O grande volume de dados que temos disponíveis em ambientes virtuais pode ser excelente fonte de novos recursos para estudos em diversas tarefas de Processamento de Linguagem Natural, como a Análise de Sentimentos. Infelizmente é elevado o custo de anotação de novos córpus, que envolve desde investimentos financeiros até demorados processos de revisão. Nossa pesquisa propõe uma abordagem de anotação semissupervisionada, ou seja, anotação automática de um grande córpus não anotado partindo de um conjunto de dados anotados manualmente. Para tal, introduzimos o TweetSentBR, um córpus de tweets no domínio de programas televisivos que possui anotação em três classes e revisões parciais feitas por até sete anotadores. O córpus representa um importante recurso linguístico de português brasileiro, e fica entre os maiores córpus anotados na literatura para classificação de polaridades. Além da anotação manual do córpus, realizamos a implementação de um framework de aprendizado semissupervisionado que faz uso de dados anotados e, de maneira iterativa, expande o mesmo usando dados não anotados. O TweetSentBR, que possui 15:000 tweets anotados é assim expandido cerca de oito vezes. Para a expansão, foram treinados modelos de classificação usando seis classificadores de polaridades, assim como foram avaliados diferentes parâmetros e representações a fim de obter um córpus confiável. Realizamos experimentos gerando córpus expandidos por cada classificador, tanto para a classificação em três polaridades (positiva, neutra e negativa) quanto para classificação binária. Avaliamos os córpus gerados usando um conjunto de held-out e comparamos a FMeasure da classificação usando como treinamento os córpus anotados manualmente e semiautomaticamente. O córpus semissupervisionado que obteve os melhores resultados para a classificação em três polaridades atingiu 62;14% de F-Measure média, superando a média obtida com as avaliações no córpus anotado manualmente (61;02%). Na classificação binária, o melhor córpus expandido obteve 83;11% de F1-Measure média, superando a média obtida na avaliação do córpus anotado manualmente (79;80%). Além disso, simulamos nossa expansão em córpus anotados da literatura, medindo o quão corretas são as etiquetas anotadas semi-automaticamente. Nosso melhor resultado foi na expansão de um córpus de reviews de produtos que obteve FMeasure de 93;15% com dados binários. Por fim, comparamos um córpus da literatura obtido por meio de supervisão distante e nosso framework semissupervisionado superou o primeiro na classificação de polaridades binária em cross-domain. / The high volume of data available in the Internet can be a good resource for studies of several tasks in Natural Language Processing as in Sentiment Analysis. Unfortunately there is a high cost for the annotation of new corpora, involving financial support and long revision processes. Our work proposes an approach for semi-supervised labeling, an automatic annotation of a large unlabeled set of documents starting from a manually annotated corpus. In order to achieve that, we introduced TweetSentBR, a tweet corpora on TV show programs domain with annotation for 3-point (positive, neutral and negative) sentiment classification partially reviewed by up to seven annotators. The corpus is an important linguistic resource for Brazilian Portuguese language and it stands between the biggest annotated corpora for polarity classification. Beyond the manual annotation, we implemented a semi-supervised learning based framework that uses this labeled data and extends it using unlabeled data. TweetSentBR corpus, containing 15:000 documents, had its size augmented in eight times. For the extending process, we trained classification models using six polarity classifiers, evaluated different parameters and representation schemes in order to obtain the most reliable corpora. We ran experiments generating extended corpora for each classifier, both for 3-point and binary classification. We evaluated the generated corpora using a held-out subset and compared the obtained F-Measure values with the manually and the semi-supervised annotated corpora. The semi-supervised corpus that obtained the best values for 3-point classification achieved 62;14% on average F-Measure, overcoming the results obtained by the same classification with the manually annotated corpus (61;02%). On binary classification, the best extended corpus achieved 83;11% on average F-Measure, overcoming the results on the manually corpora (79;80%). Furthermore, we simulated the extension of labeled corpora in literature, measuring how well the semi-supervised annotation works. Our best results were in the extension of a product review corpora, achieving 93;15% on F1-Measure. Finally, we compared a literature corpus which was labeled by using distant supervision with our semi-supervised corpus, and this overcame the first in binary polarity classification on cross-domain data.

