Hidden Markov models with first-order equalization for noisy speechrecognition

A particularly effective distortion measure that takes into account the norm shrinkage bias in the noisy cepstrum is considered. A first-order equalization mechanism, specifically aiming at avoiding the norm shrinkage problem, is incorporated in a hidden Markov model (HMM) framework to model the speech cepstral sequence. Such a modeling technique requires special care, as the formulation inevitably involves parameter estimation from a set of data with singular dispersion. Solutions to this HMM stochastic modeling problem are provided, and algorithms for estimating the necessary model parameters are given. It is experimentally shown that incorporation of the first-order mean equalization model makes the HMM-based speech recognizer robust to noise. With respect to a conventional HMM recognizer, this leads to an improvement in recognition performance which is equivalent to a gain of about 15-20 dB in signal-to-noise ratio     

Pattern recognition using a family of design algorithms based uponthe generalized probabilistic descent method

This paper provides a comprehensive introduction to a novel approach to pattern recognition which is based on the generalized probabilistic descent method (GPD) and its related design algorithms. The paper contains a survey of recent recognizer design techniques, the formulation of GPD, the concept of minimum classification error learning that is closely related to the GPD formalization, a relational analysis between GPD and other important design methods, and various embodiments of GPD-based design, including segmental-GPD, minimum spotting error training, discriminative utterance verification, and discriminative feature extraction. GPD development has its origins in basic pattern recognition and Bayes decision theory. It represents a simple but careful re-investigation of the classical theory and successfully leads to an innovative framework. For clarity of presentation, detailed discussions about its embodiments are provided for examples of speech pattern recognition tasks that use a distance-based classifier. Experimental results in speech pattern recognition tasks clearly demonstrate the remarkable utility of the family of GPD-based design algorithms     

Pattern recognition using discriminative feature extraction

We propose a new design method, called discriminative feature extraction for practical modular pattern recognizers. A key concept of discriminative feature extraction is the design of an overall recognizer in a manner consistent with recognition error minimization. The utility of the method is demonstrated in a Japanese vowel recognition task     

Speech, audio, and acoustic processing for multimedia

The primary advances in speech and audio signal processing that contributed to the maturing of multimedia applications are discussed in the areas of speech and audio signal compression, speech synthesis, acoustic processing and echo control, and network echo cancellation     

Equilibrium and kinetic studies of the extraction of chelated copper(II) anions with Aliquat 336

The equilibrium and kinetics of solvent extraction of Cu²⁺ from aqueous solutions containing equimolar EDTA with Aliquat 336 in n‐decanol and kerosene at 298 K were investigated. The concentrations of Cu²⁺ (8–50 mol m^?3), Cl^? (5–60 mol m^?3), and Aliquat 336 (20–100 mol m^?3) were varied. A semi‐empirical model with three parameters was proposed to describe the equilibrium behavior, in which the non‐idealities in both aqueous and organic phases were considered. Over the ranges studied, the model agreed reasonably well with the experimental data (standard deviation, 15%). The forward and backward reaction rate constants were determined as (5.31 ± 0.16)×10^?6 m^9/4 mol^?3/4 s^?1 and (2.62 ± 0.09)×10^?7 s^?1, respectively, at 298 K. An interfacial reaction mechanism was proposed, which revealed that the reaction between the chelated anions and trimeric amine molecules at the interface was rate limiting. The derived rate laws were consistent with the experimental results. © 2002 Society of Chemical Industry     

Characterization and comparison of diblock and triblock amphiphilic copolymers of poly(δ‐valerolactone)

