Antioxidant compounds of Kielmeyera coriacea Mart. with α-amylase, lipase and advanced glycation end-product inhibitory activities.

Chronic hyperglycemia and hyperlipidemia are associated with excessive formation of reactive oxygen species and advanced glycation end-products. The present study aimed to evaluate the potential in vitro antidiabetic properties of Kielmeyera coriacea inner bark. The main phytochemical compounds were identified by UHPLC-ESI/MSn and the ethanol extract and its fractions were used to evaluate their antioxidant and anti-glycation capacities, as well as their inhibitory potential against glycoside and lipid hydrolases activities. The polar fractions, especially the n-butanol fraction, had free radical scavenging and quenching properties (ORAC and FRAP values>1800 and 1000 µmol trolox eq/g, respectively, and DPPH IC50<4 µg/mL), and inhibited ROS production (p < 0.01), lipid peroxidation (p < 0.001), glycation (IC50 ~ 10 µg/mL in the BSA-fructose assay; IC50 ~ 200 µg/mL in the BSA-methylglyoxal and arginine-methylglyoxal assays), α-amylase (IC50<0.1 µg/mL) and lipase (IC50<5 µg/mL), with no cytotoxicity. Biomolecules well-known as potent antioxidants were identified for the first time in the inner bark of K. coriacea, such as protocatechuic acid, epicatechin and procyanidins A, B and C. Together, our results support the antioxidant, anti-glycation and glycoside and lipid hydrolases inhibitory properties of the inner bark of K. coriacea, a species found in the Brazilian savanna, which makes it especially useful to combat oxidative stress and hyperglycemia and hyperlipidemia.

Exploiting Spatio-Temporal Structure With Recurrent Winner-Take-All Networks.

We propose a convolutional recurrent neural network (ConvRNNs), with winner-take-all (WTA) dropout for high-dimensional unsupervised feature learning in multidimensional time series. We apply the proposed method for object recognition using temporal context in videos and obtain better results than comparable methods in the literature, including the deep predictive coding networks (DPCNs) previously proposed by Chalasani and Principe. Our contributions can be summarized as a scalable reinterpretation of the DPCNs trained end-to-end with backpropagation through time, an extension of the previously proposed WTA autoencoders to sequences in time, and a new technique for initializing and regularizing ConvRNNs.

Joint optimization of algorithmic suites for EEG analysis.

Electroencephalogram (EEG) data analysis algorithms consist of multiple processing steps each with a number of free parameters. A joint optimization methodology can be used as a wrapper to fine-tune these parameters for the patient or application. This approach is inspired by deep learning neural network models, but differs because the processing layers for EEG are heterogeneous with different approaches used for processing space and time. Nonetheless, we treat the processing stages as a neural network and apply backpropagation to jointly optimize the parameters. This approach outperforms previous results on the BCI Competition II - dataset IV; additionally, it outperforms the common spatial patterns (CSP) algorithm on the BCI Competition III dataset IV. In addition, the optimized parameters in the architecture are still interpretable.