Autor: Mario Graff
In the Genetic Programming (GP) community there has been a great interest in developing semantic genetic operators. These type of operators use information of the phenotype to create ospring. The most recent approaches of semantic GP include the GP framework based on the alignment of error space, the geometric semantic genetic operators, and backpropagation genetic operators. Our contribution proposes two semantic operators based on projections in the phenotype space. The proposed operators have the characteristic, by construction, that the ospring's tness is as at least as good as the tness of the best parent; using as tness the euclidean distance. The semantic operators proposed increment the learning capabilities of GP. These operators are compared against a traditional GP and Geometric Semantic GP in the Human oral bioavailability regression problem and 13 classication problems. The results show that a GP system with our novel semantic operators has the best performance in the training phase in all the problems tested.
El análisis de redes sociales para el estudio de mercados financieros se ha vuelto un tema de investigación y desarrollo de herramientas que permite a los agentes financieros usar las opiniones de la gente para aumentar la precisión en las predicciones de mercado. Nuestra investigación se enfoca en la predicción de la tendencia de índices financieros usando la minería de opinión, basado en el análisis de blogs especializados en finanzas para el idioma inglés. Los comenta-rios vertidos en estos blogs son clasificados en términos de su opinión respecto a la tendencia de mercado (a la alza, estable o a la baja). Se evalúan distintas téc-nicas de aprendizaje computacional y minería de textos para la clasificación de los comentarios realizados durante un periodo de tres meses. Los resultados ob-tenidos muestran que este análisis puede ser incorporado como un factor en la toma de decisión de los agentes financieros y mejorar la precisión de sus proyec-ciones.
This paper describes a novel approach to learning term-weighting schemes (TWSs) in the context of text classification. In text mining a TWS determines the way in which documents will be represented in a vector space model, before applying a classifier. Whereas acceptable performance has been obtained with standard TWSs (e.g., Boolean and term-frequency schemes), the definition of TWSs has been traditionally an art. Further, it is still a difficult task to determine what is the best TWS for a particular problem and it is not clear yet, whether better schemes, than those currently available, can be generated by combining known TWS. We propose in this article a genetic program that aims at learning effective TWSs that can improve the performance of current schemes in text classification. The genetic program learns how to combine a set of basic units to give rise to discriminative TWSs. We report an extensive experimental study comprising data sets from thematic and non-thematic text classification as well as from image classification. Our study shows the validity of the proposed method; in fact, we show that TWSs learned with the genetic program outperform traditional schemes and other TWSs proposed in recent works. Further, we show that TWSs learned from a specific domain can be effectively used for other tasks.
Prototype generation (PG) methods aim to find a subset of instances taken from a large training data set, in such a way that classification performance (commonly, using a 1NN classifier) when using prototypes is equal or better than that obtained when using the original training set. Several PG methods have been proposed so far, most of them consider a small subset of training instances as initial prototypes and modify them trying to maximize the classification performance on the whole training set. Although some of these methods have obtained acceptable results, training instances may be under-exploited, because most of the times they are only used to guide the search process. This paper introduces a PG method based on genetic programming in which many training samples are combined through arithmetic operators to build highly effective prototypes. The genetic program aims to generate prototypes that maximize an estimate of the generalization performance of an 1NN classifier. Experimental results are reported on benchmark data to assess PG methods. Several aspects of the genetic program are evaluated and compared to many alternative PG methods. The empirical assessment shows the effectiveness of the proposed approach outperforming most of the state of the art PG techniques when using both small and large data sets. Better results were obtained for data sets with numeric attributes only, although the performance of the proposed technique on mixed data was very competitive as well.
Proximity searching is the problem of retrieving,from agiven data base,those objects closest to aquery.To avoid exhaustive searching, data structures called indexes are builton the data base prior to serving queries.The curse of dimensionality is awell-known problem
for indexes: in spaces with sufficiently concentrated distance histograms,no index out performs anexhaustive scan of the data base.
Quality tests applied to hydraulic concrete such as compressive, tension, and bending strength are used to guarantee proper
characteristics ofmaterials. All these assessments are performed by destructive tests (DTs). The trend is to carry out quality analysis
using nondestructive tests (NDTs) as has been widely used for decades.This paper proposes a framework for predicting concrete
compressive strength and modulus of rupture by combining data from four NDTs: electrical resistivity, ultrasonic pulse velocity,
resonant frequency, and hammer test rebound withDTs data.Themodel, determined fromthemultiple linear regression technique,
produces accurate indicators predictions and categorizes the importance of each NDT estimate. However, the model is identified
fromall the possible linear combinations of the available NDT, and it was selected using a cross-validation technique. Furthermore,
the generality of the model was assessed by comparing results from additional specimens fabricated afterwards.
In recent years, a variety of semantic operators have been successfully developed to improve the performance of GP. This work presents a new semantic operator based on the semantic crossover based on the partial derivative error. The operator presented here uses the information of the second partial derivative to choose a crossover point in the second parent. The results show an improvement with respect to previous semantic operator.