We wish to understand brain; discover its sophisticated ways of calculations to invent improved computational methods. To decipher any complex system, first its components should be understood. Brain comprises neurons. Neurobiologists use morphologic properties like “somatic perimeter”, “axonal length”, and “number of dendrites” to classify neurons. They have discerned two types of neurons: “interneurons” and “pyramidal cells”, and have a consensus about five classes of interneurons: PV, 2/3, Martinotti, Chandelier, and NPY. They still need a more refined classification of interneurons because they suppose its known classes may contain subclasses or new classes may arise. This is a difficult process because of the great number and diversity of interneurons and lack of objective indices to classify them. Machine learning—automatic learning from data—can overcome the mentioned difficulties, but it needs a data set to learn from. To meet this demand neurobiologists compiled a data set from measuring 67 morphologic properties of 220 interneurons of mouse brains; they also labeled some of the samples—i.e. added their opinion about the sample’s classes. This project aimed to use machine learning to determine the true number of classes within the data set, classes of the unlabeled samples, and the accuracy of the available class labels. We used K-means, seeded K-means, and constrained K-means, and clustering validity techniques to achieve our objectives. Our results indicate that: the data set contains seven classes; seeded K-means outperforms K-means and constrained K-means; chandelier and 2/3 are the most consistent classes, whereas PV and Martinotti are the least consistent ones.

We have a wide-range of classifiers today that are employed in numerous applications, from credit scoring to speech-processing, with great technical and commercial success. No classifier, however, exists that will outperform all other classifiers on all classification tasks, and the process of classifier selection is still mainly one of trial and error. The optimal classifier for a classification task is determined by the characteristics of the data set employed; understanding the relationship between data characteristics and the performance of classifiers is therefore crucial to the process of classifier selection. Empirical and theoretical approaches have been employed in the literature to define this relationship. None of these approaches have, however, been very successful in accurately predicting or explaining classifier performance on real-world data. We use theoretical properties of classifiers to identify data characteristics that influence classifier performance; these data properties guide us in the development of measures that describe the relationship between data characteristics and classifier performance. We employ these data measures on real-world and artificial data to construct a meta-classification system. We use theoretical properties of classifiers to identify data characteristics that influence classifier performance; these data properties guide us in the development of measures that describe the relationship between data characteristics and classifier performance. We employ these data measures on real-world and artificial data to construct a meta-classification system. The purpose of this meta-classifier is two-fold: (1) to predict the classification performance of real-world classification tasks, and (2) to explain these predictions in order to gain insight into the properties of real-world data. We show that these data measures can be employed successfully to predict the classification performance of real-world data sets; these predictions are accurate in some instances but there is still unpredictable behaviour in other instances. We illustrate that these data measures can give valuable insight into the properties and data structures of real-world data; these insights are extremely valuable for high-dimensional classification problems. / Dissertation (MEng)--University of Pretoria, 2008. / Electrical, Electronic and Computer Engineering / unrestricted

This thesis proposes a novel approach to attribute clustering. It exploits the strength of semi-supervised learning to improve the quality of attribute clustering particularly when labeled data is limited. The significance of this work derives in part from the broad, and increasingly important, usage of attribute clustering to address outstanding problems within the machine learning community. This form of clustering has also been shown to have strong practical applications, being usable in heavyweight industrial applications. Although researchers have focused on supervised and unsupervised attribute clustering in recent years, semi-supervised attribute clustering has not received substantial attention. In this research, we propose an innovative two step iterative semi-supervised attribute clustering framework. This new framework, in each iteration, uses the result of attribute clustering to improve a classifier. It then uses the classifier to augment the training data used by attribute clustering in next iteration. This iterative framework outputs an improved classifier and attribute clustering at the same time. It gives more accurate clusters of attributes which better fit the real relations between attributes. In this study we proposed two new usages for attribute clustering to improve classification: solving the automatic view definition problem for multi-view learning and improving missing attribute-value handling at induction and prediction time. The application of these two new usages of attribute clustering in our proposed semi-supervised attribute clustering is evaluated using real world data sets from different domains.

