A detailed investigation of classification methods for vowel speech imagery recognition

Accurate and fast decoding of speech imagery from electroencephalographic (EEG) data could serve as a basis for a new generation of brain computer interfaces (BCIs), more portable and easier to use. However, decoding of speech imagery from EEG is a hard problem due to many factors. In this paper we focus on the analysis of the classification step of speech imagery decoding for a three-class vowel speech imagery recognition problem. We empirically show that different classification subtasks may require different classifiers for accurately decoding and obtain a classification accuracy that improves the best results previously published. We further investigate the relationship between the classifiers and different sets of features selected by the common spatial patterns method. Our results indicate that further improvement on BCIs based on speech imagery could be achieved by carefully selecting an appropriate combination of classifiers for the subtasks involved.

TRANSFER OF ANALYTICAL METHODS FOR THE CHARACTERISATION OF NATURAL PRODUCTS TO R&D PROCESSES OF IDOKI

IDOKI SCF Technologies S.L. is a technology-based company, set up on September 2006 in Derio (Biscay) with the main scope of developing extraction and purification processes based on the use of supercritical fluid extraction technology (SFE) in food processing, extraction of natural products and the production of personal care products. IDOKI¿s researchers have been working on many different R&D projects so far, most of them using this technology. However, the optimization of a SFE method for the different matrices cannot be performed unless we have an analytical method for the characterisation of the extracts obtained in each experiment. The analytical methods are also essential for the quality control of the raw materials that are going to be used and also for the final product. This PhD thesis was born to tackle this problem and therefore, it is based on the development of different analytical methods for the characterisation of the extracts and products. The projects that we could include in this thesis were the following: the extraction propolis, the recovery of agroindustrial residues (soy and wine) and the dealcoholisation of wine.On the one hand, for the extraction of propolis, several UV-Vis spectroscopic methods were used in order to measure the antioxidant capacity and the total polyphenol and flavonoid content of the extracts. A SFC method was also developed in order to measure more specific phenolic compounds. On the other hand, for the recovery of agroindustrial residues UV-Vis spectroscopy was used to determine the total polyphenol content and two SFC methods were developed to analyse different phenolic compounds. Extraction methods such as MAE, FUSE and rotary agitation were also evaluated for the characterisation of the raw materials.Finally, for the dealcoholisation of wine, the development of a SBSE-TD-GC-MS and DHS-TD-GC-MS methods for the analysis of aromas and a NIR spectroscopic method for the determination of ethanol content with the help of chemometrics was necessary. Most of these methods are typically used in IDOKI¿s lab as routine analyses apart from others not included in this PhD thesis.

Properties of convergence of a class of iterative processes generated by sequences of self-mappings with applications to switched dynamic systems

This article investigates the convergence properties of iterative processes involving sequences of self-mappings of metric or Banach spaces. Such sequences are built from a set of primary self-mappings which are either expansive or non-expansive self-mappings and some of the non-expansive ones can be contractive including the case of strict contractions. The sequences are built subject to switching laws which select each active self-mapping on a certain activation interval in such a way that essential properties of boundedness and convergence of distances and iterated sequences are guaranteed. Applications to the important problem of stability of dynamic switched systems are also given.

An investigation of imputation methods for discrete databases and multi-variate time series

When it comes to information sets in real life, often pieces of the whole set may not be available. This problem can find its origin in various reasons, describing therefore different patterns. In the literature, this problem is known as Missing Data. This issue can be fixed in various ways, from not taking into consideration incomplete observations, to guessing what those values originally were, or just ignoring the fact that some values are missing. The methods used to estimate missing data are called Imputation Methods. The work presented in this thesis has two main goals. The first one is to determine whether any kind of interactions exists between Missing Data, Imputation Methods and Supervised Classification algorithms, when they are applied together. For this first problem we consider a scenario in which the databases used are discrete, understanding discrete as that it is assumed that there is no relation between observations. These datasets underwent processes involving different combina- tions of the three components mentioned. The outcome showed that the missing data pattern strongly influences the outcome produced by a classifier. Also, in some of the cases, the complex imputation techniques investigated in the thesis were able to obtain better results than simple ones. The second goal of this work is to propose a new imputation strategy, but this time we constrain the specifications of the previous problem to a special kind of datasets, the multivariate Time Series. We designed new imputation techniques for this particular domain, and combined them with some of the contrasted strategies tested in the pre- vious chapter of this thesis. The time series also were subjected to processes involving missing data and imputation to finally propose an overall better imputation method. In the final chapter of this work, a real-world example is presented, describing a wa- ter quality prediction problem. The databases that characterized this problem had their own original latent values, which provides a real-world benchmark to test the algorithms developed in this thesis.

hp-Adaptive simulation and inversion of magnetotelluric measurements

xlix, 121 p.