Hierarchical Hidden Markov Model Research Articles

Health-State Estimation and Prognostics in Machining Processes

Failure mechanisms of electromechanical systems usually involve several degraded health-states. Tracking and forecasting the evolution of health-states and impending failures, in the form of remaining-useful-life (RUL), is a critical challenge and regarded as the Achilles' heel of condition-based-maintenance (CBM). This paper demonstrates how this difficult problem can be addressed through Hidden Markov models (HMMs) that are able to estimate unobservable health-states using observable sensor signals. In particular, implementation of HMM based models as dynamic Bayesian networks (DBNs) facilitates compact representation as well as additional flexibility with regard to model structure. Both regular HMM pools and hierarchical HMMs are employed here to estimate online the health-state of drill-bits as they deteriorate with use on a CNC drilling machine. Hierarchical HMM is composed of sub-HMMs in a pyramid structure, providing functionality beyond an HMM for modeling complex systems. In the case of regular HMMs, each HMM within the pool competes to represent a distinct health-state and adapts through competitive learning. In the case of hierarchical HMMs, health-states are represented as distinct nodes at the top of the hierarchy. Monte Carlo simulation, with state transition probabilities derived from a hierarchical HMM, is employed for RUL estimation. Detailed results on health-state and RUL estimation are very promising and are reported in this paper. Hierarchical HMMs seem to be particularly effective and efficient and outperform other HMM methods from literature.

A Ubiquitous Power Management System to Balance Energy Savings and Response Time Based on Device- level Usage Prediction

Power conservation has become a serious concern during people's daily life. Ubiquitous computing technologies clearly provide a potential way to help us realize a more environment-friendly lifestyle. In this paper, we propose a ubiquitous power management system called Gynapse, which uses multi-modal sensors to predict the exact usage of each device, and then switches their power modes based on predicted usage to maximize the total energy saving under the constraint of user required response time. We build a three-level Hierarchical Hidden Markov Model (HHMM) to represent and learn the device level usage patterns from multi-modal sensors. Based on the learned HHMM, we develop our predictive mechanism in Dynamic Bayesian Network (DBN) scheme to precisely predict the usage of each device, with user required response time under consideration. Based on the predicted usage, we follow a four-step process to balance the total energy saving and response time of devices by switching their power modes accordingly. Preliminary results demonstrate that Gynapse has the capability to reduce power consumption while keeping the response time within user's requirement, and provides a complementary approach to previous power management systems.

Bi-dimension decomposed hidden Markov models for multi-person activity recognition

We present a novel model for recognizing long-term complex activities involving multiple persons. The proposed model, named ‘decomposed hidden Markov model’ (DHMM), combines spatial decomposition and hierarchical abstraction to capture multi-modal, long-term dependent and multi-scale characteristics of activities. Decomposition in space and time offers conceptual advantages of compaction and clarity, and greatly reduces the size of state space as well as the number of parameters. DHMMs are efficient even when the number of persons is variable. We also introduce an efficient approximation algorithm for inference and parameter estimation. Experiments on multi-person activities and multi-modal individual activities demonstrate that DHMMs are more efficient and reliable than familiar models, such as coupled HMMs, hierarchical HMMs, and multi-observation HMMs.

Hierarchical hidden Markov model with application to joint analysis of ChIP-chip and ChIP-seq data

Chromatin immunoprecipitation (ChIP) experiments followed by array hybridization, or ChIP-chip, is a powerful approach for identifying transcription factor binding sites (TFBS) and has been widely used. Recently, massively parallel sequencing coupled with ChIP experiments (ChIP-seq) has been increasingly used as an alternative to ChIP-chip, offering cost-effective genome-wide coverage and resolution up to a single base pair. For many well-studied TFs, both ChIP-seq and ChIP-chip experiments have been applied and their data are publicly available. Previous analyses have revealed substantial technology-specific binding signals despite strong correlation between the two sets of results. Therefore, it is of interest to see whether the two data sources can be combined to enhance the detection of TFBS. In this work, hierarchical hidden Markov model (HHMM) is proposed for combining data from ChIP-seq and ChIP-chip. In HHMM, inference results from individual HMMs in ChIP-seq and ChIP-chip experiments are summarized by a higher level HMM. Simulation studies show the advantage of HHMM when data from both technologies co-exist. Analysis of two well-studied TFs, NRSF and CCCTC-binding factor (CTCF), also suggests that HHMM yields improved TFBS identification in comparison to analyses using individual data sources or a simple merger of the two. Source code for the software ChIPmeta is freely available for download at http://www.umich.edu/~hwchoi/HHMMsoftware.zip, implemented in C and supported on linux.

