Rui Leite

Predicting relative performance of classifiers from samples

This paper is concerned with the problem of predicting relative performance of classification algorithms. It focusses on methods that use results on small samples and discusses the shortcomings of previous approaches. A new variant is proposed that exploits, as some previous approaches, meta-learning. The method requires that experiments be conducted on few samples. The information gathered is used to identify the nearest learning curve for which the sampling procedure was carried out fully. This in turn permits to generate a prediction regards the relative performance of algorithms. Experimental evaluation shows that the method competes well with previous approaches and provides quite good and practical solution to this problem.

An iterative process for building learning curves and predicting relative performance of classifiers

This paper concerns the problem of predicting the relative performance of classification algorithms. Our approach requires that experiments are conducted on small samples. The information gathered is used to identify the nearest learning curve for which the sampling procedure was fully carried out. This allows the generation of a prediction regarding the relative performance of the algorithms. The method automatically establishes how many samples are needed and their sizes. This is done iteratively by taking into account the results of all previous experiments - both on other datasets and on the new dataset obtained so far. Experimental evaluation has shown that the method achieves better performance than previous approaches.

Depression and social functioning among HIV-infected and uninfected persons in South Africa

Depression and other health problems are common co-morbidities among persons living with human immunodeficiency virus infection/acquired immunodeficiency syndrome (HIV/AIDS). The aim of this study was to investigate depression, health status, and substance use in relation to HIV-infected and uninfected individuals in South Africa. Using a cross-sectional case-control design, we compared depression, physical health, mental health, problem alcohol use, and tobacco use in a sample of HIV-infected (N = 143) and HIV-uninfected (N = 199) respondents who had known their HIV status for two months. We found that depression was higher, and physical health and mental health were lower in HIV-positive than HIV-negative individuals. Poor physical health also moderated the effect of HIV infection on depression; HIV-positive individuals were significantly more depressed than HIV-negative controls, but only when general physical health was also poor. We did not find an association between alcohol or tobacco use and HIV status. These results suggest the importance of incorporating the management of psychological health in the treatment of HIV.

Determining the Best Classification Algorithm with Recourse to Sampling and Metalearning

Currently many classification algorithms exist and no algorithm exists that would outperform all the others. Therefore it is of interest to determine which classification algorithm is the best one for a given task. Although direct comparisons can be made for any given problem using a cross-validation evaluation, it is desirable to avoid this, as the computational costs are significant. We describe a method which relies on relatively fast pairwise comparisons involving two algorithms. This method is based on a previous work and exploits sampling landmarks, that is information about learning curves besides classical data characteristics. One key feature of this method is an iterative procedure for extending the series of experiments used to gather new information in the form of sampling landmarks. Metalearning plays also a vital role. The comparisons between various pairs of algorithm are repeated and the result is represented in the form of a partially ordered ranking. Evaluation is done by comparing the partial order of algorithm that has been predicted to the partial order representing the supposedly correct result. The results of our analysis show that the method has good performance and could be of help in practical applications.

Do Customers Flee From HIV? A Survey of HIV Stigma and Its Potential Economic Consequences on Small Businesses in Tshwane (Pretoria), South Africa

HIV stigma and discrimination affect care-seeking behavior and may also affect entrepreneurial activity. We interview 2382 individuals in Pretoria, South Africa, and show that respondents believe that businesses with known HIV+ workers may lose up to half of their customers, although the impact depends on the type of business. Survey respondents’ fear of getting HIV from consuming everyday products sold by the business—despite a real infection risk of zero—was a major factor driving perceived decline in customers, especially among food businesses. Respondents’ perceptions of the decline in overall life satisfaction when one gets sick from HIV and the respondent’s dislike of people with HIV were also important predictors of potential customer exit. We suggest policy mechanisms that could improve the earnings potential of HIV+ workers: reducing public health scare tactics that exacerbate irrational fear of HIV infection risk and enriching public health education about HIV and ARVs to improve perceptions about people with HIV.ResumenEl efecto discriminatorio y estigma hacia el VIH puede afectar a la búsqueda de ayuda y a la actividad emprendedora. Entrevistamos a 2.382 personas en Pretoria, Sudáfrica. Los encuestados creen que los empresarios con trabajadores con VIH+ conocido pueden perder hasta la mitad de sus clientes, aunque el impacto depende del tipo de negocio. El que los encuestados teman adquirir VIH por el consumo diarios de productos de un negocio, a pesar del riesgo de infección ser nulo, es un factor importante para el declive en el número de clientes, especialmente en el sector alimentario. Otros de los factores que predicen la pérdida de clientes son la percepción de los encuestados de que la satisfacción general de la vida disminuye cuando uno se enferma a causa del VIH y la aversión a las personas con VIH. Sugerimos mecanismos políticos que puedan mejorar el potencial de ingresos de los trabajadores con VIH+: tácticas de reducción de alarma sanitaria que exacerba el miedo irracional de riesgo de infección de VIH y enriquecer la educación de la sociedad en salud sobre el VIH y los ARV para mejorar las percepciones sobre las personas con VIH.

Improving progressive sampling via meta-learning on learning curves

This paper describes a method that can be seen as an improvement of the standard progressive sampling. The standard method uses samples of data of increasing size until accuracy of the learned concept cannot be further improved. The issue we have addressed here is how to avoid using some of the samples in this progression. The paper presents a method for predicting the stopping point using a meta-learning approach. The method requires just four iterations of the progressive sampling. The information gathered is used to identify the nearest learning curves, for which the sampling procedure was carried out fully. This in turn permits to generate the prediction regards the stopping point. Experimental evaluation shows that the method can lead to significant savings of time without significant losses of accuracy.

Improving Progressive Sampling via Meta-learning

We present a method that can be seen as an improvement of standard progressive sampling method. The method exploits information concerning performance of a given algorithm on past datasets, which is used to generate predictions of the stopping point. Experimental evaluation shows that the method can lead to significant time savings without significant losses in accuracy.

An Agent-Based Model for Detection in Economic Networks.

The economic impact of fraud is wide and fraud can be a critical problem when the prevention procedures are not robust. In this paper we create a model to detect fraudulent transactions, and then use a classification algorithm to assess if the agent is fraud prone or not. The model (BOND) is based on the analytics of an economic network of agents of three types: individuals, businesses and financial intermediaries. From the dataset of transactions, a sliding window of rows previously aggregated per agent has been used and machine learning (classification) algorithms have been applied. Results show that it is possible to predict the behavior of agents, based on previous transactions.