Search

Recently, significant improvements on face detection have been achieved by convolutional neural networks. However, the speed of a CNN-based face detector is hampered by the large computational complexity and the huge number of parameters. It is still challenging to achieve real-time face detection as well as maintain high accuracy. In this work, we introduce a single-stage face detector with an extremely lightweight CNN to achieve fast and accurate detection. Specifically, our method has a structure to integrate the network with the image pyramid for fully utilizing the calculated features. Benefiting from weight sharing, the network size still can keep small. We also analysis the detection capability of anchors with various scales, and reserve the most effective anchors in our model. Besides, to avoid the too difficult training samples which the small network can’t learn, each ground truth face is assigned with a 0-or-1 weight during training. When tested on WIDERFACE and FDDB, it outperforms existing lightweight face detectors on accuracy with the smallest model size. The outstanding detection performance and lightweight model size signify its effectiveness and practicability.

Spiking neural networks (SNNs) are currently one of the most successful approaches to model the behavior and learning potential of the brain. Recently, they have obtained marvelous research interest thanks to their event-driven and energy-efficient characteristics. While difficult to directly train SNNs from scratch because of their non-differentiable spike operations, many works have focused on converting a trained DNN to the target SNN. However, there is no efficient method to convert the deformable convolutional layer which is frequently used in many applications. The deformable convolution layer enables deformation of the convolutional sampling grid by adding offsets to the regular sampling locations, which enhances the geometric transformation modeling capability of CNNs. In this work, we propose a novel deformable spiking CNN, which can successfully convert DNNs with deformable convolution layers to SNNs with much shorter simulation time and have low latency during inference while maintaining high accuracy. To be specific, we design an effective method dedicated for deformable convolution layers to be converted. By treating the offset prediction module as an embedded SNN, we calculate the spiking offsets multi times and use the average values as the final offsets for deformable convolution. We also propose a stage-wise DNN-SNN conversion strategy to further reduce the conversion error. We divide the network into several stages and convert each stage sequentially with retraining to diminish the difference between the source DNN and the target SNN as much as possible. The experiments on CIFAR-10 and CIFAR-100 datasets show that our method surpasses the state-of-the-art works both in conversion accuracy and inference latency.

In recent years, deformable convolutional networks are widely adopted in object detection tasks and have achieved outstanding performance. Compared with conventional convolution, the deformable convolution has an irregular receptive field to adapt to objects with different sizes and shapes. However, the irregularity of the receptive field causes inefficient access to memory and increases the complexity of control logic. Toward hardware-friendly implementation, prior works change the characteristics of deformable convolution by restricting the receptive field, leading to accuracy degradation. In this paper, we develop a dedicated Sampling Core to sample and rearrange the input pixels, enabling the convolution array to access the inputs regularly. In addition, a memory-efficient dataflow is introduced to match the processing speed of the Sampling Core and convolutional array, which improves hardware utilization and reduces access to off-chip memory. Based on these optimizations, we propose a novel hardware architecture for the deformable convolution network, which is the first work to accelerate the original deformable convolution network. With the design of the memory-efficient architecture, the access to the off-chip memory is reduced significantly. We implement it on Xilinx Virtex-7 FPGA, and experiments show that the energy efficiency reaches 50.29 GOPS/W, which is 2.5 times higher compared with executing the same network on GPU.

Background: Invasive alien insects threaten agriculture, biodiversity, and human livelihoods globally. Unfortunately, insect invasiveness still cannot be reliably predicted. Empirical policies of insect pest quarantine and inspection are mainly designed against species that are already problematic.Results: We conducted a comparative genomic analysis of 37 invasive insect species and six non-invasive insect species, showing that the gene families associated with defense, protein and nucleic acid metabolism, chemosensory function, and transcriptional regulation were significantly expanded in invasive insects, suggesting that enhanced abilities in self-protection, nutrition exploitation, and locating food or mates are intrinsic features conferring invasiveness in insects. By using these intrinsic genome features, we proposed an invasiveness index and estimated the invasiveness of 99 other insect species with genome data, classifying them as highly, moderately, or minimally invasive. Insects possessing all these aforementioned enhanced abilities are predicted to be highly invasive, and vice versa. Next, a logistic-regression classifier was trained to predict insect invasiveness, achieving 93.2% accuracy. Conclusions: We present evidence that several traits may confer invasiveness in insects and these features can be used to predict insect invasiveness accurately, and we quantify insect invasiveness with an invasiveness index.

