Recommending media content based on machine learning methods

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Machine learning Gaussian short rate

Dissertação para obtenção do Grau de Doutor em Estatística e Gestão do Risco

Automated image tagging through tag propagation

Trabalho apresentado no âmbito do Mestrado em Engenharia Informática, como requisito parcial Para obtenção do grau de Mestre em Engenharia Informática

Context-aware personalization environment for mobile computing

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

New visualization model for large scale biosignals analysis

Benefits of long-term monitoring have drawn considerable attention in healthcare. Since the acquired data provides an important source of information to clinicians and researchers, the choice for long-term monitoring studies has become frequent. However, long-term monitoring can result in massive datasets, which makes the analysis of the acquired biosignals a challenge. In this case, visualization, which is a key point in signal analysis, presents several limitations and the annotations handling in which some machine learning algorithms depend on, turn out to be a complex task. In order to overcome these problems a novel web-based application for biosignals visualization and annotation in a fast and user friendly way was developed. This was possible through the study and implementation of a visualization model. The main process of this model, the visualization process, comprised the constitution of the domain problem, the abstraction design, the development of a multilevel visualization and the study and choice of the visualization techniques that better communicate the information carried by the data. In a second process, the visual encoding variables were the study target. Finally, the improved interaction exploration techniques were implemented where the annotation handling stands out. Three case studies are presented and discussed and a usability study supports the reliability of the implemented work.

Human activity recognition with accelerometry: novel time and frequency features

Human Activity Recognition systems require objective and reliable methods that can be used in the daily routine and must offer consistent results according with the performed activities. These systems are under development and offer objective and personalized support for several applications such as the healthcare area. This thesis aims to create a framework for human activities recognition based on accelerometry signals. Some new features and techniques inspired in the audio recognition methodology are introduced in this work, namely Log Scale Power Bandwidth and the Markov Models application. The Forward Feature Selection was adopted as the feature selection algorithm in order to improve the clustering performances and limit the computational demands. This method selects the most suitable set of features for activities recognition in accelerometry from a 423th dimensional feature vector. Several Machine Learning algorithms were applied to the used accelerometry databases – FCHA and PAMAP databases - and these showed promising results in activities recognition. The developed algorithm set constitutes a mighty contribution for the development of reliable evaluation methods of movement disorders for diagnosis and treatment applications.

Learning facial-expression models with crowdsourcing

The computational power is increasing day by day. Despite that, there are some tasks that are still difficult or even impossible for a computer to perform. For example, while identifying a facial expression is easy for a human, for a computer it is an area in development. To tackle this and similar issues, crowdsourcing has grown as a way to use human computation in a large scale. Crowdsourcing is a novel approach to collect labels in a fast and cheap manner, by sourcing the labels from the crowds. However, these labels lack reliability since annotators are not guaranteed to have any expertise in the field. This fact has led to a new research area where we must create or adapt annotation models to handle these weaklylabeled data. Current techniques explore the annotators’ expertise and the task difficulty as variables that influences labels’ correction. Other specific aspects are also considered by noisy-labels analysis techniques. The main contribution of this thesis is the process to collect reliable crowdsourcing labels for a facial expressions dataset. This process consists in two steps: first, we design our crowdsourcing tasks to collect annotators labels; next, we infer the true label from the collected labels by applying state-of-art crowdsourcing algorithms. At the same time, a facial expression dataset is created, containing 40.000 images and respective labels. At the end, we publish the resulting dataset.

Mining protein structure data

The principal topic of this work is the application of data mining techniques, in particular of machine learning, to the discovery of knowledge in a protein database. In the first chapter a general background is presented. Namely, in section 1.1 we overview the methodology of a Data Mining project and its main algorithms. In section 1.2 an introduction to the proteins and its supporting file formats is outlined. This chapter is concluded with section 1.3 which defines that main problem we pretend to address with this work: determine if an amino acid is exposed or buried in a protein, in a discrete way (i.e.: not continuous), for five exposition levels: 2%, 10%, 20%, 25% and 30%. In the second chapter, following closely the CRISP-DM methodology, whole the process of construction the database that supported this work is presented. Namely, it is described the process of loading data from the Protein Data Bank, DSSP and SCOP. Then an initial data exploration is performed and a simple prediction model (baseline) of the relative solvent accessibility of an amino acid is introduced. It is also introduced the Data Mining Table Creator, a program developed to produce the data mining tables required for this problem. In the third chapter the results obtained are analyzed with statistical significance tests. Initially the several used classifiers (Neural Networks, C5.0, CART and Chaid) are compared and it is concluded that C5.0 is the most suitable for the problem at stake. It is also compared the influence of parameters like the amino acid information level, the amino acid window size and the SCOP class type in the accuracy of the predictive models. The fourth chapter starts with a brief revision of the literature about amino acid relative solvent accessibility. Then, we overview the main results achieved and finally discuss about possible future work. The fifth and last chapter consists of appendices. Appendix A has the schema of the database that supported this thesis. Appendix B has a set of tables with additional information. Appendix C describes the software provided in the DVD accompanying this thesis that allows the reconstruction of the present work.

