Search for: All records

SUMMARY Nectar volume and sugar composition are key determinants of the strength of plant–pollinator mutualisms. The main nectar sugars are sucrose, glucose and fructose, which can vary widely in ratio and concentration across species.Brassica spp. produce a hexose‐dominant nectar (high in the monosaccharides glucose and fructose) with very low levels of the disaccharide sucrose. Cell wall invertases (CWINVs) catalyze the irreversible hydrolysis of sucrose into glucose and fructose in the apoplast. We found thatBrCWINV4Ais highly expressed in the nectaries ofBrassica rapa. Moreover, abrcwinv4anull mutant: (i) has greatly reduced CWINV activity in the nectaries; (ii) produces a sucrose‐rich nectar; but (iii) with significantly less volume. These results definitively demonstrate that CWINV activity is not only essential for the production of a hexose‐rich nectar, but also support a hypothetical model of nectar secretion in which its hydrolase activity is required for maintaining a high intracellular‐to‐extracellular sucrose ratio that facilitates the continuous export of sucrose into the nectary apoplast. The extracellular hydrolysis of each sucrose into two hexoses by BrCWINV4A also likely creates the osmotic potential required for nectar droplet formation. These results cumulatively indicate that modulation of CWINV activity can at least partially account for naturally occurring differences in nectar volume and sugar composition. Finally, honeybees prefer nectars with some sucrose, but wild‐typeB. rapaflowers were much more heavily visited than flowers ofbrcwinv4a, suggesting that the potentially attractive sucrose‐rich nectar ofbrcwinv4acould not compensate for its low volume. 

Deep learning is an important technique for extracting value from big data. However, the effectiveness of deep learning requires large volumes of high quality training data. In many cases, the size of training data is not large enough for effectively training a deep learning classifier. Data augmentation is a widely adopted approach for increasing the amount of training data. But the quality of the augmented data may be questionable. Therefore, a systematic evaluation of training data is critical. Furthermore, if the training data is noisy, it is necessary to separate out the noise data automatically. In this paper, we propose a deep learning classifier for automatically separating good training data from noisy data. To effectively train the deep learning classifier, the original training data need to be transformed to suit the input format of the classifier. Moreover, we investigate different data augmentation approaches to generate sufficient volume of training data from limited size original training data. We evaluated the quality of the training data through cross validation of the classification accuracy with different classification algorithms. We also check the pattern of each data item and compare the distributions of datasets. We demonstrate the effectiveness of the proposed approach through an experimental investigation of automated classification of massive biomedical images. Our approach is generic and is easily adaptable to other big data domains.