LIAAD

Permanent URI for this community

http://localhost:4000/handle/123456789/31
This service focuses its activity on decision support systems, with particular emphasis on data mining, forecasting, adaptive modeling and optimization techniques, with applications on marketing, finance, process scheduling, health, text information extraction, and many other areas

Browse

Browsing LIAAD by Title

  • Item
    2D-Interval Predictions for Time Series
    ( 2011) Luís Torgo ; Orlando Shigueo Junior
  • Item
    Accelerating Recommender Systems using GPUs
    ( 2015) André Valente Rodrigues ; Alípio Jorge ; Inês Dutra
    We describe GPU implementations of the matrix recommender algorithms CCD++ and ALS. We compare the processing time and predictive ability of the GPU implementations with existing multi- core versions of the same algorithms. Results on the GPU are better than the results of the multi- core versions (maximum speedup of 14.8).
  • Item
    Accelerating Recommender Systems using GPUs
    ( 2015) André Valente Rodrigues ; Alípio Jorge ; Inês Dutra
  • Item
    Active Learning from Video Streams in a Multi-Camera Scenario
    ( 2014) Samaneh Khoshrou ; Jaime Cardoso ; Luís Filipe Teixeira
    While video surveillance systems are spreading everywhere, extracting meaningful information from what they are recording is still prohibitively expensive. There is a major effort under way in order to make this process economical by including an intelligent software that eases the burden of the system. In this paper, we introduce an incremental learning framework to classify parallel data streams generated in a multi-camera surveillance scenario. The framework exploits active learning strategies in order to interact wisely with operators to address various problems that exist in such non-stationary environments, such as concept drift and concept evolution. If we look at the problem as mining parallel streams, the framework can address learning from uneven parallel streams applying a class-based ensemble, a problem that has not been addressed before. Favourable results indicate the success of the framework.
  • Item
    Active Mining of Parallel Video Streams
    ( 2014) Samaneh Khoshrou ; Jaime Cardoso ; Luís Filipe Teixeira
  • Item
    Acute Kidney Injury Detection: An Alarm System to Improve Early Treatment
    ( 2017) Ana Rita Nogueira ; Carlos Ferreira ; João Gama
  • Item
    Adaptive model rules from data streams
    ( 2013) Ezilda Duarte Almeida ; Carlos Ferreira ; João Gama
    Decision rules are one of the most expressive languages for machine learning. In this paper we present Adaptive Model Rules (AMRules), the first streaming rule learning algorithm for regression problems. In AMRules the antecedent of a rule is a conjunction of conditions on the attribute values, and the consequent is a linear combination of attribute values. Each rule uses a Page-Hinkley test to detect changes in the process generating data and react to changes by pruning the rule set. In the experimental section we report the results of AMRules on benchmark regression problems, and compare the performance of our system with other streaming regression algorithms. © 2013 Springer-Verlag.
  • Item
    Adaptive Model Rules From High-Speed Data Streams
    ( 2016) Duarte,J ; João Gama ; Bifet,A
  • Item
    An Agent-based Model of the Earth System & Climate Change
    ( 2016) Yassine Baghoussi ; Pedro Campos ; Rossetti,RJF
    Simulation is a computer-based experimentation tool suitable to determine the efficacy of a previously untried decision. In this paper, we present a model of climate change. The goal behind this project is to provide a test-bed to evaluate theories related to the Earth system so as to test and evaluate metrics such as greenhouse gases and climate change in general. The proposed approach is based on a multi-agent model which has as input a representation of nature and as output the changes that will occur on Earth within a given instant of time. Most views about climate change do not take into account the real severity of the subject matter; however, the present perspective is given in a way so as to make non-experts aware of the risks that are threatening life on Earth. Just recently, the general population has developed considerable sensitivity to these issues. One important contribution of this work is to use agent-based modeling and simulation as an instructional tool that will allow people to easily understand all aspects involved in the preservation of the environment in a more aware and responsible way.
  • Item
    Algorithm selection via meta-learning and sample-based active testing
    ( 2015) Salisu Mamman Abdulrhaman ; Pavel Brazdil ; Van Rijn,JN ; Vanschoren,J
    Identifying the best machine learning algorithm for a given problem continues to be an active area of research. In this paper we present a new method which exploits both meta-level information acquired in past experiments and active testing, an algorithm selection strategy. Active testing attempts to iteratively identify an algorithm whose performance will most likely exceed the performance of previously tried algorithms. The novel method described in this paper uses tests on smaller data sample to rank the most promising candidates, thus optimizing the schedule of experiments to be carried out. The experimental results show that this approach leads to considerably faster algorithm selection.
  • Item
  • Item
    ANALYSIS OF WELL-BEING IN OECD COUNTRIES THROUGH STATIS METHODOLOGY
    ( 2016) Rivadeneira,FJ ; Adelaide Figueiredo ; Figueiredo,FOS ; Carvajal,SM ; Rivadeneira,RA
    This paper presents the main concepts and results of a Master thesis in Data Analysis which aims to analyze the evolution of some developed countries and also of some emerging countries that are members of the Organisation for Economic Co-operation and Development (OECD) in what concerns some indicators or variables of well-being during the period 2011-2015, through the STATIS (Structuring Three-way data sets in Statistics) methodology. This methodology allows to analyze the presence of a common structure in several data tables obtained over time, to identify the differences and similarities along the period of time under study and according to well-being indicators included in the "Your Better Life Index" of the OECD, and to analyze the trajectories of the countries.
  • Item
