Carbon fiber was surface-functionalized by a facile dopamine self-polymerization method to improve the interfacial interaction with maleic anhydride grafted polypropylene modified PP.

Lipase production and cell growth of Serratia marcescens ECU1010 were optimized in shake flasks, with lipase production being enhanced 9.5-fold (4,780 U/l) compared with the initial activity (500 U/l). Optimal carbon and nitrogen sources were Tween-80 and peptone, and the optimal ratio of Tween-80 to peptone was 1:3. The optimized cultivation conditions were 25 degrees C and pH 6.5. Lipase activity, particularly specific activity, could be improved by decreasing the cultivation temperature from 35 to 25 degrees C. Enzyme stability was significantly improved by simple immobilization with synthetic adsorption resin no. 8244. After five reaction cycles, enzyme activity decreased only very slightly, while enantioselectivity of the preparation remained constant, and the ees (enantiomeric excess of the remaining substrate) achieved in all cases was higher than 97%. The resin-8244-lipase preparation can be used for efficient enantioselective hydrolysis of trans-3-(4'-methoxyphenyl)glycidic acid methyl ester [(+/-)-MPGM], a key intermediate in the synthesis of Diltiazem.

Modern adhesives usually need to have integrated functions to match the growing application needs, such as stimulus responsiveness, reversible adhesion ability, self-healing function, and adhesion ability under humid environment. Compared...

No abstract

No abstract

No abstract

Despite the recent prevalence of keyword spotting (KWS) in smart-home, open-vocabulary KWS remains a keen but unmet need among the users. In this paper, we propose an RNN Transducer (RNN-T) based keyword spotting system with a constrained attention mechanism biasing module that biases the RNN-T model towards a specific keyword of interest. The atonal syllables are adopted as the modeling units, which addresses the out-ofvocabulary (OOV) problem. A multi-level detection is applied to the posterior probabilities for the judgement. Evaluating on the AISHELL-2 dataset shows our proposed method outperforms the RNN-T-based approach by 2.70% in false reject rate (FRR) at 1 false alarm (FA) per hour. We further provide insights into the role of each stage of the detection cascade, where most negative samples are filtered out by the first stage with high computational efficiency.

No abstract