Identifying differing opinions on a given topic as expressed by multiple people (as in a set of written reviews for a given product, for example) presents challenges. Opinions about a particular subject are often nuanced: a person may have both negative and positive opinions about different aspects of the subject of interest, and these aspect-specific opinions can be independent of the overall opinion on the subject. Being able to identify, collect, and count these nuanced opinions in a large set of data offers more insight into the strengths and weaknesses of competing products and services than does aggregating the overall ratings of such products and services. I make two useful and useable contributions in working with opinionated text. First, I present my implementation of a semi-supervised co-training machine classification method for identifying both product aspects (features of products) and sentiments expressed about such aspects. It offers better precision than fully-supervised methods while requiring much less text to be manually tagged (a time-consuming process). This algorithm can also be run in a fully supervised manner when more data is available. Second, I apply this co-training approach to reviews of restaurants and various electronic devices; such text contains both factual statements and opinions about features/aspects of products. The algorithm automatically identifies the product aspects and the words that indicate aspect-specific opinion polarity, while largely avoiding the problem of misclassifying the products themselves as inherently positive or negative. This method performs well compared to other approaches. When run on a set of reviews of five technology products collected from Amazon, the system performed with some demonstrated competence (with an average precision of 0.83) at the difficult task of simultaneously identifying aspects and sentiments, though comparison to contemporaries' simpler rules-based approaches was difficult. When run on a set of opinionated sentences about laptops and restaurants that formed the basis of a shared challenge in the SemEval-2014 Task 4 competition, it was able to classify the sentiments expressed about aspects of laptops better than any team that competed in the task (achieving 0.72 accuracy). It was above the mean in its ability to identify the aspects of restaurants about which people expressed opinions, even when co-training using only half of the labelled training data at the outset. While the SemEval-2014 aspect-based sentiment extraction task considered only separately the tasks of identifying product aspects and determining their polarities, I take an extra step and evaluate sentences as a whole, inferring aspects and the aspect-specific sentiments expressed simultaneously, a more difficult task that seems more applicable to real-world tasks. I present first results of this sentence-level task. The algorithm uses both lexical and syntactic information in a manner that is shown to be able to handle new words that it has never before seen. It offers some demonstrated ability to adapt to new subject domains for which it has no training data. The system is characterizable by very high precision and weak-to-average recall and it estimates its own confidence in its predictions; this characteristic should make the algorithm suitable for use on its own or for combination in a confidence-based voting ensemble. The software created for and described in the course of this dissertation is made available online.

Ces dernières années, l’intérêt pour l’apprentissage automatique n’a cessé d’augmenter dans des domaines aussi variés que la reconnaissance d’images ou l’analyse de données médicales. Cependant, une limitation du cadre classique PAC a récemment été mise en avant. Elle a entraîné l’émergence d’un nouvel axe de recherche : l’Adaptation de Domaine, dans lequel on considère que les données d’apprentissage proviennent d’une distribution (dite source) différente de celle (dite cible) dont sont issues les données de test. Les premiers travaux théoriques effectués ont débouché sur la conclusion selon laquelle une bonne performance sur le test peut s’obtenir en minimisant à la fois l’erreur sur le domaine source et un terme de divergence entre les deux distributions. Trois grandes catégories d’approches s’en inspirent : par repondération, par reprojection et par auto-étiquetage. Dans ce travail de thèse, nous proposons deux contributions. La première est une approche de reprojection basée sur la théorie du boosting et s’appliquant aux données numériques. Celle-ci offre des garanties théoriques intéressantes et semble également en mesure d’obtenir de bonnes performances en généralisation. Notre seconde contribution consiste d’une part en la proposition d’un cadre permettant de combler le manque de résultats théoriques pour les méthodes d’auto-étiquetage en donnant des conditions nécessaires à la réussite de ce type d’algorithme. D’autre part, nous proposons dans ce cadre une nouvelle approche utilisant la théorie des (epsilon, gamma, tau)-bonnes fonctions de similarité afin de contourner les limitations imposées par la théorie des noyaux dans le contexte des données structurées / During the past few years, an increasing interest for Machine Learning has been encountered, in various domains like image recognition or medical data analysis. However, a limitation of the classical PAC framework has recently been highlighted. It led to the emergence of a new research axis: Domain Adaptation (DA), in which learning data are considered as coming from a distribution (the source one) different from the one (the target one) from which are generated test data. The first theoretical works concluded that a good performance on the target domain can be obtained by minimizing in the same time the source error and a divergence term between the two distributions. Three main categories of approaches are derived from this idea : by reweighting, by reprojection and by self-labeling. In this thesis work, we propose two contributions. The first one is a reprojection approach based on boosting theory and designed for numerical data. It offers interesting theoretical guarantees and also seems able to obtain good generalization performances. Our second contribution consists first in a framework filling the gap of the lack of theoretical results for self-labeling methods by introducing necessary conditions ensuring the good behavior of this kind of algorithm. On the other hand, we propose in this framework a new approach, using the theory of (epsilon, gamma, tau)- good similarity functions to go around the limitations due to the use of kernel theory in the specific context of structured data

