Raúl P. Rentería

Time-series forecasting through wavelets transformation and a mixture of expert models

Abstract This paper describes a system formed by a mixture of expert models (MEM) for time-series forecasting. We deal with several different competing models, such as partial least squares, K-nearest neighbours and carbon copy. The input space, after changing its base using the Haar wavelets transform, is partitioned into disjoint regions by a clustering algorithm. For each region, a benchmark is performed among the different competing models aiming at selecting the most adequate one. MEM has improved the forecast performance when compared with the single models as experimentally demonstrated through two different time series: laser data and exchange rate data.

Portuguese Part-of-Speech Tagging Using Entropy Guided Transformation Learning

Entropy Guided Transformation Learning (ETL) is a new machine learning strategy that combines the advantages of Decision Trees (DT) and Transformation Based Learning (TBL). In this work, we apply the ETL framework to Portuguese Part-of-Speech Taggging. We use two different corpora: Mac-Morpho and Tycho Brahae. ETL achieves the best results reported so far for Machine Learning based POS tagging of both corpora. ETL provides a new training strategy that accelerates transformation learning. For the Mac-Morpho corpus this corresponds to a factor of three speedup. ETL shows accuracies of 96.75% and 96.64% for Mac-Morpho and Tycho Brahae, respectively.

A fast and simple method for extracting relevant content from news webpages

We propose NCE, an efficient algorithm to identify and extract relevant content from news webpages. We define relevant as the textual sections that more objectively describe the main event in the article. This includes the title and the main body section, and excludes comments about the story and presentation elements. Our experiments suggest that NCE is competitive, in terms of extraction quality, with the best methods available in the literature. It achieves F1 = 90.7% in our test corpus containing 324 news webpages from 22 sites. The main advantages of our method are its simplicity and its computational performance. It is at least an order of magnitude faster than methods that use visual features. This characteristic is very suitable for applications that process a large number of pages.

A machine learning approach to the identification of appositives

Appositives are structures composed by semantically related noun phrases. In Natural Language Processing, the identification of appositives contributes to the building of semantic lexicons, noun phrase coreference resolution and information extraction from texts. In this paper, we present an appositive identifier for the Portuguese language. We describe experimental results obtained by applying two machine learning techniques: Transformation-based learning (TBL) and Hidden Markov Models (HMM). The results obtained with these two techniques are compared with that of a full syntactic parser, PALAVRAS. The TBL-based system outperformed the other methods. This suggests that a machine learning approach can be beneficial for appositive identification, and also that TBL performs well for this language task.

DPLS and PPLS: two PLS algorithms for large data sets

Two enhancements to the PLS regression algorithm are presented. The first, direct PLS (DPLS), offers a direct approximate formulation for the calculation of the required eigenvectors when dealing with more than one dependent variable. The second enhancement is parallel PLS (PPLS), a parallel version of the PLS algorithm restricted to the case of only one dependent variable for the regression model. In the experiments, DPLS shows a 40% faster running time, while the PPLS produces a speedup of 3 for the first four machines in a computer cluster architecture.

Semi-supervised learning for portuguese noun phrase extraction

Semi-supervised learning is frequently used when we have a small labeled training set but a large set of unlabeled samples. In this paper, we combine Hidden Markov Models and Transformation Based Learning in a semi-supervised learning approach. Self-training and Co-training are the two semi-supervised techniques that we apply to our scheme in order to classify Portuguese noun phrases. Our main goal here is to show that we can achieve effective noun phrase extraction using fewer tagged examples by applying a semi-supervised technique. Our models show good improvement with a small labeled corpus and little with a large one.

Fast calculation of optimal strategies for searching with non-uniform costs

Proposes an algorithm for finding a binary search tree that minimizes the worst-case cost when the access costs are non-uniform and depend on the last accessed key. For this kind of problem, which is commonly found when accessing data stored on magnetic or optical disks, we present an algorithm that finds an optimal search strategy with an expected running time of O(n/sup 2/log n), under some reasonable assumptions on the cost matrix. It is worth mentioning that the best previous algorithm for this problem runs in /spl Theta/(n/sup 3/) time.

RelHunter: a machine learning method for relation extraction from text

We propose RelHunter, a machine learning-based method for the extraction of structured information from text. RelHunter’s key idea is to model the target structures as a relation over entities. Hence, the modeling effort is reduced to the identification of entities and the generation of a candidate relation, which are simpler problems than the original one. RelHunter fits a very broad spectrum of complex computational linguistic problems. We apply it to five tasks: phrase chunking, clause identification, hedge detection, quotation extraction, and dependency parsing. We compare RelHunter to token classification approaches through several computational experiments on seven multilingual corpora. RelHunter outperforms the token classification approaches by 2.14% on average. Moreover, we compare the derived systems against state-of-the-art systems for each corpus. Our systems achieve state-of-the-art performances for three corpora: Portuguese phrase chunking, Portuguese clause identification, and English quotation extraction. Additionally, the derived systems show good quality performance for the other four corpora.

Multi-Kernel Based PLS regression

MKPLS, a non-linear version of the Partial LeastSquares regression is presented. The non-linearity is introduced in the classical algorithm through the use of multiple kernel functions, thus providing an straightforward non-linear adaptation. MKPLS provides a multikernel based version for the PLS algorithm with a competitive modeling error. Experimental results show that the use of different kernels for the regression model enhances the predictive power when compared to a PLS regression based on only one function kernel.