Three types of pegylated amphiphilic copolymers of poly(δ‐valerolactone) (PVL) were copolymerized with methoxy poly(ethylene glycol) (MePEG) and poly(ethylene glycol) (PEG₄₀₀₀ and PEG_10,000), respectively. Pegylation of PVL allowed copolymers possessing amphiphilic property and efficiently self‐assembled to form micelles with a low critical micelle concentration (CMC) in the range of 10^?7–10^?8M. The average molecular weight of copolymers was in the range of 10,000–20,000 Da, and the polydispersity of copolymers was about 1.7–1.8. Higher mobility of low molecular weight PEG (i.e., MePEG and PEG₄₀₀₀) than high molecular weight PEG_10,000 allowed valerolactone ring opening more efficient in terms of PVL/MePEG and PVL/PEG₄₀₀₀ copolymers possessing longer chain length in hydrophobic domain. Pegylated PVL with low CMC and triblock structure was preferred to encapsulate drug during micelle formation. Although all of these amphiphilic copolymers exhibited controlled release character, the micelles formed by triblock copolymer possessed a more stable core‐shell conformation than that by diblock copolymer, and resulted in the release of drug from triblock micelles slower than that from diblock micelles. © 2006 Wiley Periodicals, Inc. J Appl Polym Sci 100: 1836–1841, 2006     

Influence of operating parameters on photocatalytic degradation of phenol in UV/TiO2 process

The use of hydrogen peroxide (H₂O₂) for improved photocatalytic degradation of phenol in aqueous suspension of commercial TiO₂ powders (Degussa P-25) was investigated. Photodegradation was compared using direct photolysis (UV alone), H₂O₂/UV, TiO₂/UV, and H₂O₂/TiO₂/UV processes in a batch reactor with high-pressure mercury lamp irradiation. The effects of operating parameters such as catalyst dosage, light intensity, pH of the solution, the initial phenol, and H₂O₂ concentrations on photodegradation process were examined. It was shown that photodegradation using H₂O₂/TiO₂/UV process was much more effective than using either H₂O₂/UV or TiO₂/UV process. The effect of the initial phenol concentration on TOC removal was also studied, demonstrating that more than 8 h was required to completely mineralize phenol into water and carbon dioxide. For all the four oxidation processes studied, photodegradation followed the first-order kinetics. The apparent rate constants with 400-W UV ranged from 5.0 × 10⁻⁴ min⁻¹ by direct photolysis to 1.4 × 10⁻² min⁻¹ using H₂O₂/TiO₂/UV process. The role of H₂O₂ on such enhanced photodegradation of phenol in aqueous solution was finally discussed.     

Optimization of site investigation program for improved statistical characterization of geotechnical property based on random field theory

This paper presents a framework for optimization of site investigation program, within which the robustness of the site investigation program and the investigation effort are optimized. A site investigation program is judged robust if the derived statistics of the geotechnical property of interest are robust against the uncertainties caused by limited data availability and test error. In this study, a Markov chain Monte Carlo simulation-based Bayesian inference approach was used to characterize the statistics of the intended geotechnical property. The robustness of the site investigation program was formulated as a byproduct of the Bayesian inference of the geotechnical property statistics. The proposed framework for optimization of the site investigation program was implemented as a bi-objective optimization problem that considers both robustness and investigation effort. The concepts of Pareto Front and knee point were employed to aid in making an informed decision regarding selection of site investigation program. The effectiveness and significance of the proposed framework were demonstrated through a simulation study.     

Human Body Posture Classification by a Neural Fuzzy Network and Home Care System Application

A new classification approach for human body postures based on a neural fuzzy network is proposed in this paper, and the approach is applied to detect emergencies that are caused by accidental falls. Four main body postures are used for posture classification, including standing, bending, sitting, and lying. After the human body is segmented from the background, the classification features are extracted from the silhouette. The body silhouette is projected onto horizontal and vertical axes, and then, a discrete Fourier transform is applied to each projected histogram. Magnitudes of significant Fourier transform coefficients together with the silhouette length-width ratio are used as features. The classifier is designed by a neural fuzzy network. The four postures can be classified with high accuracy according to experimental results. Classification results are also applicable to home care emergency detection of a person who suddenly falls and remains in the lying posture for a period of time due to experiments that were performed.