A anotac~ao de papeis sem^anticos (APS) e uma tarefa do processamento de lngua natural (PLN) que permite analisar parte do signicado das sentencas atraves da detecc~ao dos participantes dos eventos (e dos eventos em si) que est~ao sendo descritos nelas, o que e essencial para que os computadores possam usar efetivamente a informac~ao codicada no texto. A maior parte das pesquisas desenvolvidas em APS tem sido feita para textos em ingl^es, considerando as particularidades gramaticais e sem^anticas dessa lngua, o que impede que essas ferramentas e resultados sejam diretamente transportaveis para outras lnguas como o portugu^es. A maioria dos sistemas de APS atuais emprega metodos de aprendizado de maquina supervisionado e, portanto, precisa de um corpus grande de senten cas anotadas com papeis sem^anticos para aprender corretamente a tarefa. No caso do portugu^es do Brasil, um recurso lexical que prov^e este tipo de informac~ao foi recentemente disponibilizado: o PropBank.Br. Contudo, em comparac~ao com os corpora para outras lnguas como o ingl^es, o corpus fornecido por este projeto e pequeno e, portanto, n~ao permitiria que um classicador treinado supervisionadamente realizasse a tarefa de anotac~ao com alto desempenho. Para tratar esta diculdade, neste trabalho emprega-se uma abordagem semissupervisionada capaz de extrair informac~ao relevante tanto dos dados anotados disponveis como de dados n~ao anotados, tornando-a menos dependente do corpus de treinamento. Implementa-se o algoritmo self-training com modelos de regress~ ao logstica (ou maxima entropia) como classicador base, para anotar o corpus Bosque (a sec~ao correspondente ao CETENFolha) da Floresta Sinta(c)tica com as etiquetas do PropBank.Br. Ao algoritmo original se incorpora balanceamento e medidas de similaridade entre os argumentos de um verbo especco para melhorar o desempenho na tarefa de classicac~ao de argumentos. Usando um benchmark de avaliac~ao implementado neste trabalho, a abordagem semissupervisonada proposta obteve um desempenho estatisticamente comparavel ao de um classicador treinado supervisionadamente com uma maior quantidade de dados anotados (80,5 vs. 82,3 de \'F IND. 1\', p > 0, 01) / Semantic role labeling (SRL) is a natural language processing (NLP) task able to analyze part of the meaning of sentences through the detection of the events they describe and the participants involved, which is essential for computers to eectively understand the information coded in text. Most of the research carried out in SRL has been done for texts in English, considering the grammatical and semantic particularities of that language, which prevents those tools and results to be directly transported to other languages such as Portuguese. Most current SRL systems use supervised machine learning methods and require a big corpus of sentences annotated with semantic roles in order to learn how to perform the task properly. For Brazilian Portuguese, a lexical resource that provides this type of information has recently become available: PropBank.Br. However, in comparison with corpora for other languages such as English, the corpus provided by that project is small and it wouldn\'t allow a supervised classier to perform the labeling task with good performance. To deal with this problem, in this dissertation we use a semi-supervised approach capable of extracting relevant information both from annotated and non-annotated data available, making it less dependent on the training corpus. We implemented the self-training algorithm with logistic regression (or maximum entropy) models as base classier to label the corpus Bosque (section CETENFolha) from the Floresta Sintá(c)tica with the PropBank.Br semantic role tags. To the original algorithm, we incorporated balancing and similarity measures between verb-specic arguments so as to improve the performance of the system in the argument classication task. Using an evaluation benchmark implemented in this research project, the proposed semi-supervised approach has a statistical comparable performance as the one of a supervised classier trained with more annotated data (80,5 vs. 82,3 de \'F IND. 1\', p > 0, 01).

Multi-Task Learning is today an interesting and promising field which many mention as a must for achieving the next level advancement within machine learning. However, in reality, Multi-Task Learning is much more rarely used in real-world implementations than its more popular cousin Transfer Learning. The questionis why that is and if Multi-Task Learning outperforms its Single-Task counterparts. In this thesis different Multi-Task Learning architectures were utilized in order to build a model that can handle labeling real technical issues within two categories. The model faces a challenging imbalanced data set with many labels to choose from and short texts to base its predictions on. Can task-sharing be the answer to these problems? This thesis investigated three Multi-Task Learning architectures and compared their performance to a Single-Task model. An authentic data set and two labeling tasks was used in training the models with the method of supervised learning. The four model architectures; Single-Task, Multi-Task, Cross-Stitched and the Shared-Private, first went through a hyper parameter tuning process using one of the two layer options LSTM and GRU. They were then boosted by auxiliary tasks and finally evaluated against each other.

