Surrogate-Assisted and Filter-Based Multiobjective Evolutionary Feature Selection for Deep Learning.

Feature selection (FS) for deep learning prediction models is a difficult topic for researchers to tackle. Most of the approaches proposed in the literature consist of embedded methods through the use of hidden layers added to the neural network architecture that modify the weights of the units associated with each input attribute so that the worst attributes have less weight in the learning process. Other approaches used for deep learning are filter methods, which are independent of the learning algorithm, which can limit the precision of the prediction model. Wrapper methods are impractical with deep learning due to their high computational cost. In this article, we propose new attribute subset evaluation FS methods for deep learning of the wrapper, filter and wrapper-filter hybrid types, where multiobjective and many-objective evolutionary algorithms are used as search strategies. A novel surrogate-assisted approach is used to reduce the high computational cost of the wrapper-type objective function, while the filter-type objective functions are based on correlation and an adaptation of the reliefF algorithm. The proposed techniques have been applied in a time series forecasting problem of air quality in the Spanish south-east and an indoor temperature forecasting problem in a domotic house, with promising results compared to other FS techniques used in the literature.

Antioxidant activity of phenolic compounds added to a functional emulsion containing omega-3 fatty acids and plant sterol esters.

The effect of eleven compounds extracted from red propolis on the oxidative stability of a functional emulsion was evaluated. Emulsions prepared with Echium oil as omega 3 (ω-3 FA) source, containing 1.63 g/100mL of α-linolenic acid (ALA), 0.73 g/100 mL of stearidonic acid (SDA) and 0.65 g/100mL of plant sterol esters (PSE) were prepared without or with phenolic compounds (vanillic acid, caffeic acid, trans-cinnamic acid, 2,4-dihydroxycinnamic acid, p-coumaric acid, quercetin, trans-ferulic acid, trans,trans-farnesol, rutin, gallic acid or sinapic acid). tert-Butylhydroquinone and a mixture containing ascorbic acid and FeSO4 were applied as negative and positive controls of the oxidation. Hydroperoxide, thiobarbituric acid reactive substances (TBARS), malondialdehyde and phytosterol oxidation products (POPs) were evaluated as oxidative markers. Based on hydroperoxide and TBARS analysis, sinapic acid and rutin (200 ppm) showed the same antioxidant activity than TBHQ, representing a potential alternative as natural antioxidant to be applied in a functional emulsion containing ω-3 FA and PSE.

Comparison of the antinociceptive effect of celecoxib, diclofenac and resveratrol in the formalin test.

The peripheral antinociceptive effect of the selective COX-2 inhibitor celecoxib in the formalin-induced inflammatory pain was compared with that of resveratrol (COX-1 inhibitor) and diclofenac (non-selective COX inhibitor). Rats received local pretreatment with saline, celecoxib, diclofenac or resveratrol followed by 50 microl of either 1% or 5% formalin. Peripheral administration of celecoxib did not produce antinociception at either formalin concentration. In contrast, diclofenac and resveratrol produced a dose-dependent antinociceptive effect in the second phase of both 1% and 5% formalin test. The peripheral antinociception produced by diclofenac or resveratrol was due to a local action, as drug administration in the contralateral paw was ineffective. Results indicate that the selective COX-2 inhibitor celecoxib does not produce peripheral antinociception in formalin-induced inflammatory pain. In contrast, selective COX-1 and non-selective COX inhibitors (resveratrol and diclofenac, respectively) are effective drugs in this model of pain.

Híperplasia adrenal congénita

Se reportan 5 casos de hiperplasia adrenal congénita....(AU)