Attention mechanisms are widely used in current encoder/decoder frameworks of image captioning, where a weighted average on encoded vectors is generated at each time step to guide the caption decoding process. However, the decoder has little idea of whether or how well the attended vector and the given attention query are related, which could make the decoder give misled results. In this paper, we propose an "Attention on Attention" (AoA) module, which extends the conventional attention mechanisms to determine the relevance between attention results and queries. AoA first generates an "information vector" and an "attention gate" using the attention result and the current context, then adds another attention by applying element-wise multiplication to them and finally obtains the "attended information", the expected useful knowledge. We apply AoA to both the encoder and the decoder of our image captioning model, which we name as AoA Network (AoANet). Experiments show that AoANet outperforms all previously published methods and achieves a new state-ofthe-art performance of 129.8 CIDEr-D score on MS COCO "Karpathy" offline test split and 129.6 CIDEr-D (C40) score on the official online testing server. Code is available at https://github.com/husthuaan/AoANet.

Odorant-binding proteins (OBPs) mediate both perception and release of semiochemicals in insects. These proteins are the ideal targets for understanding the olfactory code of insects as well as for interfering with their communication system in order to control pest species. The two sibling Lepidopteran species Helicoverpa armigera and H. assulta are two major agricultural pests. As part of our aim to characterize the OBP repertoire of these two species, here we focus our attention on a member of this family, OBP10, particularly interesting for its expression pattern. The protein is specifically expressed in the antennae of both sexes, being absent from other sensory organs. However, it is highly abundant in seminal fluid, is transferred to females during mating and is eventually found on the surface of fertilised eggs. Among the several different volatile compounds present in reproductive organs, OBP10 binds 1-dodecene, a compound reported as an insect repellent. These results have been verified in both H. armigera and H. assulta with no apparent differences between the two species. The recombinant OBP10 binds, besides 1-dodecene, some linear alcohols and several aromatic compounds. The structural similarity of OBP10 with OBP1 of the mosquito Culex quinquefasciatus, a protein reported to bind an oviposition pheromone, and its affinity with 1-dodecene suggest that OBP10 could be a carrier for oviposition deterrents, favouring spreading of the eggs in these species where cannibalism is active among larvae.

No abstract

Background Dietary pattern analysis is a promising approach to understanding the complex relationship between diet and health. While many statistical methods exist, the literature predominantly focuses on classical methods such as dietary quality scores, principal component analysis, factor analysis, clustering analysis, and reduced rank regression. There are some emerging methods that have rarely or never been reviewed or discussed adequately. Methods This paper presents a landscape review of the existing statistical methods used to derive dietary patterns, especially the finite mixture model, treelet transform, data mining, least absolute shrinkage and selection operator and compositional data analysis, in terms of their underlying concepts, advantages and disadvantages, and available software and packages for implementation. Results While all statistical methods for dietary pattern analysis have unique features and serve distinct purposes, emerging methods warrant more attention. However, future research is needed to evaluate these emerging methods’ performance in terms of reproducibility, validity, and ability to predict different outcomes. Conclusion Selection of the most appropriate method mainly depends on the research questions. As an evolving subject, there is always scope for deriving dietary patterns through new analytic methodologies.

No abstract

Male moths possess highly sensitive and selective olfactory systems that detect sex pheromones produced by their females. Pheromone receptors (PRs) play a key role in this process. The PR HassOr14b is found to be tuned to (Z)−9-hexadecenal, the major sex-pheromone component, in Helicoverpa assulta. HassOr14b is co-localized with HassOr6 or HassOr16 in two olfactory sensory neurons within the same sensilla. As HarmOr14b, the ortholog of HassOr14b in the closely related species Helicoverpa armigera, is tuned to another chemical (Z)−9-tetradecenal, we study the amino acid residues that determine their ligand selectivity. Two amino acids located in the transmembrane domains F232I and T355I together determine the functional difference between the two orthologs. We conclude that species-specific changes in the tuning specificity of the PRs in the two Helicoverpa moth species could be achieved with just a few amino acid substitutions, which provides new insights into the evolution of closely related moth species.

Sex pheromone communication of moths helps to understand the mechanisms underlying reproductive isolation and speciation. Helicoverpa armigera and Helicoverpa assulta use (Z)-11-hexadecenal (Z11-16:Ald) and (Z)-9-hexadecenal (Z9-16:Ald) as pheromone components in reversed ratios, 97:3 and 5:95, respectively. H. armigera also produces trace amount of (Z)-9-tetradecenal (Z9-14:Ald) in the sex pheromone gland, but H. assulta does not. Wind tunnel studies revealed that the addition of small amounts (0.3%) of Z9-14:Ald to the main pheromone blend of H. armigera increased the males' attraction, but at higher doses (1%, 10%) the same compound acted as an inhibitor. In H. assulta, Z9-14:Ald reduced male attraction when presented as 1% to the pheromone blend, but was ineffective at lower concentrations (0.3%). Three types (A–C) of sensilla trichodea in antennae were identified by single sensillum recording, responding to Z11-16:Ald, Z9-14:Ald, and both Z9-16:Ald and Z9-14:Ald, respectively. Calcium imaging in the antennal lobes (ALs) revealed that the input information of the three chemicals was transmitted to three units of the macroglomerular complex (MGC) in ALs in both species: a large glomerulus for the major pheromone components, a small one for the minor pheromone components, and a third one for the behavioral antagonists. The type A and C neurons tuned to Z11-16:Ald and Z9-16:Ald had a reversed target in the MGC between the two species. In H. armigera, low doses (1, 10 μg) of Z9-14:Ald dominantly activated the glomerulus which processes the minor pheromone component, while a higher dose (100 μg) also evoked an equal activity in the antagonistic glomerulus. In H. assulta, instead, Z9-14:Ald always strongly activated the antagonistic glomerulus. These results suggest that Z9-14:Ald plays different roles in the sexual communication of two Helicoverpa species through activation of functionally different olfactory pathways.

Plant volatiles play a key role in host plant location of phytophagous insects. Cydia molesta is an important pest of pear fruit late in the growing season. We identified and quantified volatiles from immature and mature fruits of six pear varieties by using gas chromatography-mass spectrometry (GC-MS). Attractiveness of synthetic blends to adults based on gas chromatography-electroantennogram detection (GC-EAD) activity was investigated in both field and laboratory. Consistent electroantennographic activity was obtained for 12 compounds from headspace collections of the mature fruits of the six pear varieties. Qualitative and quantitative differences were found among six odor profiles. Among the six mixtures, the mixture of 1-hexanol, nonanal, ethyl butanoate, butyl acetate, ethyl hexanoate, hexyl acetate, hexyl butanoate, and farnesene (different isomers) with a 1:1:100:70:7:5:1:4 ratio from the variety Jimi and the mixture of nonanal, ethyl butanoate, 3-methylbutyl acetate, ethyl hexanoate, hexyl acetate, and farnesene with a 1:100:1:32:1:2 ratio from the variety Huangjin were highly attractive to both sexes in the field. However, male captures were much higher than those of females. Further wind tunnel tests proved that both sexes exhibited upwind flight to the lures, but only males landed on the source. Our finding indicates that mixtures mimicking Jimi and Huangjin volatiles attract both females and males of C. molesta, and these host volatiles may be involved in mate finding behavior.