Algorithm for the Parkinson’s disease behavioural models characterization using a biosensor

Dissertação para obtenção do Grau de Mestre em Engenharia Biomédica

Adapter module for self-learning production systems

Dissertação para obtenção do Grau de Mestre em Engenharia Electrotécnica, Sistemas e Computadores

Automatic musical instrument recognition for multimedia indexing

Trabalho apresentado no âmbito do Mestrado em Engenharia Informática, como requisito parcial para obtenção do grau de Mestre em Engenharia Informática

Classificação de tumores de cancro na mama através de radar de banda ultra-larga de microondas

Dissertação para obtenção do Grau de Mestre em Engenharia Biomédica

Programação genética

A Programação Genética (PG) é uma técnica de Aprendizagem de Máquina (Machine Learning (ML)) aplicada em problemas de otimização onde pretende-se achar a melhor solução num conjunto de possíveis soluções. A PG faz parte do paradigma conhecido por Computação Evolucionária (CE) que tem como inspiração à teoria da evolução natural das espécies para orientar a pesquisa das soluções. Neste trabalho, é avaliada a performance da PG no problema de previsão de parâmetros farmacocinéticos utilizados no processo de desenvolvimento de fármacos. Este é um problema de otimização onde, dado um conjunto de descritores moleculares de fármacos e os valores correspondentes dos parâmetros farmacocinéticos ou de sua atividade molecular, utiliza-se a PG para construir uma função matemática que estima tais valores. Para tal, foram utilizados dados de fármacos com os valores conhecidos de alguns parâmetros farmacocinéticos. Para avaliar o desempenho da PG na resolução do problema em questão, foram implementados diferentes modelos de PG com diferentes funções de fitness e configurações. Os resultados obtidos pelos diferentes modelos foram comparados com os resultados atualmente publicados na literatura e os mesmos confirmam que a PG é uma técnica promissora do ponto de vista da precisão das soluções encontradas, da capacidade de generalização e da correlação entre os valores previstos e os valores reais.

Botnet cluster identification

Botnets are a group of computers infected with a specific sub-set of a malware family and controlled by one individual, called botmaster. This kind of networks are used not only, but also for virtual extorsion, spam campaigns and identity theft. They implement different types of evasion techniques that make it harder for one to group and detect botnet traffic. This thesis introduces one methodology, called CONDENSER, that outputs clusters through a self-organizing map and that identify domain names generated by an unknown pseudo-random seed that is known by the botnet herder(s). Aditionally DNS Crawler is proposed, this system saves historic DNS data for fast-flux and double fastflux detection, and is used to identify live C&Cs IPs used by real botnets. A program, called CHEWER, was developed to automate the calculation of the SVM parameters and features that better perform against the available domain names associated with DGAs. CONDENSER and DNS Crawler were developed with scalability in mind so the detection of fast-flux and double fast-flux networks become faster. We used a SVM for the DGA classififer, selecting a total of 11 attributes and achieving a Precision of 77,9% and a F-Measure of 83,2%. The feature selection method identified the 3 most significant attributes of the total set of attributes. For clustering, a Self-Organizing Map was used on a total of 81 attributes. The conclusions of this thesis were accepted in Botconf through a submited article. Botconf is known conferênce for research, mitigation and discovery of botnets tailled for the industry, where is presented current work and research. This conference is known for having security and anti-virus companies, law enforcement agencies and researchers.

A era de um mercado social

O crescimento e a expansão das redes sociais trouxe novas formas de interação entre os seres humanos que se repercutem na vida real. Os textos partilhados nas redes sociais e as interações resultantes de todas as atividades virtuais têm vindo a ganhar um grande impacto no quotidiano da sociedade e no âmbito económico e financeiro, as redes sociais tem sido alvo de diversos estudos, particularmente em termos de previsão e descrição do mercado acionista (Zhang, Fuehres, & Gloor, 2011) (Bollen, Mao & Zheng, 2010). Nesta investigação percebemos se o sentimento do Twitter, rede social de microblogging, se relaciona diretamente com o mercado acionista, querendo assim compreender qual o impacto das redes sociais no mercado financeiro. Tentámos assim relacionar duas dimensões, social e financeira, de forma a conseguirmos compreender de que forma poderemos utilizar os valores de uma para prever a outra. É um tópico especialmente interessante para empresas e investidores na medida em que se tenta compreender se o que se diz de determinada empresa no Twitter pode ter relação com o valor de mercado dessa empresa. Usámos duas técnicas de análise de sentimentos, uma de comparação léxica de palavras e outra de machine learning para compreender qual das duas tinha uma melhor precisão na classificação dos tweets em três atributos, positivo, negativo ou neutro. O modelo de machine learning foi o modelo escolhido e relacionámos esses dados com os dados do mercado acionista através de um teste de causalidade de Granger. Descobrimos que para certas empresas existe uma relação entre as duas variáveis, sentimento do Twitter e alteração da posição da ação entre dois períodos de tempo no mercado acionista, esta última variável estando dependente da dimensão temporal em que agrupamos o nosso sentimento do Twitter. Este estudo pretendeu assim dar seguimento ao trabalho desenvolvido por Bollen, Mao e Zheng (2010) que descobriram que uma dimensão de sentimento (calma) consegue ser usada para prever a direção das ações do mercado acionista, apesar de terem rejeitado que o sentimento geral (positivo, negativo ou neutro) não se relacionava de modo global com o mercado acionista. No seu trabalho compararam o sentimento de todos os tweets de um determinado período sem exclusão com o índice geral de ações no mercado enquanto a metodologia adotada nesta investigação foi realizada por empresa e apenas nos interessaram tweets que se relacionavam com aquela empresa em específico. Com esta diferença obtemos resultados diferentes e certas empresas demonstravam que existia relação entre várias combinações, principalmente para empresas tecnológicas. Testamos o agrupamento do sentimento do Twitter em 3 minutos, 1 hora e 1 dia, sendo que certas empresas só demonstravam relação quando aumentávamos a nossa dimensão temporal. Isto leva-nos a querer que o sentimento geral da empresa, e se a mesma for uma empresa tecnológica, está ligado ao mercado acionista estando condicionada esta relação à dimensão temporal que possamos estar a analisar.