abstract: Humans have a great ability to recognize objects in different environments irrespective of their variations. However, the same does not apply to machine learning models which are unable to generalize to images of objects from different domains. The generalization of these models to new data is constrained by the domain gap. Many factors such as image background, image resolution, color, camera perspective and variations in the objects are responsible for the domain gap between the training data (source domain) and testing data (target domain). Domain adaptation algorithms aim to overcome the domain gap between the source and target domains and learn robust models that can perform well across both the domains. This thesis provides solutions for the standard problem of unsupervised domain adaptation (UDA) and the more generic problem of generalized domain adaptation (GDA). The contributions of this thesis are as follows. (1) Certain and Consistent Domain Adaptation model for closed-set unsupervised domain adaptation by aligning the features of the source and target domain using deep neural networks. (2) A multi-adversarial deep learning model for generalized domain adaptation. (3) A gating model that detects out-of-distribution samples for generalized domain adaptation. The models were tested across multiple computer vision datasets for domain adaptation. The dissertation concludes with a discussion on the proposed approaches and future directions for research in closed set and generalized domain adaptation. / Dissertation/Thesis / Masters Thesis Computer Science 2020

The study and the analysis of marine ecosystems is a significant part of the marine science research. These systems are valuable resources for fisheries, improving water quality and can even be used in drugs production. The investigation of ichthyoplankton inhabiting these ecosystems is also an important research field. Ichthyoplankton are fish in their early stages of life. In this stage, the fish have relatively similar shape and are small in size. The currently used way of identifying them is not optimal. Marine scientists typically study such organisms by sending a team that collects samples from the sea which is then taken to the lab for further investigation. These samples need to be studied by an expert and usually end needing a DNA sequencing. This method is time-consuming and requires a high level of experience. The recent advances in AI have helped to solve and automate several difficult tasks which motivated us to develop a classification tool for ichthyoplankton. We show that using machine learning techniques, such as generative adversarial networks combined with transfer learning solves such a problem with high accuracy. We show that using traditional machine learning algorithms fails to solve it. We also give a general framework for creating a classification tool when the dataset used for training is a limited dataset. We aim to build a user-friendly tool that can be used by any user for the classification task and we aim to give a guide to the researchers so that they can follow in creating a classification tool.

Data streams are large sequences of data, possibly endless and temporarily ordered, that are common-place in Internet of Things (IoT) applications such as intrusion detection in computer networking, fraud detection in financial institutions, real-time tumor tracking in radiotherapy and social media analysis. Algorithms learning from such streams need to be able to construct near real-time models that continuously adapt to potential changes in patterns, in order to retain high performance throughout the stream. It follows that there are numerous challenges involved in supervised learning (or so-called classification) in such environments. One of the challenges in learning from streams is multi-class imbalance, in which the rates of instances in the different class labels differ substantially. Notably, classification algorithms may become biased towards the classes with more frequent instances, sacrificing the performance of the less frequent or so-called minority classes. Further, minority instances often arrive infrequently and in bursts, making accurate model construction problematic. For example, network intrusion detection systems must be able to distinguish between normal traffic and multiple minority classes corresponding to a variety of different types of attacks. Further, having labels for all instances are often infeasible, since we might have missing or late-arriving labels. For instance, when learning from a stream regarding the task of detecting network intrusions, the true label for all instances might not be available, or it might take time until the label is made available, especially for new types of attacks. In this thesis, we contribute to the advancements of online learning from evolving streams by focusing on the above-mentioned areas of multi-class imbalance and missing labels. First, we introduce a multi-class online ensemble algorithm designed to maintain a balanced performance over all classes. Specifically, our approach samples instances with replacement while dynamically increasing the weights of under-represented classes, in order to produce models that benefit all classes. Our experimental results show that our online ensemble method performs well against multi-class imbalanced data in various datasets. We further continue our study by introducing an approach to dealing with missing labels that utilize both labelled and unlabelled data to increase a model’s performance. That is, our method utilizes labelled data for pseudo-labelling unlabelled instances, allowing the model to perform better in environments where labels are scarce. More specifically, our approach features a meta-reinforcement learning agent, trained on multiple-source streams, that can effectively select the prediction of a K nearest neighbours (K-NN) classifier as the label for unlabelled instances. Extensive experiments on benchmark datasets demonstrate the value and effectiveness of our approach and confirm that our method outperforms state-of-the-art.

