Statistical and Hybrid Methods for Speech Recognition in Romanian

The present paper describes the evolution of our work concerning the problem of speech recognition. Beginning with a classical hidden Markov model (HMM), we have investigated two ways to improve the performance of this basic structure. The first way was to realize a neuro-statistical hybrid by integrating a multilayer perceptron (MLP) as a posteriori probability estimator. The system was further refined by adding supplementary discriminative training (DT) based on the minimum classification error (MCE). Tests performed on a 15,000 isolated spoken-word database, showed an increase in the recognition rate from 92.2% for the HMM-based recognition system, to 94.7% for the HMM-MLP system, and then to 98.1% for the refined HMM-MLP-DT system. The second way to improve the classical HMM was to build a fuzzy-statistical hybrid, FHMM, based on a fuzzy similarity measure instead of the probabilistic measure specific to the usual statistical model. The benefits of the fuzzy measure introduction were evaluated on a vowel recognition task, and a decrease of approximately 3% in the error rate is reported.     

Doubly coupled systems of elliptic hemivariational inequalities: Existence and location

We develop an approach based on the subsolution–supersolution method for an elliptic system of hemivariational inequalities. The system exhibits full dependence on the gradient of the solution and is doubly coupled on both the source and multivalued terms. We prove the existence of solutions in a prescribed trapping region and, as an application, provide a criterion for obtaining positive solutions.     

Computationally designed monomers for molecular imprinting of chemical warfare agents - Part V

The main objective of this research is to develop and apply state-of-the-art computational tools to achieve an understanding of intermolecular interactions in molecular imprinting of chemical warfare (CW) agents into complex monomeric systems. Molecular dynamic (MD) simulations were carried out for different monomeric molecular systems in order to predict the interaction energies, the closest approach distances and the active site groups between the simulated molecular systems and different CW agents. The minimized structures of CW agents have been obtained with the use of molecular mechanics approach. NVT MD simulations at room temperature were carried out to obtain equilibrated conformations in all cases. The simulated molecular systems consisted of a ligand (CW agents) and commonly used functional monomers.During this study, it was found that electrostatic interactions play the most significant role in the formation of molecular imprinting materials. The simulated systems indicate that the functional groups of monomers interacting with ligands tend to be either -COOH or CH₂CH-.