<div>Structured prediction makes coherent decisions as structured objects to present the interrelations of these predicted variables. They have been widely used in many areas, such as bioinformatics, computer vision, speech recognition, and natural language processing. Machine Learning with reduced supervision aims to leverage the laborious and error-prone annotation effects and benefit the low-resource languages. In this dissertation we study structured prediction with reduced supervision for two sets of problems, sequence labeling and dependency parsing, both of which are representatives of structured prediction problems in NLP. We investigate three different approaches.</div><div> </div><div>The first approach is learning with modular architecture by task decomposition. By decomposing the labels into location sub-label and type sub-label, we designed neural modules to tackle these sub-labels respectively, with an additional module to infuse the information. The experiments on the benchmark datasets show the modular architecture outperforms existing models and can make use of partially labeled data together with fully labeled data to improve on the performance of using fully labeled data alone.</div><div><br></div><div>The second approach builds the neural CRF autoencoder (NCRFAE) model that combines a discriminative component and a generative component for semi-supervised sequence labeling. The model has a unified structure of shared parameters, using different loss functions for labeled and unlabeled data. We developed a variant of the EM algorithm for optimizing the model with tractable inference. The experiments on several languages in the POS tagging task show the model outperforms existing systems in both supervised and semi-supervised setup.</div><div><br></div><div>The third approach builds two models for semi-supervised dependency parsing, namely local autoencoding parser (LAP) and global autoencoding parser (GAP). LAP assumes the chain-structured sentence has a latent representation and uses this representation to construct the dependency tree, while GAP treats the dependency tree itself as a latent variable. Both models have unified structures for sentence with and without annotated parse tree. The experiments on several languages show both parsers can use unlabeled sentences to improve on the performance with labeled sentences alone, and LAP is faster while GAP outperforms existing models.</div>

abstract: In recent years, Convolutional Neural Networks (CNNs) have been widely used in not only the computer vision community but also within the medical imaging community. Specifically, the use of pre-trained CNNs on large-scale datasets (e.g., ImageNet) via transfer learning for a variety of medical imaging applications, has become the de facto standard within both communities. However, to fit the current paradigm, 3D imaging tasks have to be reformulated and solved in 2D, losing rich 3D contextual information. Moreover, pre-trained models on natural images never see any biomedical images and do not have knowledge about anatomical structures present in medical images. To overcome the above limitations, this thesis proposes an image out-painting self-supervised proxy task to develop pre-trained models directly from medical images without utilizing systematic annotations. The idea is to randomly mask an image and train the model to predict the missing region. It is demonstrated that by predicting missing anatomical structures when seeing only parts of the image, the model will learn generic representation yielding better performance on various medical imaging applications via transfer learning. The extensive experiments demonstrate that the proposed proxy task outperforms training from scratch in six out of seven medical imaging applications covering 2D and 3D classification and segmentation. Moreover, image out-painting proxy task offers competitive performance to state-of-the-art models pre-trained on ImageNet and other self-supervised baselines such as in-painting. Owing to its outstanding performance, out-painting is utilized as one of the self-supervised proxy tasks to provide generic 3D pre-trained models for medical image analysis. / Dissertation/Thesis / Masters Thesis Computer Science 2020

The supervised and semi-supervised machine learning paradigms hinge on the idea that the training data is labeled. The label quality is often brought into question, and problems related to noisy, inaccurate, or missing labels are studied. One of these is an interesting and prevalent problem in the semi-supervised classification area where only some positive labels are known. At the same time, the remaining and often the majority of the available data is unlabeled, i.e., there are no negative examples. Known as Positive-Unlabeled (PU) learning, this problem has been identified with increasing frequency across many disciplines, including but not limited to health science, biology, bioinformatics, geoscience, physics, business, and politics. Also, there are several closely related machine learning problems, such as cost-sensitive learning and mixture proportion estimation. This dissertation explores the PU learning problem from the perspective of density estimation and proposes a new modular method compatible with the relabeling framework that is common in PU learning literature. This approach is compared with two existing algorithms throughout the manuscript, one from a seminal work by Elkan and Noto and a current state-of-the-art algorithm by Ivanov. Furthermore, this thesis identifies two machine learning application domains that can benefit from PU learning approaches, which were not previously seen that way: predicting length of stay in hospitals and automatic music tagging. Experimental results with multiple synthetic and real-world datasets from different application domains validate the proposed approach. Accurately predicting the in-hospital length of stay (LOS) at the time of admission can positively impact healthcare metrics, particularly in novel response scenarios such as the Covid-19 pandemic. During the regular steady-state operation, traditional classification algorithms can be used for this purpose to inform planning and resource management. However, when there are sudden changes to the admission and patient statistics, such as during the onset of a pandemic, these approaches break down because reliable training data becomes available only gradually over time. This thesis demonstrates the effectiveness of PU learning approaches in such situations through experiments by simulating the positive-unlabeled scenario using two fully-labeled publicly available LOS datasets. Music auto-tagging systems are typically trained using tag labels provided by human listeners. In many cases, this labeling is weak, which means that the provided tags are valid for the associated tracks, but there can be tracks for which a tag would be valid but not present. This situation is analogous to PU learning with the additional complication of being a multi-label scenario. Experimental results on publicly available music datasets with tags representing three different labeling paradigms demonstrate the effectiveness of PU learning techniques in recovering the missing labels and improving auto-tagger performance. / Graduate