No abstract

No abstract

A nucleotide sequence (GenBank AI134194) was identified in the database of the Berkeley Drosophila Genome Project based on the similarity of its translated sequence to insulin-like peptides known for other insects. The putative A chain of the Drosophila peptide was synthesized and used to produce an antiserum for immunocytochemistry. Medial neurosecretory cells and their axons were immunostained in whole brains of Drosophila larvae and adults. In larvae, immunostained axons from these cells extended to the corpus cardiacum in the ring gland and the hypocerebral ganglion and along the aorta in the head. In adults, axons from these cells extended along the aorta to the corpus cardiacum-hypocerebral ganglion complex near the cardiac valve and branched along the midgut and crop. In contrast, two clusters of lateral neurosecretory cells and their axons were immunostained weakly in brains of female mosquitoes, Aedes aegypti. No other cells were stained in the nervous systems or midguts of Drosophila larvae and adults or Ae. aegypti females. These specific localizations substantiate the existence of an endogenous insulin-like peptide in Drosophila and suggest that a similar peptide is present in the distantly related mosquitoes.

No abstract

With the increasing adoption of Deep Neural Network (DNN) models as integral parts of software systems, efficient operational testing of DNNs is much in demand to ensure these models' actual performance in field conditions. A challenge is that the testing often needs to produce precise results with a very limited budget for labeling data collected in field.Viewing software testing as a practice of reliability estimation through statistical sampling, we re-interpret the idea behind conventional structural coverages as conditioning for variance reduction. With this insight we propose an efficient DNN testing method based on the conditioning on the representation learned by the DNN model under testing. The representation is defined by the probability distribution of the output of neurons in the last hidden layer of the model. To sample from this high dimensional distribution in which the operational data are sparsely distributed, we design an algorithm leveraging cross entropy minimization.Experiments with various DNN models and datasets were conducted to evaluate the general efficiency of the approach. The results show that, compared with simple random sampling, this approach requires only about a half of labeled inputs to achieve the same level of precision. CCS CONCEPTS• Software and its engineering → Software testing and debugging; • Computing methodologies → Neural networks.