Performance analysis of ASR system in hybrid DNN-HMM framework using a PWL euclidean activation function

Automatic Speech Recognition (ASR) is the process of mapping an acoustic speech signal into a human readable text format. Traditional systems exploit the Acoustic Component of ASR using the Gaussian Mixture Model- Hidden Markov Model (GMM-HMM) approach.Deep NeuralNetwork (DNN) opens up new possibilities to overcome the shortcomings of conventional statistical algorithms. Recent studies modeled the acoustic component of ASR system using DNN in the so called hybrid DNN-HMM approach. In the context of activation functions used to model the non-linearity in DNN, Rectified Linear Units (ReLU) and maxout units are mostly used in ASR systems. This paper concentrates on the acoustic component of a hybrid DNN-HMM system by proposing an efficient activation function for the DNN network. Inspired by previous works, euclidean norm activation function is proposed to model the non-linearity of the DNN network. Such non-linearity is shown to belong to the family of Piecewise Linear (PWL) functions having distinct features. These functions can capture deep hierarchical features of the pattern. The relevance of the proposal is examined in depth both theoretically and experimentally. The performance of the developed ASR system is evaluated in terms of Phone Error Rate (PER) using TIMIT database. Experimental results achieve a relative increase in performance by using the proposed function over conventional activation functions.     

On the method of calibration of the energy dispersive EXAFS beamline at Indus-2 and fitting theoretical model to the EXAFS spectrum

A procedure for calibration of the recently developed energy dispersive EXAFS beamline at the Indus-2 synchrotron source at RRCAT, Indore, India has been described. The procedure involves recording of absorption spectra of two standard samples, whose absorption edge energies are well-established. Two methods have been considered for calibration. In the first method, the position of the first maximum of the derivative of absorption curve is taken as the position of the edge energy. In the second method, the position of the point at half edge step in the absorption curve is taken as the position of the edge energy. It has been shown that only the first method gives same values of dispersion even when the beam current is varied and should be used for calibrating the experimental spectra. Further, it is recommended that the performance of the beamline for a particular setting should be checked by recording and analysing EXAFS of a standard. Hence, the procedure for analysis and extracting information about the various parameters that can be determined by fitting the EXAFS data with a theoretical model has also been described, by taking the example of K-absorption spectra of copper metal foil recorded on this beamline.     

Chitosan–silver oxide nanocomposite film: Preparation and antimicrobial activity

The chitosan–silver oxide encapsulated nanocomposite film was prepared by solution casting method. The prepared film was characterized by FTIR, scanning electron microscopy (SEM), thermal studies, and UV-Vis spectroscopy. The elemental composition of the film was studied by energy dispersive X-ray analysis (EDAX). The antibacterial activity of the composite film against pathogenic bacteria viz. Escherichia coli, Staphylococcus aureus, Bacillus subtilis and Pseudomonas aeruginosa was measured by agar diffusion method. Our observations suggest that chitosan as biomaterial based nanocomposite film containing silver oxide has an excellent antibacterial ability for food packaging applications.     

Conductivity study of solid polyelectrolytes based on hydroiodide salt of poly(4-vinyl pyridine-co-butylmethacrylate), poly(4-vinyl pyridine-co-butylacrylate)

The chain flexibility of poly(4-vinylpyridine) was tried to increase by lowering its glass transition temperature ( T _g ) and by increasing its amorphous region by copolymerizing with butyl methacrylate and butylacrylate which act as internal plasticizer. The copolymers were prepared in five different feed molar ratios to optimize the required properties such as higher room temperature conductivity and better film-forming capacity. The conductivity and conduction behaviour of the copolymers, as well as their hydroiodide salts have been reported. There was about 10³–10⁴-fold increase in room temperature conductivity of these plasticized polyelectrolytes.     

Seismic behavior of cantilever wall embedded in dry and saturated sand

The embedded cantilever retaining walls are often required for excavation to construct the underground facilities. Significant numbers of numerical and experimental studies have been performed to understand the behavior of embedded cantilever retaining walls under static condition. However, very limited studies have been conducted on the behavior of embedded retaining walls under seismic condition. In this paper, the behavior of a small scale model embedded cantilever retaining wall in dry and saturated sand under seismic loading condition is investigated by shake table tests in the laboratory and numerically using software FLAC2D. The embedded cantilever walls are subjected to sinusoidal dynamic motions. The behaviors of the cantilever walls in terms of lateral displacement and bending moment are studied with the variation of the two important design parameters, peak amplitude of the base motions and excavation depth. The variation of the pore water pressures within the sand is also observed in the cases of saturated sand. The maximum lateral displacement of a cantilever wall due to seismic loading is below 1% of the total height of the wall in dry sand, but in case of saturated sand, it can go up to 12.75% of the total height of the wall.     