Background Parasitoid wasps are well-known natural enemies of major agricultural pests and arthropod borne diseases. The parasitoid wasp Macrocentrus cingulum (Hymenoptera: Braconidae) has been widely used to control the notorious insect pests Ostrinia furnacalis (Asian Corn Borer) and O. nubilalis (European corn borer). One striking phenomenon exhibited by M. cingulum is polyembryony, the formation of multiple genetically identical offspring from a single zygote. Moreover, M. cingulum employs a passive parasitic strategy by preventing the host’s immune system from recognizing the embryo as a foreign body. Thus, the embryos evade the host’s immune system and are not encapsulated by host hemocytes. Unfortunately, the mechanism of both polyembryony and immune evasion remains largely unknown. Results We report the genome of the parasitoid wasp M. cingulum. Comparative genomics analysis of M. cingulum and other 11 insects were conducted, finding some gene families with apparent expansion or contraction which might be linked to the parasitic behaviors or polyembryony of M. cingulum. Moreover, we present the evidence that the microRNA miR-14b regulates the polyembryonic development of M. cingulum by targeting the c-Myc Promoter-binding Protein 1 (MBP-1), histone-lysine N-methyltransferase 2E (KMT2E) and segmentation protein Runt. In addition, Hemomucin, an O-glycosylated transmembrane protein, protects the endoparasitoid wasp larvae from being encapsulated by host hemocytes. Motif and domain analysis showed that only the hemomucin in two endoparasitoids, M. cingulum and Venturia canescens, possessing the ability of passive immune evasion has intact mucin domain and similar O-glycosylation patterns, indicating that the hemomucin is a key factor modulating the immune evasion. Conclusions The microRNA miR-14b participates in the regulation of polyembryonic development, and the O-glycosylation of the mucin domain in the hemomucin confers the passive immune evasion in this wasp. These key findings provide new insights into the polyembryony and immune evasion. Electronic supplementary material The online version of this article (10.1186/s12864-018-4783-x) contains supplementary material, which is available to authorized users.

Laminated ZrB2–SiC ultra-high temperature ceramics (UHTCs) with different layer thickness ratios and strong interfaces were fabricated by hot-pressing. Residual stresses developed on the surfaces of laminated ZrB2-SiC ceramics were evaluated by X-ray diffraction (XRD) and indentation techniques. Results showed that the characteristic XRD peaks of the ZrB2 phase on the surface of the laminated ceramics presented a shift due to the residual surface stress existence. Both XRD and indentation tested results suggested that the surface residual stress increases with the increasing layer thickness ratio confirming the theoretical analysis and calculation reported by our previous work. The presence of a compressive residual stress in the external layer of the heterogeneous material is expected to improve the mechanical properties of laminated ZrB2-SiC ceramics.

Background Insecticide resistance is a substantial problem in controlling agricultural and medical pests. Detecting target site mutations is crucial to manage insecticide resistance. Though PCR-based methods have been widely used in this field, they are time-consuming and inefficient, and typically have a high false positive rate. Acetylcholinesterases (Ace) is the neural target of the widely used organophosphate (OP) and carbamate insecticides. However, there is not any software available to detect insecticide resistance associated mutations in RNA-Seq data at present. Results A computational pipeline ACE was developed to detect resistance mutations of ace in insect RNA-Seq data. Known ace resistance mutations were collected and used as a reference. We constructed a Web server for ACE, and the standalone software in both Linux and Windows versions is available for download. ACE was used to analyse 971 RNA-Seq data from 136 studies in 7 insect pests. The mutation frequency of each RNA-Seq dataset was calculated. The results indicated that the resistance frequency was 30%–44% in an eastern Ugandan Anopheles population, thus suggesting this resistance-conferring mutation has reached high frequency in these mosquitoes in Uganda. Analyses of RNA-Seq data from the diamondback moth Plutella xylostella indicated that the G227A mutation was positively related with resistance levels to organophosphate or carbamate insecticides. The wasp Nasonia vitripennis had a low frequency of resistant reads (

