Prematurity, especially if extreme, is one of the leading causes of problems and/or death after delivery. Among all the problems encountered by premature infants, feeding difficulties are very common. Many premature infants are fed intravenously at first, and they progress to milk feedings provided by a tube passed into the stomach. At around 34 weeks of gestation, premature infants should be able to breastfeed or take a bottle. At the same time such premature infants are usually faced with difficulty making the transition from tube-feeding to full oral feeding. In this study three physiological measurements of premature infants including sucking, swallowing and breathing were measured. The objective of this work was to detect, identify and classify these three signals independently and in relation to each other. The goal was to look at the specification of sucking, swallowing and breathing signals to extract the ratio of suck swallow-breath coordination. The results of this study were used to predict the readiness of a premature infant for introduction to oral feeding.To accomplish this, three different methods were examined. In the first method, the integration of the wavelet packet transform and a neural network was investigated. According to results of the first approach, integration of the wavelet packet transform and the neural network failed due to the inefficiency of the feature extraction method. Thus, the wavelet packet energy nodes did not provide a good feature extraction tool in this specific application.In the second approach, the frequency content of each signal was investigated to study the relationship between the shape of each waveform and the frequency content of that specific signal. Spectral analysis for suck, swallow and breathing signals showed that the shape of the signal was not tightly related to the frequency content of that specific waveform. Therefore, the frequency content could not be used as a method of feature extraction in this specific application.In the third method, the integration of correlation and matched filtering techniques was investigated and demonstrated promising result for the detection of suck and breathing signal but not for the swallowing waveform. Based on the results for sucking and breathing signals, this method should also work for good quality swallowing signal. To understand the relationship between the suck, swallow and breathing signals a matrix containing information on the time of occurrence of each event was developed.

This thesis studies a family of unsupervised learning algorithms called feature distribution learning and their extension to perform covariate shift adaptation. Unsupervised learning is one of the most active areas of research in machine learning, and a central challenge in this field is to develop simple and robust algorithms able to work in real-world scenarios. A traditional assumption of machine learning is the independence and identical distribution of data. Unfortunately, in realistic conditions this assumption is often unmet and the performances of traditional algorithms may be severely compromised. Covariate shift adaptation has then developed as a lively sub-field concerned with designing algorithms that can account for covariate shift, that is for a difference in the distribution of training and test samples. The first part of this dissertation focuses on the study of a family of unsupervised learning algorithms that has been recently proposed and has shown promise: feature distribution learning; in particular, sparse filtering, the most representative feature distribution learning algorithm, has commanded interest because of its simplicity and state-of-the-art performance. Despite its success and its frequent adoption, sparse filtering lacks any strong theoretical justification. This research questions how feature distribution learning can be rigorously formalized and how the dynamics of sparse filtering can be explained. These questions are answered by first putting forward a new definition of feature distribution learning based on concepts from information theory and optimization theory; relying on this, a theoretical analysis of sparse filtering is carried out, which is validated on both synthetic and real-world data sets. In the second part, the use of feature distribution learning algorithms to perform covariate shift adaptation is considered. Indeed, because of their definition and apparent insensitivity to the problem of modelling data distributions, feature distribution learning algorithms seems particularly fit to deal with covariate shift. This research questions whether and how feature distribution learning may be fruitfully employed to perform covariate shift adaptation. After making explicit the conditions of success for performing covariate shift adaptation, a theoretical analysis of sparse filtering and another novel algorithm, periodic sparse filtering, is carried out; this allows for the determination of the specific conditions under which these algorithms successfully work. Finally, a comparison of these sparse filtering-based algorithms against other traditional algorithms aimed at covariate shift adaptation is offered, showing that the novel algorithm is able to achieve competitive performance. In conclusion, this thesis provides a new rigorous framework to analyse and design feature distribution learning algorithms; it sheds light on the hidden assumptions behind sparse filtering, offering a clear understanding of its conditions of success; it uncovers the potential and the limitations of sparse filtering-based algorithm in performing covariate shift adaptation. These results are relevant both for researchers interested in furthering the understanding of unsupervised learning algorithms and for practitioners interested in deploying feature distribution learning in an informed way.