HHMMiR: efficient de novo prediction of microRNAs using hierarchical hidden Markov models

BackgroundMicroRNAs (miRNAs) are small non-coding single-stranded RNAs (20–23 nts) that are known to act as post-transcriptional and translational regulators of gene expression. Although, they were initially overlooked, their role in many important biological processes, such as development, cell differentiation, and cancer has been established in recent times. In spite of their biological significance, the identification of miRNA genes in newly sequenced organisms is still based, to a large degree, on extensive use of evolutionary conservation, which is not always available.ResultsWe have developed HHMMiR, a novel approach for de novo miRNA hairpin prediction in the absence of evolutionary conservation. Our method implements a Hierarchical Hidden Markov Model (HHMM) that utilizes region-based structural as well as sequence information of miRNA precursors. We first established a template for the structure of a typical miRNA hairpin by summarizing data from publicly available databases. We then used this template to develop the HHMM topology.ConclusionOur algorithm achieved average sensitivity of 84% and specificity of 88%, on 10-fold cross-validation of human miRNA precursor data. We also show that this model, trained on human sequences, works well on hairpins from other vertebrate as well as invertebrate species. Furthermore, the human trained model was able to correctly classify ~97% of plant miRNA precursors. The success of this approach in such a diverse set of species indicates that sequence conservation is not necessary for miRNA prediction. This may lead to efficient prediction of miRNA genes in virtually any organism.

Incorporation of acoustic level analysis with sonar operator workstation workflow using hierarchical hidden Markov models

Future Anti Submarine Warfare (ASW) systems will be highly automated, and utilize multiple sonar signal processing, non‐acoustic information, off platform information, and environmental information to maximize the detection, tracking, and classification of undersea threats. Recently, a hidden Markov model (HMM) was developed to describe and characterize the workflow strategies that expert sonar operators used during the IUSW‐21 (Integrated UnderSea Warfare for the 21st Century) Sea Test. Abandoning the normative workflow designed into the system, the sonar operators developed and utilized unanticipated strategies to gather evidence to confirm or disconfirm automated classifications or to manually classify contacts. The workflow included detailed visual inspection of and listening to the sonar return. This initial HMM did not include a characterization of the auditory processing. Hidden Markov models are well suited to this purpose. Adding this component to the operator workflow was achieved by a hierarchical HMM which represents the acoustic processing as an embedded HMM in the overall workflow HMM. The resulting model is not just an operator workflow but defines a system processing model that could be used as a framework to develop automated processing of selected components.

A daily behavior enabled hidden Markov model for human behavior understanding

This paper presents a Hierarchical Context Hidden Markov Model (HC-HMM) for behavior understanding from video streams in a nursing center. The proposed HC-HMM infers elderly behaviors through three...

MAppleT: simulation of apple tree development using mixed stochastic and biomechanical models.

Construction of tree architectural databases over years is time consuming and cannot easily capture event dynamics, especially when both tree topology and geometry are considered. The present project aimed to bring together models of topology and geometry in a single simulation such that the architecture of an apple tree may emerge from process interactions. This integration was performed using L-systems. A mixed approach was developed based on stochastic models to simulate plant topology and mechanistic model for the geometry. The succession of growth units (GUs) along axes and their branching structure were jointly modelled by a hierarchical hidden Markov model. A biomechanical model, derived from previous studies, was used to calculate stem form at the metamer scale, taking into account the intra-year dynamics of primary, secondary and fruit growth. Outputs consist of 3-D mock-ups - geometric models representing the progression of tree form over time. To asses these models, a sensitivity analysis was performed and descriptors were compared between simulated and digitised trees, including the total number of GUs in the entire tree, descriptors of shoot geometry (basal diameter, length), and descriptors of axis geometry (inclination, curvature). In conclusion, despite some limitations, MAppleT constitutes a useful tool for simulating development of apple trees in interaction with gravity.

