Experience generalization for multi-agent reinforcement learning

On-line learning methods have been applied successfully in multi-agent systems to achieve coordination among agents. Learning in multi-agent systems implies in a non-stationary scenario perceived by the agents, since the behavior of other agents may change as they simultaneously learn how to improve their actions. Non-stationary scenarios can be modeled as Markov Games, which can be solved using the Minimax-Q algorithm a combination of Q-learning (a Reinforcement Learning (RL) algorithm which directly learns an optimal control policy) and the Minimax algorithm. However, finding optimal control policies using any RL algorithm (Q-learning and Minimax-Q included) can be very time consuming. Trying to improve the learning time of Q-learning, we considered the QS-algorithm. in which a single experience can update more than a single action value by using a spreading function. In this paper, we contribute a Minimax-QS algorithm which combines the Minimax-Q algorithm and the QS-algorithm. We conduct a series of empirical evaluation of the algorithm in a simplified simulator of the soccer domain. We show that even using a very simple domain-dependent spreading function, the performance of the learning algorithm can be improved.

Identification of dementia cases in the community: A Brazilian experience

OBJECTIVES: To describe the application in Brazil of a simple, low-cost procedure, developed in India by the 10/66 Dementia Research Group, for the identification of dementia cases in the community.DESIGN: Community-based dementia case-finding method.SETTING: Piraju, São Paulo, Brazil.PARTICIPANTS: Twenty-five community health workers were trained to identify dementia cases in 2,222 people aged 65 and older in Piraju, a Brazilian town with 27,871 inhabitants.MEAUREMENTS: After the training, the health workers prepared a list of possible cases that afterward an experienced psychiatrist clinically evaluated, according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), criteria and the Clinical Dementia Rating.RESULTS: of the 72 cases that were clinically assessed, 45 met the DSM-IV diagnostic criteria for dementia. Therefore, the positive predictive value of this case finding method was 62.5%; the estimated frequency of dementia was 2%. Most of the confirmed cases met clinical criteria for Alzheimer's disease and vascular dementia.CONCLUSION: This simple method was appropriate to identify cases of dementia in the general population and can possibly be extended to other developing countries with limited resources to be applied in health programs.

Neonatal meningitis according to the microbiological diagnosis: a decade of experience in a tertiary center

The aim of this study was to evaluate the incidence of and mortality due to meningitis and compare data according to microbiological diagnosis. This was a ten-year retrospective study conducted at a neonatal intensive care unit (NICU). Newborns with meningitis confirmed by positive CSF culture were included; those with congenital infection or malformations that made lumbar puncture impossible were excluded. The variables investigated were birth weight, gestational and postnatal age, procedures, hematological and CSF parameters, and complications. Parametric and non-parametric tests were used (statistical value p<0.05). The incidence of meningitis was 0.6% and mortality was 27%. of the 22 cases, 59% involved Gram-negative bacteria; 36% Gram-positive and 5% fungi. The groups did not differ in relation to birth weight, gestational and postnatal age, procedures or hematological and CSF parameters. Sepsis, convulsions and deaths were frequent in both groups, without statistical difference. Gram-negative cases showed abscesses and higher frequency of ventriculitis and hydrocephaly. Meningitis was infrequent, but presented high mortality and frequent complications.

An integrated approach to identification and conservation of medicinal plants in the tropical forest - A Brazilian experience

An integrated and interdisciplinary research programme with native medicinal plants from tropical forests has been performed in order to obtain new forest products for sustainable use in regional markets vis-à-vis ecosystem conservation. For the success of this programme ethnopharmacological studies are very important with respect to (i) identification of useful plants including medicinal and aromatic species; (ii) recuperation and preservation of traditional knowledge about native plants; and (iii) identification of potential plants with economic value. The plants are selected with a view to evaluate efficacy and safety (pharmacological and toxicological studies), and phytochemical profile and quality control (phytochemical and chromatographic characterization). These studies are very important to add value to plant products and also to mitigate unscrupulous exploitation of medicinal plants by local communities, since multiple use of plants represents an excellent strategy for sustaining the tropical ecosystem through ex situ and in situ conservation. Thus, conservation of tropical resources is possible in conjunction with improvements in the quality of life of the traditional communities and production of new products with therapeutic, cosmetic and 'cosmeceutic' value. © NIAB 2005.