나타냄을 실험으로 보인다. This paper proposes a new stochastic optimization algorithm for hidden Markov models (HMMs) used as a recognizer of automatic lipreading. The proposed method combines a global stochastic optimization method, the simulated annealing technique, and the local optimization method, which produces fast convergence and good solution quality. We mathematically show that the proposed algorithm converges to the global optimum. Experimental results show that training HMMs by the method yields better lipreading performance compared to the conventional training methods based on local optimization.

Selecting hidden Markov model state number with cross-validated likelihood

The problem of estimating the number of hidden states in a hidden Markov model is considered. Emphasis is placed on cross-validated likelihood criteria. Using cross-validation to assess the number of hidden states allows to circumvent the well-documented technical difficulties of the order identification problem in mixture models. Moreover, in a predictive perspective, it does not require that the sampling distribution belongs to one of the models in competition. However, computing cross-validated likelihood for hidden Markov models for which only one training sample is available, involves difficulties since the data are not independent. Two approaches are proposed to compute cross-validated likelihood for a hidden Markov model. The first one consists of using a deterministic half-sampling procedure, and the second one consists of an adaptation of the EM algorithm for hidden Markov models, to take into account randomly missing values induced by cross-validation. Numerical experiments on both simulated and real data sets compare different versions of cross-validated likelihood criterion and penalised likelihood criteria, including BIC and a penalised marginal likelihood criterion. Those numerical experiments highlight a promising behaviour of the deterministic half-sampling criterion.

The message of the genes [medical signal processing

This paper presents the use of signal processing technique in DNA analysing. All the rules of information theory that apply to sending data down a phone line also apply to analysing DNA sequences. The QuantiSNP model describes how advanced signal processing techniques can help to analyse DNA. It is an innovative combination of objective Bayes and hidden Markov model algorithms.

Batch and Sequential Bayesian Estimators of the Number of Active Terminals in an IEEE 802.11 Network

The performance of the IEEE 802.11 protocol based on the distributed coordination function (DCF) has been shown to be dependent on the number of competing terminals and the backoff parameters. Better performance can be expected if the parameters are adapted to the number of active users. In this paper we develop both off-line and online Bayesian signal processing algorithms to estimate the number of competing terminals. The estimation is based on the observed use of the channel and the number of competing terminals is modeled as a Markov chain with unknown transition matrix. The off-line estimator makes use of the Gibbs sampler whereas the first online estimator is based on the sequential Monte Carlo (SMC) technique. A deterministic variant of the SMC estimator is then developed, which is simpler to implement and offers superior performance. Finally a novel approximate maximum a posteriori (MAP) algorithm for hidden Markov models (HMM) with unknown transition matrix is proposed. Realistic IEEE 802.11 simulations using the ns-2 network simulator are provided to demonstrate the excellent performance of the proposed estimators

Aggregate a posteriori linear regression adaptation

We present a new discriminative linear regression adaptation algorithm for hidden Markov model (HMM) based speech recognition. The cluster-dependent regression matrices are estimated from speaker-specific adaptation data through maximizing the aggregate a posteriori probability, which can be expressed in a form of classification error function adopting the logarithm of posterior distribution as the discriminant function. Accordingly, the aggregate a posteriori linear regression (AAPLR) is developed for discriminative adaptation where the classification errors of adaptation data are minimized. Because the prior distribution of regression matrix is involved, AAPLR is geared with the Bayesian learning capability. We demonstrate that the difference between AAPLR discriminative adaptation and maximum a posteriori linear regression (MAPLR) adaptation is due to the treatment of the evidence. Different from minimum classification error linear regression (MCELR), AAPLR has closed-form solution to fulfil rapid adaptation. Experimental results reveal that AAPLR speaker adaptation does improve speech recognition performance with moderate computational cost compared to maximum likelihood linear regression (MLLR), MAPLR, MCELR and conditional maximum likelihood linear regression (CMLLR). These results are verified for supervised adaptation as well as unsupervised adaptation for different numbers of adaptation data.