MASS TRANSFER AUGMENTATION DUE TO WALL SLIP IN HAEMODIALYSERS†

The inadequacy of the current theories to accurately predict the haemodialyser performance is shown to be a result of the effective slip at the membrane surface. This slip phenomenon, manifested by the Fahraeus-Lindqvist effect is peculiar to (heterogeneous) blood flow. Literature data have been analyzed to substantiate the slip hypothesis proposed in this work. Both the theoretical analysis and the experimental observations indicate that slip enhances haemodialyser performance. Some design considerations have been provided, which will account for the beneficial effect of wall slip in practical haemodialyser operations.     

Plant growth-promoting rhizobacteria (PGPR) and its mechanisms against plant diseases for sustainable agriculture and better productivity

Plant growth-promoting rhizobacteria (PGPR) are specialized bacterial communities inhabiting the root rhizosphere and the secretion of root exudates helps to, regulate the microbial dynamics and their interactions with the plants. These bacteria viz., Agrobacterium, Arthobacter, Azospirillum, Bacillus, Burkholderia, Flavobacterium, Pseudomonas, Rhizobium, etc., play important role in plant growth promotion. In addition, such symbiotic associations of PGPRs in the rhizospheric region also confer protection against several diseases caused by bacterial, fungal and viral pathogens. The biocontrol mechanism utilized by PGPR includes direct and indirect mechanisms direct PGPR mechanisms include the production of antibiotic, siderophore, and hydrolytic enzymes, competition for space and nutrients, and quorum sensing whereas, indirect mechanisms include rhizomicrobiome regulation via. secretion of root exudates, phytostimulation through the release of phytohormones viz., auxin, cytokinin, gibberellic acid, 1-aminocyclopropane-1-carboxylate and induction of systemic resistance through expression of antioxidant defense enzymes viz., phenylalanine ammonia lyase (PAL), peroxidase (PO), polyphenyloxidases (PPO), superoxide dismutase (SOD), chitinase and β-glucanases. For the suppression of plant diseases potent bio inoculants can be developed by modulating the rhizomicrobiome through rhizospheric engineering. In addition, understandings of different strategies to improve PGPR strains, their competence, colonization efficiency, persistence and its future implications should also be taken into consideration.

     

Velocity Distributions of Food Particle Suspensions in Holding Tube Flow: Experimental and Modeling Studies on Average Particle Velocities

Average particle velocities of model food particles were investigated by videotaping particles suspended in sodium carboxymethylcellulose solutions during passage through a transparent holding tube similar in dimension to that of commercial aseptic processing systems. Results indicated the mean normalized velocity of particles was significantly affected by particle Froude number and a dimensionless viscosity. Within the range of this study, particle concentration effects were significant only with respect to standard deviations of distributions. Generalized Reynolds number poorly correlated with mean normalized particle velocity. A mathematical model for particle action was in qualitative agreement with experimental results.     

Adaptive learning algorithm in classification of fuzzy patterns An application to vowels in CNC context

An adaptive algorithm for recognition of ill-defined patterns using weak representative points and single pattern training procedure is presented from the standpoint of fuzzy set theory. The method includes both supervised and non-supervised schemes, A non-adaptive algorithm with fixed reference and weight vectors is also described to describe the efficiency of the system's adaptiveness to a new input.

This was implemented to machine recognition of vowel sounds of a number of speakers in Consonant-Vowel Nucleus-Consonant (CNC) context considering the first three vowel formants as input features. The decision of the machine is governed by the maximum value of fuzzy membership function. A recognition rate, particularly for weak initial representative vectors, was seen to be dependent on the sequence of incoming patterns. As the process of classification continued, the learned moan vectors approached their respective true values of the clusters. Again, once the optimum size of training set is obtained, the role of the external supervisor became insignificant.