The codling moth Cydia pomonella, a major invasive pest of pome fruit, has spread around the globe in the last half century. We generated a chromosome-level scaffold assembly including the Z chromosome and a portion of the W chromosome. This assembly reveals the duplication of an olfactory receptor gene (OR3), which we demonstrate enhances the ability of C. pomonella to exploit kairomones and pheromones in locating both host plants and mates. Genome-wide association studies contrasting insecticide-resistant and susceptible strains identify hundreds of single nucleotide polymorphisms (SNPs) potentially associated with insecticide resistance, including three SNPs found in the promoter of CYP6B2. RNAi knockdown of CYP6B2 increases C. pomonella sensitivity to two insecticides, deltamethrin and azinphos methyl. The high-quality genome assembly of C. pomonella informs the genetic basis of its invasiveness, suggesting the codling moth has distinctive capabilities and adaptive potential that may explain its worldwide expansion., The codling moth, Cydia pomonella, is one of the major pests of pome fruit (apples and pears) and walnuts. Here, the authors sequence and analyze its genome, providing insights on olfactory and detoxification processes that may underlie its worldwide expansion.

To fabricate metallic 316L/HA (hydroxyapatite) materials which meet the requirements of an implant&rsquo, s mechanical properties and bioactivity for its function as human bone replacement, selective laser melting (SLM) has been employed in this study to prepare a 316L stainless steel matrix, which was subsequently covered with a hydroxyapatite (HA) coating using the sol-gel method. High density (98.9%) as-printed parts were prepared using a laser power of 230 W and a scanning speed of 800 mm/s. Austenite and residual acicular ferrite existed in the microstructure of the as-printed 316L stainless steel, and the sub-grain was uniform, whose primary dendrite spacing was around 0.35 &mu, m. The as-printed 316L stainless steel showed the highest Vickers hardness, elastic modulus, and tensile strength at ~ (~ means about, same applies below unless stated otherwise) 247 HV, ~214.2 GPa, and ~730 MPa, respectively. The elongation corresponding to the highest tensile strength was ~38.8%. The 316L/HA structure, measured by the Relative Growth Rate (RGR) value, exhibited no cell cytotoxicity, and presented better biocompatibility than the uncoated as-printed and as-cast 316L samples.

Insect pests reduce yield and cause economic losses, which are major problems in agriculture. Parasitic wasps are the natural enemies of many agricultural pests and thus have been widely used as biological control agents. Plants, phytophagous insects and parasitic wasps form a tritrophic food chain. Understanding the interactions in this tritrophic system should be helpful for developing parasitic wasps for pest control and deciphering the mechanisms of parasitism. However, the genomic resources for this tritrophic system are not well organized. Here, we describe the WaspBase, a new database that contains 573 transcriptomes of 35 parasitic wasps and the genomes of 12 parasitic wasps, 5 insect hosts and 8 plants. In addition, we identified long non-coding RNA, untranslated regions and 25 widely studied gene families from the genome and transcriptome data of these species. WaspBase provides conventional web services such as Basic Local Alignment Search Tool, search and download, together with several widely used tools such as profile hidden Markov model, Multiple Alignment using Fast Fourier Transform, automated alignment trimming and JBrowse. We also present a collection of active researchers in the field of parasitic wasps, which should be useful for constructing scientific networks in this field.

