Machine learning has enjoyed astounding practical success in a wide range of applications in recent years-practical success that often hurries ahead of our theoretical understanding. The standard framework for machine learning theory assumes full supervision, that is, training data consists of correctly labeled iid examples from the same task that the learned classifier is supposed to be applied to. However, many practical applications successfully make use of the sheer abundance of data that is currently produced. Such data may not be labeled or may be collected from various sources. The focus of this thesis is to provide theoretical analysis of machine learning regimes where the learner is given such (possibly large amounts) of non-perfect training data. In particular, we investigate the benefits and limitations of learning with unlabeled data in semi-supervised learning and active learning as well as benefits and limitations of learning from data that has been generated by a task that is different from the target task (domain adaptation learning). For all three settings, we propose Probabilistic Lipschitzness to model the relatedness between the labels and the underlying domain space, and we discuss our suggested notion by comparing it to other common data assumptions.

The analysis of microarray data is a challenging task because of the large dimensionality and small sample size involved. Although a few methods are available to address the problem of small sample size, they are not sufficiently successful in dealing with microarray data from extremely small (~<20) sample sizes. We propose a method to incorporate information from diverse sources to analyze the microarray data so as to improve the predictability of significant genes. A transformed data set, including statistical parameters, literature mining and gene ontology data, is evaluated. We performed classification experiments to identify potential allergy-related genes. Feature selection is used to identify the effect of features on classifier behaviour. An exploratory and domain knowledge analysis was performed on noisy real-life allergy data, and a subset of genes was selected as positive and negative class. A new set of transformed variables, depending on the mean and standard deviation statistics of the data distribution and other data sources, was identified. Significant allergy- and immune-related genes from the microarray data were selected. Experiments showed that classification predictability of significant genes can be improved. Important features from the transformed variable set were also identified.

No description available.

We hypothesized that temporal features of EEG are altered in sleep apnea subjects comparedto normal subjects. The initial aim was to develop a measure to discriminate sleep stages innormals. The longer-term goal was to apply these methods to identify differences in EEGactivity in sleep apnea subjects from normals. We analyzed the C3A2 EEG and anelectrooculogram (EOG) recorded from 9 normal adults awake and in rapid eye movement(REM) and non-REM sleep. The EEG signals were filtered to remove EOG contamination. Twomeasures of the irregularity of EEG signals, Sample Entropy (SpEn) and Tsallis Entropy, wereevaluated for their ability to discriminate sleep stages. SpEn changes with sleep state, beinglargest in Wake. Stage 3/4 had the smallest SpEn (0.57??0.11) normalized to Wake values,followed by Stage 2 (0.72??0.09), REM (0.75??0.1) and Stage 1 (0.89??0.05). This pattern wasconsistent in all the polysomnogram records analyzed. Similar pattern was observed in leadO1A2 as well. We conclude that SpEn may be useful as part of a montage for assessing sleepstate. We analyzed data from sleep apnea subjects having obstructive and central apnea eventsand have made some preliminary observations; the SpEn values were more similar across sleepstages and also high correlation with oxygen saturation was observed.

Background: Hepatitis E (HEV) is a small, non-enveloped virus that belongs to the viral genus Hepevirus. HEV is a positive sense single-stranded RNA virus and there is insufficient information regarding its replication. This is mainly because the virus has low capacity to grow in normally used cell cultures. Many specific strand assay detection studies have been done in order to understand more about HEV replication. Unfortunately, these assays have the disadvantage of giving false positive results. Aim: The aim of this project was to improve the positive strand assay to increase specificity and eliminate false positivity which is due to high sensitivity of the polymerase chain reaction (PCR). False positivity occurs as remains of transfected material in the cell are amplified. Method: The samples used in this project were swine samples from Sweden and a human sample (plasmid clone of genotype 1) from India. Negative samples, extracted positive samples and transcribed RNA positive sense samples were used. The methods applied were cDNA synthesis, exonuclease I and RNase treatments, DNA purification kits followed by first and nested PCR. Result: The results of this study indicated great improvement of the detection assay especially for the transcribed RNA samples. Best results were obtained at a final concentration of 1.5mM MgCl2 in the mastermix.  Conclusion: Changing the concentration of MgCl2 appeared to have a great effect on PCR specificity. Improving detection assays is very essential as they can be applied in the research field and in public health centers either for diagnosis or tracking disease outbreaks.

The cold fiber solid phase microextraction (SPME) system was improved by minimizing the coating temperature fluctuation range, and the performance of the system was evaluated by investigating the extraction of PAHs from spiked sand samples. The coating temperature can be made relatively constant and the relative standard division (RSD) for most compounds was smaller than 2%. A simplified cold fiber system without the solenoid valve was modified to connect CO2 delivery tubing directly to the liquid CO2 tank. The robustness of this system was evaluated with different sizes of CO2 delivery tubings. The system is stable, low cost and can be easily controlled, which provides a supplementary extraction strategy to the traditional cold fiber system. The extraction amount of the analyte in a specific system was calculated theoretically in advance. The extraction amount for the experiment agreed with that of the calculated result. By using theoretical calculations as a guide, desorption efficiency for aged spiked samples was investigated. In order to achieve better extraction efficiency for PAHs, a programmed coating temperature method was developed and optimized, which led to higher extraction efficiency for most studied analytes compared to the traditional methods. In real sample analysis, certified reference soils were analyzed using cold fiber SPME and the addition of diethylamine successfully realized the exhaustive extraction for volatile compounds and enhanced the recoveries for semi-volatile compounds. Satisfactory extraction amounts for all compounds were achieved by the proposed method after method optimization.

