Derivative Kernels for Noise Robust ASR

Recently there has been interest in combined gen- erative/discriminative classifiers. In these classifiers features for the discriminative models are derived from generative kernels. One advantage of using generative kernels is that systematic approaches exist how to introduce complex dependencies beyond conditional independence assumptions. Furthermore, by using generative kernels model-based compensation/adaptation tech- niques can be applied to make discriminative models robust to noise/speaker conditions. This paper extends previous work with combined generative/discriminative classifiers in several directions. First, it introduces derivative kernels based on context- dependent generative models. Second, it describes how derivative kernels can be incorporated in continuous discriminative models. Third, it addresses the issues associated with large number of classes and parameters when context-dependent models and high- dimensional features of derivative kernels are used. The approach is evaluated on two noise-corrupted tasks: small vocabulary AURORA 2 and medium-to-large vocabulary AURORA 4 task.

Derivative kernels for noise robust ASR

Recently there has been interest in combining generative and discriminative classifiers. In these classifiers features for the discriminative models are derived from the generative kernels. One advantage of using generative kernels is that systematic approaches exist to introduce complex dependencies into the feature-space. Furthermore, as the features are based on generative models standard model-based compensation and adaptation techniques can be applied to make discriminative models robust to noise and speaker conditions. This paper extends previous work in this framework in several directions. First, it introduces derivative kernels based on context-dependent generative models. Second, it describes how derivative kernels can be incorporated in structured discriminative models. Third, it addresses the issues associated with large number of classes and parameters when context-dependent models and high-dimensional feature-spaces of derivative kernels are used. The approach is evaluated on two noise-corrupted tasks: small vocabulary AURORA 2 and medium-to-large vocabulary AURORA 4 task. © 2011 IEEE.

Photocurrent derivative spectra of ZnCdSe-ZnSe double multi-quantum wells

Photocurrent (PC) spectra of ZnCdSe-ZnSe double multi-quantum wells are measured at different temperature. Its corresponding photocurrent derivative (PCD) spectra are obtained by computing, and the PCD spectra have greatly enhanced the sensitivity of the relative weak PC signals. The polarization dependence of the PC spectra shows that the transitions observed in the PC spectra are heavy-hole related, and the transition energy coincide well with the results obtained by envelope function approximation including strain. The temperature dependence of the photocurrent curves indicates that the thermal activation is the dominant transport mechanism of the carriers in our samples. The concept of saturation temperature region is introduced to explain why the PC spectra have different temperature dependence in the samples with different structure parameters. It is found to be very useful in designing photovoltaic devices.

Leakage current analysis in AlGaInP/GaInP multi-quantum well lasers by the electrical derivative method - art. no. 60202F

In AlGaInP/GaInP multi-quantum well (MQW) lasers, the electron leakage current is a much more serious problem than that in laser diodes with longer wavelength. To further improve the output performance, the leakage current should be analyzed. In this letter, the temperature dependence of electrical derivative characteristics in AlGaInP/GaInP multi-quantum well lasers was measured, and the potential barrier for electron leakage was obtained. With the help of secondary ion mass spectroscopy (SIMS) measurement, theoretical analysis of the potential barrier was presented and compared with the measurement result. The influence of p-cladding doping level and doping profile on the potential barrier was discussed, and this can be helpful in metalorganic chemical vapor deposition (MOCVD) growth.

Photocurrent derivative spectra of ZnCdSe-ZnSe double multi-quantum wells

Photocurrent (PC) spectra of ZnCdSe-ZnSe double multi-quantum wells are measured at different temperature. Its corresponding photocurrent derivative (PCD) spectra are obtained by computing, and the PCD spectra have greatly enhanced the sensitivity of the relative weak PC signals. The polarization dependence of the PC spectra shows that the transitions observed in the PC spectra are heavy-hole related, and the transition energy coincide well with the results obtained by envelope function approximation including strain. The temperature dependence of the photocurrent curves indicates that the thermal activation is the dominant transport mechanism of the carriers in our samples. The concept of saturation temperature region is introduced to explain why the PC spectra have different temperature dependence in the samples with different structure parameters. It is found to be very useful in designing photovoltaic devices.

Capillary electrochromatographic separation of enantiomers on chemically bonded type of cellulose derivative chiral stationary phases with a positively charged spacer

Positively charged chiral stationary phases (CSPs) were prepared for capillary electrochromatography (CEC) separation of enantiomers by chemically immobilizing cellulose derivatives onto diethylenetriaminopropylated silica (DEAPS) with tolylene-2,4-diisocyanate (TDI) as a spacer reagent. Anodic electroosmotic mobility was observed in both nonaqueous and aqueous mobile phases due to the positively charged amines on the surface of the prepared CSPs. For comparison, the traditionally used 3-aminopropyl silica (APS) was also adopted as the base material instead of DEAPS to prepare CSP. It was observed that the EOF on the DEAPS-based CSP was 18%-60% higher than that on the APS-based CSP under nonaqueous mobile phase conditions. Separation of enantiomers in CEC was performed on the positively charged CSPs with the nonaqueous mobile phases of pure ethanol or mixture of hexane-alcohol and the aqueous phases of acetonitrile-water or 95% ethanol. Fast separation of enantiomers was achieved on the newly prepared CSPs.

Synthesis of covalently bonded cellulose derivative chiral stationary phases with a bifunctional reagent of 3-(triethoxysilyl)propyl isocyanate

A bifunctional reagent of 3-(triethoxysilyl)propyl isocyanate (TEPI) was initially adopted as a spacer reagent to prepare the bonded types of chiral stationary phases (CSPs) with cellulose derivatives. The silica-based CSPs were chemically prepared with non-regioselective and regioselective approaches and their chiral resolving capabilities were evaluated in terms of HPLC resolution of test enantiomers. It was observed that the chiral recognition capabilities of the non-regioselectively prepared CSPs were influenced by the amount of TEPI used. And also, the regioselectively prepared CSP generally showed a slightly higher resolution power than the non-regioselectively prepared CSP, while the non-regioselective procedures were highly advantageous to rapid preparation. In addition, chiral recognition of the prepared CSPs was affected by the properties of the used silica matrices. (C) 2003 Elsevier B.V. All rights reserved.

Syntheses of opened hollow clay microspheres through a spray-drying approach and their derivative clay/carbon nanotubes composites

Opened hollow microspheres of organoclays were prepared via spray drying the suspension of modified Na+-montmorillonite (Na+-MMT) with alkylsulfonate. The microstructure and thermal properties of these opened hollow spheres were characterized by means of wide-angle X-ray diffraction, field emission scanning electron microscopy, and thermogravimetric analysis. The results showed that the organoclays had larger interlayer spacing compared with pure Na+-MMT and higher thermal stability relative to the alkylsufonate.