Figure S1. Flow cytometry estimation of the genome size for the M. cingulum. Figure S2. The distribution of 17-mer frequency in M. cingulum genome sequencing reads. Figure S3. Distribution of GC content, CpG Obs/ExpRatios of M. cingulum(Mcin), N. vitripennis(Nvit) and A. mellifera(Amel). Figure S4. COG function classification of the OGS in M. cingulum. Figure S5. KEGG pathway analysis of the OGS in M. cingulum. Figure S6. GO classification of the OGS in M. cingulum. Figure S7. Venn diagram of the homologous protein-coding genes among three wasps (M. cingulum, C. solmsi, N. vitripennis) and fruit fly (D. melanogaster). Figure S8. Phylogenetic relationship of CSP proteins from A. mellifera, C. floridanum, C. solmsi, M. cingulum, N.vitripennis, S.invicta. Figure S9. Phylogenetic relationship of GR proteins from A. mellifera, C. floridanum, C. solmsi, M. cingulum, N.vitripennis, S.invicta. Figure S10. Phylogenetic relationship of IR proteins from A. mellifera, C. floridanum, C. solmsi, M. cingulum, N.vitripennis, S.invicta. Figure S11. Phylogenetic relationship of OR proteins from C. floridanum, D. melanogaster and M. cingulum. Figure S12. Phylogenetic relationship of OBP proteins from A. mellifera, C. floridanum, C. solmsi, M. cingulum, N.vitripennis, S.invicta. Figure S13. Phylogenetic relationship of SNMP proteins from A.mellifera, C. floridanum, C. solmsi, M. cingulum, N.vitripennis, S.invicta. Figure S14. Phylogenetic relationship of GST proteins from A. mellifera, C. floridanum, C. solmsi, M. cingulum, N.vitripennis, S.invicta. Figure S15. Phylogenetic relationship of P450 proteins from N. vitripennis, D. melanogaster and M. cingulum. Figure S16. Phylogenetic relationship of ABC proteins from M. cingulum and D. melanogaster. Figure S17. Different expression levels of miR-14b in different developmental stages of M. cingulum.Table S1. Genome sequencing data of M. cingulum. Table S2. Estimation of M. cingulum genome size using K-mer analysis. Table S3. Summary of the M. cingulum genome assembly. Table S4. The published insect genomes. Table S5. The genome assembly assessment on different insects. Table S6. Classification of repeat sequences identified in the M. cingulum genome. Table S7. Genome features of the M. cingulum, N. vitripennis and A. mellifera. Table S8. Gene features of M. cingulum, N. vitripennis and A. mellifera. Table S9. The insects with OGSs in InsectBase. Table S10. Hemomucin genes in eight wasps. Table S11. The different gene expression of embryo and pseudogerm transcriptomes in KEGG pathway. Table S12. The differently expressed miRNAs in embryo and mixed embryo transcriptomes. Table S13. Comparison of gene numbers for chemoreception in A.mellifera, C. floridanum, C. solmsi, M. cingulum, N. vitripennis and S. invicta. Table S14. Comparison of gene numbers for Gene families associated with insecticide resistance and detoxification in D. melanogaster, A. mellifera, C. floridanum, C. solmsi, M. cingulum, N. vitripennis and S. invicta. Table S15. Comparison of gene numbers of insect immune in A. mellifera, C. floridanum, C. solmsi, M. cingulum, N. vitripennis and S. invicta. Table S16. The PCR primer for target genes of mci-miR-14b. (PDF 6076 kb)

The multi-axial forging (MAF) process was introduced into the strain induced metal activation (SIMA) process to replace conventional forging. Microstructure evolution of MAF formed AZ80 magnesium alloy during partial remelting was investigated. Furthermore, the tensile mechanical properties for AZ80 magnesium alloy thixoextruded from the starting materials treated by MAF were determined. For comparison, as-cast AZ80 magnesium alloy was also thixoextruded. The results show that the SIMA route produced ideal, fine semi-solid microstructure, in which almost completely spheroidal primary solid grains had a little amount of entrapped liquid. The microstructure of the as-cast alloy in the semi-solid state is less spheroidized compared with the MAF alloy under the similar isothermal holding conditions. With prolonged holding time, the size of the solid grain increases and the degree of spheroidization is improved in the MAF formed alloys. However, the solid grain size of the as-cast alloys decreases initially, and then increases with further increasing temperature. The tensile mechanical properties for AZ80 magnesium alloy thixoextruded from the starting material produced by MAF are better than those of AZ80 magnesium alloy thixoextruded from the starting material produced by casting. The ultimate tensile strength, yield strength and elongation of the alloy thixoextruded from the starting material produced by MAF are 314 MPa, 238 MPa and 14%, respectively.

The NCBI accession numbers of insect ace-1 and ace-2 genes

The SRA accession numbers of 971 RNA-Seq data used for detecting mutations

In multi-label classification, high-dimensional and sparse binary label vectors usually make existing multi-label classifiers perform unsatisfactorily, which induces a group of label compression coding (LCC) techniques particularly. So far, several linear LCC methods have been introduced via considering linear relations among labels. In this paper, we extend traditional three-layer auto-encoder to construct a five-layer one (i.e., five-layer symmetrical neural network), and then apply the training principle in extreme learning machine to determine all network weights. Therefore, a non-linear LCC approach is proposed to capture non-linear relations of labels, where the first three-layer network is regarded as a encoder and the last two layers act as a decoder. The experimental results on three benchmark data sets show that our proposed method performs better than four existing linear LCC methods according to five performance measures.