In this paper, we present a new method for emotion recognition from facial expressions. The proposed algorithm concentrates on only two specific areas (eyes and mouth), reducing features and descriptors and focusing only on these areas. The algorithm extracts characteristics from these two regions of the face and, in a subsequent process, eliminates the less significant characteristics or those that introduce noise into the classifier. The system allows obtaining a reduced set of features to improve the performance of the classification. In the experiments carried out, we obtain precisions of 99.56%. We evaluated the proposed algorithm on two benchmark datasets; we find that SVM consistently outperforms traditional machine learning techniques.

In the Software industry, big software projects are carried out with hundreds of developers. The fast change in technologies and development environments increase the complexity. Usually, there are project teams with a project leader. However, it is very difficult to know the profile of each developer. The development tasks also have their profile. Hence, it is necessary to assign each task to the most suitable developer. The erroneous assigning of tasks can cause delays and increase the project costs. Thus, a bad assigning of tasks can cause stress and low productivity. Therefore, we make a proposal to enhance tasks assignment to developers regarding the task and developer profiles. The task profile includes characteristics such as: knowledge, kind of task, complexity, experience, etc., in other aspects as codification: paradigm, programming language, version, etc. Using algorithms with similarity coefficients, we look for the best match to assign the tasks to developers. In this work, we used five techniques of similarity coefficient in order to find results and to recommend the best solution for this problem. We conclude that with the Sokal and Sneath technique we obtain better results to solve the problem.

In Micro and Small Companies for Software Development (MSCSD) of Mexico, the knowledge is generated in each software project. So, it is possible to transform tacit knowledge into explicit, to have some strategy and data storage device. Note that, when the knowledge exists only in the brains of developers there is the disadvantage that, when a developer leaves the organization, knowledge is lost. Therefore, it is possible to transform the tacit knowledge into explicit, if exist some strategy and data storage device. Developers need to generate and storage knowledge in any device and format to process, storage and exploit. In this paper, it has been called knowledge asset to an idea or solution processes or software development activities, which may be embodied as text, images, audio or video. A novel model for knowledge management is proposed that defines how must obtain, process, store and exploit knowledge resulting in the México’s MSCSD. The conclusion is that it is necessary to implement politics and processes in MSCSD to promote knowledge management and that must change the culture in organizations.

Support Vector Machines (SVM) have shown excellent generalization power in classification problems. However, on skewed data-sets, SVM learns a biased model that affects the classifier performance, which is severely damaged when the unbalanced ratio is very large. In this paper, a new external balancing method for applying SVM on skewed data sets is developed. In the first phase of the method, the separating hyperplane is computed. Support vectors are then used to generate the initial population of PSO algorithm, which is used to improve the population of artificial instances and to eliminate noise instances. Experimental results demonstrate the ability of the proposed method to improve the performance of SVM on imbalanced data-sets.

El problema del desbalance de clases puede producir un deterioro importante en la efectividad del clasificador, en particular con los patrones de las clases menos representadas. El desbalance en el conjunto de entrenamiento (CE) significa que una clase es representada por una gran cantidad de patrones mientras que otra es representada por muy pocos. Los estudios existentes se encuentran orientados principalmente a tratar problemas de dos clases, no obstante, un importante número de problemas reales se encuentran representados por múltiples clases, donde resulta más difícil su discriminación para el clasificador. El éxito de la Mezcla de Expertos (ME) se basa en el criterio de “divide y vencerás”. En su funcionamiento general, el problema es dividido en fragmentos más pequeños que serán estudiados por separado. De este modo, el modelo general es poco influenciado por las dificultades individuales de sus componentes. La idea principal del estudio aquí mostrado, es construir una Mezcla de expertos cuyos miembros serán entrenados en una parte del problema general y de este modo, mejorar el rendimiento del clasificador en el contexto de múltiples clases. Para este fin, se hace uso de los métodos conocidos como Error-correcting output codes (ECOC), que permiten realizar una codificación en parejas de clases el problema de estudio. Resultados experimentales sobre conjuntos de datos reales, muestran la viabilidad de la estrategia aquí propuesta.

El desarrollo de herramientas que faciliten la comunicacion de personas sordas es un reto de investigacion actual muy importante. Una linea de investigacion es el desarrollo de sistemas de vision con un gran poder de generalizacion. Obtener una buena precision de generalizacion requiere un conjunto de datos muy grande durante el entrenamiento, y el incremento de datos muchas veces solo anade informacion repetitiva y no representativa. En este articulo describimos el desarrollo de un sistema de reconocimiento de lenguaje de senas. El sistema propuesto permite identificar con una alta precision el Lenguaje de Senas Mexicano introduciendo datos muy representativos generados artificialmente, que permiten mejorar la capacidad de generalizacion del clasificador. Los resultados obtenidos muestran que el algoritmo propuesto mejora la precision de generalizacion de las SVM al utilizar la metodologia propuesta.

In this paper we present a new algorithm to speed up the training time of Support Vector Machines (SVM). SVM has some important properties like solid mathematical background and a better generalization capability than other machines like for example neural networks. On the other hand, the major drawback of SVM occurs in its training phase, which is computationally expensive and highly dependent on the size of input data set. The proposed algorithm uses a data filter to reduce the input data set to train a SVM. The data filter is based on an induction tree which effectively reduces the training data set for SVM, producing a very fast and high accuracy algorithm. According to the results, the algorithm produces results in a faster way than existing SVM implementations (SMO, LIBSVM and Simple-SVM) with similar accurateness.