The manufacturing process of carbide inserts at Sandvik Coromant consists of several operations. During some of these, the inserts are positioned on trays. For some inserts the trays are pre-defined but for others the insert-tray matching is partly improvised. The goal of this thesis project is to examine whether machine learning can be used to predict which tray to use for a given insert. It is also investigated which insert features are determining for the choice of tray. The study is done with insert and tray data from four blasting operations and considers a set of standardized inserts since it is assumed that the tray matching for these is well tuned. The algorithm that is used for the predictions is the supervised learning algorithm k-nearest neighbors. The problem of identifying the determining features is regarded as a feature selection problem and is done with the ReliefF algorithm. From the classification results it is seen that the classifiers are overfitting. The main reason for this is probably that the datasets contain features that together are uniquely defining for which tray is used. This was not detected during the feature selection since ReliefF identifies features that are individually relevant to the output. An idea to avoid overfitting the classifiers is to exclude these defining features from the dataset. Further work is thus recommended.

No description available.

In this thesis, three well known self-supervised methods have been implemented and trained on road scene images. The three so called pretext tasks RotNet, MoCov2, and DeepCluster were used to train a neural network self-supervised. The self-supervised trained networks where then evaluated on different amount of labeled data on two downstream tasks, object detection and semantic segmentation. The performance of the self-supervised methods are compared to networks trained from scratch on the respective downstream task. The results show that it is possible to achieve a performance increase using self-supervision on a dataset containing road scene images only. When only a small amount of labeled data is available, the performance increase can be substantial, e.g., a mIoU from 33 to 39 when training semantic segmentation on 1750 images with a RotNet pre-trained backbone compared to training from scratch. However, it seems that when a large amount of labeled images are available (>70000 images), the self-supervised pretraining does not increase the performance as much or at all.

Artificial intelligence and machine learning is used in a lot of different areas, one of those areas is image recognition. In the production of a TV-show or film, image recognition can be used to help the editors to find specific objects, scenes, or people in the video content, which speeds up the production. But image recognition is not working perfect all the time and can not be used in the production of a TV-show or film as it is intended to. Therefore the image recognition algorithms needs to be trained on large datasets to become better. But to create these datasets takes time and tools that can let users create specific datasets and retrain algorithms to become better is needed. The aim of this master thesis was to investigate if it was possible to create a tool that can annotate objects and people in video content and using the data as training sets, and a tool that can retrain the output of an image recognition to make the image recognition become better. It was also important that the tools have a good workflow for the users. The study consisted of a theoretical study to gain more knowledge about annotation, and how to make a good UX-design with a good workflow. Interviews were also held to get more knowledge of what the requirements of the product was. It resulted in a user scenario and a workflow that was used together with the knowledge from the theoretical study to create a hi-fi prototype by using an iterative process with usability testing. This resulted in a final hi-fi prototype with a good design and a good workflow for the users, where it is possible to annotate objects and people with a bounding box, and where it is possible to retrain an image recognition program that has been used on video content. / Artificiell intelligens och maskininlärning används inom många olika områden, ett av dessa områden är bildigenkänning. Vid produktionen av ett TV-program eller av en film kan bildigenkänning användas för att hjälpa redigerarna att hitta specifika objekt, scener eller personer i videoinnehållet, vilket påskyndar produktionen. Men bildigenkänningsprogram fungerar inte alltid helt perfekt och kan inte användas i produktionen av ett TV-program eller film som det är tänkt att användas i det sammanhanget. För att förbättra bildigenkänningsprogram så behöver dess algoritm tränas på stora datasets av bilder och labels. Men att skapa dessa datasets tar tid och det behövs program som kan skapa datasets och återträna algoritmer för bildigenkänning så att de fungerar bättre. Syftet med detta examensarbete var att undersöka om det var möjligt att skapa ett verktyg som kan markera(annotera) objekt och personer i video och använda datat som träningsdata för algoritmer. Men även att skapa ett verktyg som kan återträna algoritmer för bildigenkänning så att de blir bättre utifrån datat man får från ett bildigenkänningprogram. Det var också viktigt att dessa verktyg hade ett bra arbetsflöde för användarna. Studien bestod av en teoretisk studie för att få mer kunskap om annoteringar i video och hur man skapar bra UX-design med ett bra arbetsflöde. Intervjuer hölls också för att få mer kunskap om kraven på produkten och vilka som skulle använda den. Det resulterade i ett användarscenario och ett arbetsflöde som användes tillsammans med kunskapen från den teoretiska studien för att skapa en hi-fi prototyp, där en iterativ process med användbarhetstestning användes. Detta resulterade i en slutlig hi-fi prototyp med bra design och ett bra arbetsflöde för användarna där det är möjligt att markera(annotera) objekt och personer med en bounding box och där det är möjligt att återträna algoritmer för bildigenkänning som har körts på video.

