On-line adaptive learning of the continuous density hidden Markov model based on approximate recursive Bayes estimate

Huo, Qiang; Lee, Chin‐Hui

doi:10.1109/89.554778

Cited by 92 publications

(1 citation statement)

References 37 publications

(50 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the field of continuous-speech recognition, authors Lee, Rabiner, Pieraccini, and Wilpon (Lee, Rabiner, Pieraccini, & Wilpon, 1990) proposed Bayesian adaptive learning for estimating mean and variance of continuous density HMM. Authors Huo and Lee (Huo & Lee, 1997) proposed a framework of quasi-Bayes (QB) algorithm based on approximate recursive Bayes estimate for learning HMM parameters with Gaussian mixture model; they described that "The QB algorithm is designed to incrementally update the hyperparameters of the approximate posterior distribution and the continuous density HMM parameters simultaneously" (Huo & Lee, 1997, p. 161). Authors Cheng, Sha, and Saul (Cheng, Sha, & Saul, 2009) used the approach of large margin training to learn HMM parameters.…”

Section: Iv5 Continuous Observation Hidden Markov Modelmentioning

confidence: 99%

Mathematical Approaches to User Modeling

Nguyen¹

2019

Preprint

View full text Add to dashboard Cite

The first edition of the book “Mathematical Approaches to User Modeling” is developed from the PhD dissertation “A User Modeling for Adaptive Learning”. It was accepted on 4th January 2015 by Scientific Research Publishing (SCIRP) and finished on 13rd July 2016 but it is not published yet. Following is the abstract of the book.User model is description of user's information and characteristics in abstract level. User model is very important to adaptive software which aims to support user as much as possible. The process to construct user model is called user modeling. As the title suggests, the book focuses on mathematical approaches to user modeling. The book includes seven main chapters. Chapter I is a survey of user model, user modeling, and adaptive learning. Chapter II introduces the general architecture of the proposed user modeling system Zebra and Triangular Learner Model (TLM). Chapter III, IV, V describes three sub-models of TLM such as knowledge sub-model, learning style sub-model, and learning history sub-model in full of mathematical formulas and fundamental methods. These are the most important chapters. Chapter VI gives some approaches to evaluate TLM and Zebra. Chapter VII summarizes the research and discusses future trend of Zebra.The whole book is available at http://mum.locnguyen.net. In general, the first difference of the book is to introduce my innovative works in mathematics, probability, user modeling, machine learning, data mining, and adaptive learning. The second difference is that the user modeling system Zebra is implemented as computer software associated to the book, available at http://zebra.locnguyen.net. Moreover, examples, mathematical formulas, and mathematical models inside the book are implemented separately and integrated into another computer software.

show abstract

Section: Iv5 Continuous Observation Hidden Markov Modelmentioning

confidence: 99%

Mathematical Approaches to User Modeling

Nguyen¹

2019

Preprint

View full text Add to dashboard Cite

show abstract

Robust Recognition Methods

Haton¹

2009

Spoken Language Processing

View full text Add to dashboard Cite

Robust nonlinear adaptation algorithms for multitask prediction networks

Abuduweili

Liu

2020

Adaptive Control & Signal

View full text Add to dashboard Cite

Summary High fidelity behavior prediction of intelligent agents is critical in many applications, which is challenging due to the stochasticity, heterogeneity, and time‐varying nature of agent behaviors. Prediction models that work for one individual may not be applicable to another. Besides, the prediction model trained on the training set may not generalize to the testing set. These challenges motivate the adoption of online adaptation algorithms to update prediction models in real‐time to improve the prediction performance. This article considers online adaptable multitask prediction for both intention and trajectory. The goal of online adaptation is to improve the performance of both intention and trajectory predictions with only the feedback of the observed trajectory. We first introduce a generic τ‐step adaptation algorithm of the multitask prediction model that updates the model parameters with the trajectory prediction error in recent τ steps. Inspired by extended Kalman filter (EKF), a base adaptation algorithm modified EKF with forgetting factor (MEKFλ) is introduced. In order to improve the performance of MEKFλ, generalized exponential moving average filtering techniques are adopted. Then this article introduces a dynamic multiepoch update strategy to effectively utilize samples received in real time. With all these extensions, we propose a robust online adaptation algorithm: MEKF with moving average and dynamic multiepoch strategy (MEKFMA − ME). We empirically study the best set of parameters to adapt in the multitask prediction model and demonstrate the effectiveness of the proposed adaptation algorithms to reduce the prediction error.

show abstract

On-line adaptive learning of the continuous density hidden Markov model based on approximate recursive Bayes estimate

Cited by 92 publications

References 37 publications

Mathematical Approaches to User Modeling

Mathematical Approaches to User Modeling

Robust Recognition Methods

Robust nonlinear adaptation algorithms for multitask prediction networks

Contact Info

Product

Resources

About