A method for the selective concentrating of DNA targets using capillary affinity gel electrophoresis is presented. Complementary ssDNA targets are retained through hybridization with oligonucleotide probes immobilized within polyacrylamide gels while non-complementary targets are removed. The captured DNA targets were concentrated by step elution, where a localized thermal zone was applied in small steps along the capillary. Evaluation of the selective capture of a 150 nt DNA target in a complicated mixture was carried out by factorial analysis. Gels with a smaller average pore size were found to retain a higher amount of complementary targets. This was thought to be due to the ssDNA target migrating through the gel by reptation, eliminating hairpin structures, making the complementary region of the target available for hybridization. This method was applied to a series of DNA targets of different lengths, 19 nt, 150 nt, 250 nt and 400 nt. The recovery of the method ranged from 0.5 to 4% for the PCR targets, and 13 to 18% for the 19 nt oligonucleotide target. The purity was calculated to be up to 44% for the PCR targets and up to 86% for the 19 nt target. This was an improvement in purity of up to 15 times and 1100 times in comparison to the original samples for the PCR targets and 19 nt oligonucleotide, respectively. The 19 nt targets were selective concentrated and delivered into a microfluidic based DNA biosensing platform. The purity of the sample improved from 0.01% to 50% while recovery decreased from 100% to 20% for a sample with 0.5 nM complementary and 1 μM non-complementary targets. An improvement in the response of the sensing platform was demonstrated on 19 nt oligonucleotide targets delivered by selective concentration versus concentration alone into the microfluidic biosensing system.

A method for the selective concentrating of DNA targets using capillary affinity gel electrophoresis is presented. Complementary ssDNA targets are retained through hybridization with oligonucleotide probes immobilized within polyacrylamide gels while non-complementary targets are removed. The captured DNA targets were concentrated by step elution, where a localized thermal zone was applied in small steps along the capillary. Evaluation of the selective capture of a 150 nt DNA target in a complicated mixture was carried out by factorial analysis. Gels with a smaller average pore size were found to retain a higher amount of complementary targets. This was thought to be due to the ssDNA target migrating through the gel by reptation, eliminating hairpin structures, making the complementary region of the target available for hybridization. This method was applied to a series of DNA targets of different lengths, 19 nt, 150 nt, 250 nt and 400 nt. The recovery of the method ranged from 0.5 to 4% for the PCR targets, and 13 to 18% for the 19 nt oligonucleotide target. The purity was calculated to be up to 44% for the PCR targets and up to 86% for the 19 nt target. This was an improvement in purity of up to 15 times and 1100 times in comparison to the original samples for the PCR targets and 19 nt oligonucleotide, respectively. The 19 nt targets were selective concentrated and delivered into a microfluidic based DNA biosensing platform. The purity of the sample improved from 0.01% to 50% while recovery decreased from 100% to 20% for a sample with 0.5 nM complementary and 1 μM non-complementary targets. An improvement in the response of the sensing platform was demonstrated on 19 nt oligonucleotide targets delivered by selective concentration versus concentration alone into the microfluidic biosensing system.

Traditional appearance based face recognition (FR) systems use gray scale images, however recently attention has been drawn to the use of color images. Color inputs have a larger dimensionality, which increases the computational cost, and makes the small sample size (SSS) problem in supervised FR systems more challenging. It is therefore important to determine the scenarios in which usage of color information helps the FR system. In this thesis, it was found that inclusion of chromatic information in FR systems is shown to be particularly advantageous in poor illumination conditions. In supervised systems, a color input of optimal dimensionality would improve the FR performance under SSS conditions. A fusion of decisions from individual spectral planes also helps in the SSS scenario. Finally, chromatic information is integrated into a supervised ensemble learner to address pose and illumination variations. This framework significantly boosts FR performance under a range of learning scenarios.

Ongoing work in the laboratory of Professor Chris Backhouse is aimed at developing a portable and inexpensive lab on a chip instrument. A system capable of molecular biology protocols including sample preparation (SP), polymerase chain reaction (PCR), and melting curve analysis (MCA) would meet the requirements for point of care genetic analysis. The SP, PCR, and MCA modules were designed and tested on a standalone basis and then integrated for analysis of raw clinical samples. An automated XY stage was developed for magnetic bead-based DNA purification. In addition, a LED/CCD-based optical detection module was employed for real time PCR and MCA. Data analysis algorithms and protocols were implemented to remove noise and interpret data. This work culminated in proof of principle on-chip SP-PCR-MCA to detect ß2m DNA from human buccal cells in a modular and inexpensive system. / Biomedical Engineering