An intrusion detection method based on hierarchical hidden Markov models

This paper presents an anomaly detection approach to detect intrusions into computer systems. In this approach, a hierarchical hidden Markov model (HHMM) is used to represent a temporal profile of normal behavior in a computer system. The HHMM of the norm profile is learned from historic data of the system’s normal behavior. The observed behavior of the system is analyzed to infer the probability that the HHMM of the norm profile supports the observed behavior. A low probability of support indicates an anomalous behavior that may result from intrusive activities. The model was implemented and tested on the UNIX system call sequences collected by the University of New Mexico group. The testing results showed that the model can clearly identify the anomaly activities and has a better performance than hidden Markov model.

HHMM Based Recognition of Human Activity

In this paper, we present a method for recognition of human activity as a series of actions from an image sequence. The difficulty with the problem is that there is a chicken-egg dilemma that each action needs to be extracted in advance for its recognition but the precise extraction is only possible after the action is correctly identified. In order to solve this dilemma, we use as many models as actions of our interest, and test each model against a given sequence to find a matched model for each action occurring in the sequence. For each action, a model is designed so as to represent any activity containing the action. The hierarchical hidden Markov model (HHMM) is employed to represent the models, in which each model is composed of a submodel of the target action and submodels which can represent any action, and they are connected appropriately. Several experimental results are shown.

Simplified training algorithm for hierarchical hidden Markov models

AbstractThis paper considers an extension of the hidden Markov model, called the hierarchical hidden Markov model, and proposes an EM algorithm and an approximate algorithm. The EM algorithm proposed in this paper differs from the existing training algorithm, which is called the Baum–Welch algorithm, and guarantees that the likelihood is always increased by updating the parameters. The approximate algorithm has the advantage that it can be applied to problems in which observation of the training sentence and training of the parameters proceed in parallel. These algorithms and their derivations are simplified, compared to the existing training algorithm, by using stochastic context‐free grammar. © 2004 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 87(5): 59–69, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10172

Syntactic user modelling with stochastic processes

An approach to user modelling with discrete stochastic processes is presented, which aims at the dynamic individualization of user interfaces on the syntactic layer. The state of the art of syntactic user modelling is surveyed. The mathematical background of simple Markov Chains and the 'classic' Hidden Markov Model is presented. Furthermore, dynamic Bayesian Networks are introduced, which generalize these Markovian Models. A case study of the simulation experiments uses a multimodal user interface for supervisory control of advanced manufacturing cells. A corresponding simulation model is created and exploited to generate interaction cases, which are the empirical basis for the evaluation. Six topologies of dynamic Bayesian Networks are evaluated for 100 interaction cases and 50 replications each: (1) Markov Chain of order 1, (2) Hidden Markov Model, (3) autoregressive Hidden Markov Model, (4) factorial Hidden Markov Model, (5) simple hierarchical Hidden Markov Model, and (6) tree structured Hidden Markov Model. The dependent variable is the prediction accuracy for a single prediction lead. In a first step, a one-way analysis of variance in conjunction with Tukey's post-hoc test demonstrated a significant superiority of the simple hierarchical Hidden Markov Model. In a second step, an additional two-way analysis of variance also indicated a significantly better prediction accuracy of the simple hierarchical Hidden Markov Model compared to the Hidden Markov Model, but the number of interaction cases also had a significant effect. Hence, the modeller has to take both factors--model topology and number of interaction cases--into account when designing syntactic user models with stochastic processes.

The Hierarchical Hidden Markov Model: Analysis and Applications

We introduce, analyze and demonstrate a recursive hierarchical generalization of the widely used hidden Markov models, which we name Hierarchical Hidden Markov Models (HHMM). Our model is motivated by the complex multi-scale structure which appears in many natural sequences, particularly in language, handwriting and speech. We seek a systematic unsupervised approach to the modeling of such structures. By extending the standard Baum-Welch (forward-backward) algorithm, we derive an efficient procedure for estimating the model parameters from unlabeled data. We then use the trained model for automatic hierarchical parsing of observation sequences. We describe two applications of our model and its parameter estimation procedure. In the first application we show how to construct hierarchical models of natural English text. In these models different levels of the hierarchy correspond to structures on different length scales in the text. In the second application we demonstrate how HHMMs can be used to automatically identify repeated strokes that represent combination of letters in cursive handwriting.