    Analyzing the behavior dynamics of grain price indexes using Tucker tensor decomposition and spatio-temporal trajectories
    ( 2016) Correa,FE ; Márcia Barbosa Oliveira ; João Gama ; Correa,PLP ; Rady,J
    Agribusiness is an activity that generates huge amounts of temporal data. There are research centers that collect, store and create indexes of agricultural activities, providing multidimensional time series composed by years of data. In this paper, we are interested in studying the behavior of these time series, especially in what regards the evolution of agricultural price indexes over the years. We explore data mining techniques tailored to analyze temporal data, aiming to generate spatio-temporal trajectories of grains price indexes for six years of data. We propose the use of Tucker decomposition to both analyze the temporal patterns of these price indexes and map trajectories that represent their behavior over time in a concise and representative low-dimensional subspace. The case study presents an application of this methodology to real databases of price indexes of corn and soybeans in Brazil and the United States.
  • Item
    Anáiise de Variância Dual
    ( 2011) Adelaide Figueiredo
    Consideremos a abordagem dual da abordagem clássica de estatística multivariada em que os indivíduos estão fixos e as variáveis são escolhidas aleatoriamente de uma população de variáveis. Supomos k grupos de variáveis centradas e reduzidas e associamos a cada grupo uma distribuição de Watson. Para vermos se estes grupos são distintos testamos se as direções privilegiadas das distribuições de Watson diferem significativamente usando a análise de variância dual. Analisamos a potência deste teste para dois grupos e diferentes dimensões de esfera. Pretendemos aplicar esta abordagem a dados reais.
  • Item
    AND Parallelism for ILP: The APIS System
    ( 2014) Rui Camacho ; Ramos,R ; Nuno Fonseca
    Inductive Logic Programming (ILP) is a well known approach to Multi-Relational Data Mining. ILP systems may take a long time for analyzing the data mainly because the search (hypotheses) spaces are often very large and the evaluation of each hypothesis, which involves theorem proving, may be quite time consuming in some domains. To address these efficiency issues of ILP systems we propose the APIS (And ParallelISm for ILP) system that uses results from Logic Programming AND-parallelism. The approach enables the partition of the search space into sub-spaces of two kinds: sub-spaces where clause evaluation requires theorem proving; and sub-spaces where clause evaluation is performed quite efficiently without resorting to a theorem prover. We have also defined a new type of redundancy (Coverage-equivalent redundancy) that enables the prune of significant parts of the search space. The new type of pruning together with the partition of the hypothesis space considerably improved the performance of the APIS system. An empirical evaluation of the APIS system in standard ILP data sets shows considerable speedups without a lost of accuracy of the models constructed.
  • Item
    Anomaly detection from telecommunication data using three-way analysis
    ( 2012) Márcia Barbosa Oliveira ; João Gama ; Hadi Fanaee Tork
    So far, several supervised and unsupervised solutions have been provided for detecting failures in telecommunication networks. Among them, unsupervised approaches attracted more attention since no labeled data is required. Principal component analysis (PCA) is a wellknown unsupervised technique to solve this type of problem when data is organized in matrix form. However, PCA may fail to capture all the significant interactions established among di erent dimensions when applied to higher-order data. When dealing with three, instead of two dimensions, three-way factorization methods are more suitable since they are able to explicitly take into account the interactions among the three dimensions, without collapsing the raw data. Since an important property of telecommunication data is its temporal-sequential nature, the temporal dimension should be considered along with the other two dimensions in order to gain insights regarding its evolution. The aim of this paper is to demonstrate t
  • Item
    ANOSOV DIFFEOMORPHISMS
    ( 2013) João Paulo Almeida ; Fisher,AM ; Alberto Pinto ; Rand,DA
    We use Adler, Tresser and Worfolk decomposition of Anosov automorphisms to give an explicit construction of the stable and unstable C1+ self-renormalizable sequences.
  • Item
    Anosov Diffeomorphisms and -Tilings
    ( 2016) João Paulo Almeida ; Alberto Pinto
    We consider a toral Anosov automorphism G(gamma) : T-gamma --> T-gamma given by G(gamma) (x, y) = (ax + y, x) in the < v, w > base, where , a is an element of N\{1}, gamma = 1/(a + 1/(a + 1/...)), v = (gamma, 1) and w = (-1, gamma) in the canonical base of R-2 and T-gamma = R-2 / (vZ x wZ). We introduce the notion of gamma-tilings to prove the existence of a one-to-one correspondence between (i) marked smooth conjugacy classes of Anosov diffeomorphisms, with invariant measures absolutely continuous with respect to the Lebesgue measure, that are in the isotopy class of G(gamma); (ii) affine classes of gamma-tilings; and (iii) gamma-solenoid functions. Solenoid functions provide a parametrization of the infinite dimensional space of the mathematical objects described in these equivalences.
  • Item
    Application-specific traffic anomaly detection using universal background model
    ( 2015) Alizadeh,H ; Samaneh Khoshrou ; Zuquete,A
    This paper presents an application-specific intrusion detection framework in order to address the problem of detecting intrusions in individual applications when their traffic exhibits anomalies. The system is based on the assumption that authorized traffic analyzers have access to a trustworthy binding between network traffic and the source application responsible for it. Given traffic flows generated by individual genuine application, we exploit the GMM-UBM (Gaussian Mixture Model-Universal Background Model) method to build models for genuine applications, and thereby form our detection system. The system was evaluated on a public dataset collected from a real network. Favorable results indicate the success of the framework. Copyright © 2015 ACM.
  • Item
    An Approach to Extract Proper Implications Set from High-dimension Formal Contexts using Binary Decision Diagram
    ( 2018) Santos,P ; Neves,J ; Paula Raissa Silva ; Dias,SM ; Zárate,L ; Song,M ; 7134