Dans de nombreux problèmes en apprentissage supervisé, les entrées ont une structure de groupes connue et/ou clairement identifiable. Dans ce contexte, l'élaboration d'une règle de prédiction utilisant les groupes plutôt que les variables individuelles peut être plus pertinente tant au niveau des performances prédictives que de l'interprétation. L'objectif de la thèse est de développer des méthodes par arbres adaptées aux variables groupées. Nous proposons deux approches qui utilisent la structure groupée des variables pour construire des arbres de décisions. La première méthode permet de construire des arbres binaires en classification. Une coupure est définie par le choix d'un groupe et d'une combinaison linéaire des variables du dit groupe. La seconde approche, qui peut être utilisée en régression et en classification, construit un arbre non-binaire dans lequel chaque coupure est un arbre binaire. Ces deux approches construisent un arbre maximal qui est ensuite élagué. Nous proposons pour cela deux stratégies d'élagage dont une est une généralisation du minimal cost-complexity pruning. Les arbres de décision étant instables, nous introduisons une méthode de forêts aléatoires pour variables groupées. Outre l'aspect prédiction, ces méthodes peuvent aussi être utilisées pour faire de la sélection de groupes grâce à l'introduction d'indices d'importance des groupes. Ce travail est complété par une partie indépendante dans laquelle nous nous plaçons dans un cadre d'apprentissage non supervisé. Nous introduisons un nouvel algorithme de clustering. Sous des hypothèses classiques, nous obtenons des vitesses de convergence pour le risque de clustering de l'algorithme proposé. / In many problems in supervised learning, inputs have a known and/or obvious group structure. In this context, elaborating a prediction rule that takes into account the group structure can be more relevant than using an approach based only on the individual variables for both prediction accuracy and interpretation. The goal of this thesis is to develop some tree-based methods adapted to grouped variables. Here, we propose two new tree-based approaches which use the group structure to build decision trees. The first approach allows to build binary decision trees for classification problems. A split of a node is defined according to the choice of both a splitting group and a linear combination of the inputs belonging to the splitting group. The second method, which can be used for prediction problems in both regression and classification, builds a non-binary tree in which each split is a binary tree. These two approaches build a maximal tree which is next pruned. To this end, we propose two pruning strategies, one of which is a generalization of the minimal cost-complexity pruning algorithm. Since decisions trees are known to be unstable, we introduce a method of random forests that deals with groups of inputs. In addition to the prediction purpose, these new methods can be also use to perform group variable selection thanks to the introduction of some measures of group importance, This thesis work is supplemented by an independent part in which we consider the unsupervised framework. We introduce a new clustering algorithm. Under some classical regularity and sparsity assumptions, we obtain the rate of convergence of the clustering risk for the proposed alqorithm.

Semi-supervised learning (SSL) is a branch of machine learning focusing on using not only labeled data samples, but also unlabeled ones, in an effort to decrease the need for labeled data and thus allow using machine learning even when labeling large amounts of data would be too costly. Despite its quick development in the recent years, there are still issues left to be solved before it can be broadly deployed in practice. One of those issues is class distribution mismatch. It arises when the unlabeled data contains samples not belonging to the classes present in the labeled data. This confuses the training and can even lead to getting a classifier performing worse than a classifier trained on the available data in purely supervised fashion. We designed a filtration method called Unfavorable Data Filtering (UDF) which extracts important features from the data and then uses a similarity-based filter to filter the irrelevant data out according to those features. The filtering happens before any of the SSL training takes places, making UDF usable with any SSL algorithm. To judge its effectiveness, we performed many experiments, mainly on the CIFAR-10 dataset. We found out that UDF is capable of significantly improving the resulting accuracy when compared to not filtering the data, identified basic guidelines...