Convolutional neural network(CNN) has shown its powerful ability for hyperspectral image(HSI) classification, which however, is difficult to deploy on resource-limited or low-latency platforms due to ...its parameter and computation redundancy. Though binary neural network(BNN) has attracted attention for its extreme compressing and speeding up ability by binarizing both weights and activations, it has rarely been explored for HSI classification. In this study, we elaborately design a BNN with good performance for HSI classification task. Specifically, an adaptive gradient scale module is proposed to flexibly modify the gradient during training stage to better optimize the BNN and does not add any extra computation for inference. Furthermore, a curriculum learning-based progressive binarization strategy is utilized to improve the performance. Compared with the existing BNN works, our method can increase the HSI classification accuracy by a large margin while maintaining the compressing ratio. Abundant experiments on three datasets demonstrate the effectiveness of the proposed method.
Binary neural network (BNN) provides a promising solution to deploy parameter-intensive deep single image super-resolution (SISR) models onto real devices with limited storage and computational ...resources. To achieve comparable performance with the full-precision counterpart, most existing BNNs for SISR mainly focus on compensating for the information loss incurred by binarizing weights and activations in the network through better approximations to the binarized convolution. In this study, we revisit the difference between BNNs and their full-precision counterparts and argue that the key to good generalization performance of BNNs lies on preserving a complete full-precision information flow along with an accurate gradient flow passing through each binarized convolution layer. Inspired by this, we propose to introduce a full-precision skip connection, or a variant thereof, over each binarized convolution layer across the entire network, which can increase the forward expressive capability and the accuracy of back-propagated gradient, thus enhancing the generalization performance. More importantly, such a scheme can be applied to any existing BNN backbones for SISR without introducing any additional computation cost. To validate the efficacy of the proposed approach, we evaluate it using four different backbones for SISR on four benchmark datasets and report obviously superior performance over existing BNNs and even some 4-bit competitors.
Significant advancements have been made in hyperspectral image (HSI) super-resolution with the development of deep-learning techniques. However, the current application of deep neural network ...architectures to HSI super-resolution heavily relies on empirical design strategies, which can potentially impede the improvement of image reconstruction performance and introduce distortions in the results. To address this, we propose an innovative HSI super-resolution network called dual ordinary differential equations (Dual ODEs). Drawing inspiration from ordinary differential equations (ODEs), our approach offers reliable guidelines for the design of HSI super-resolution networks. The Dual ODE model leverages a spatial ODE block to extract spatial information and a spectral ODE block to capture internal spectral features. This is accomplished by redefining the conventional residual module using the multiple ODE functions method. To evaluate the performance of our model, we conducted extensive experiments on four benchmark HSI datasets. The results conclusively demonstrate the superiority of our Dual ODE approach over state-of-the-art models. Moreover, our approach incorporates a small number of parameters while maintaining an interpretable model design, thereby reducing model complexity.
Binary neural network (BNN) provides a promising solution to deploy parameter-intensive deep single image super-resolution (SISR) models onto real devices with limited storage and computational ...resources. To achieve comparable performance with the full-precision counterpart, most existing BNNs for SISR mainly focus on compensating the information loss incurred by binarizing weights and activations in the network through better approximations to the binarized convolution. In this study, we revisit the difference between BNNs and their full-precision counterparts and argue that the key for good generalization performance of BNNs lies on preserving a complete full-precision information flow as well as an accurate gradient flow passing through each binarized convolution layer. Inspired by this, we propose to introduce a full-precision skip connection or its variant over each binarized convolution layer across the entire network, which can increase the forward expressive capability and the accuracy of back-propagated gradient, thus enhancing the generalization performance. More importantly, such a scheme is applicable to any existing BNN backbones for SISR without introducing any additional computation cost. To testify its efficacy, we evaluate it using four different backbones for SISR on four benchmark datasets and report obviously superior performance over existing BNNs and even some 4-bit competitors.
Optical detection is reliable in intrinsically characterizing partial discharges (PDs). Because of the great volume and high-level power supply of the optical devices that can satisfy the ...requirements in photosensitivity, optical PD detection can merely be used in laboratory studies. To promote the practical application of the optical approach in an actual power apparatus, a silicon photomultiplier (SiPM)-based PD sensor is introduced in this paper, and its basic properties, which include the sensitivity, pulse resolution, correlation with PD severity, and electromagnetic (EM) interference immunity, are experimentally evaluated. The stochastic phase-resolved PD pattern (PRPD) for three typical insulation defects are obtained by SiPM PD detector and are compared with those obtained using a high-frequency current transformer (HFCT) and a vacuum photomultiplier tube (PMT). Because of its good performances in the above aspects and its additional advantages, such as the small size, low power supply, and low cost, SiPM offers great potential in practical optical PD monitoring.
Mosquito control based on chemical insecticides is considered as an important element of the current global strategies for the control of mosquito-borne diseases. Unfortunately, the development of ...insecticide resistance of important vector mosquito species jeopardizes the effectiveness of insecticide-based mosquito control. In contrast to target site resistance, other mechanisms are far from being fully understood. Global protein profiles among cypermethrin-resistant, propoxur-resistant, dimethyl-dichloro-vinyl-phosphate-resistant and susceptible strain of Culex pipiens pallens were obtained and proteomic differences were evaluated by using isobaric tags for relative and absolute quantification labeling coupled with liquid chromatography/tandem mass spectrometric analysis. A susceptible strain of Culex pipiens pallens showed elevated resistance levels after 25 generations of insecticide selection, through iTRAQ data analysis detected 2,502 proteins, of which 1,513 were differentially expressed in insecticide-selected strains compared to the susceptible strain. Finally, midgut differential protein expression profiles were analyzed, and 62 proteins were selected for verification of differential expression using iTRAQ and parallel reaction monitoring strategy, respectively. iTRAQ profiles of adaptation selection to three insecticide strains combined with midgut profiles revealed that multiple insecticide resistance mechanisms operate simultaneously in resistant insects of Culex pipiens pallens. Significant molecular resources were developed for Culex pipiens pallens, potential candidates were involved in metabolic resistance and reducing penetration or sequestering insecticide. Future research that is targeted towards RNA interference of the identified metabolic targets, such as cuticular proteins, cytochrome P450s, glutathione S-transferases and ribosomal proteins proteins and biological pathways (drug metabolism-cytochrome P450, metabolism of xenobiotics by cytochrome P450, oxidative phosphorylation, ribosome) could lay the foundation for a better understanding of the genetic basis of insecticide resistance in Culex pipiens pallens.
Reliable electric insulation is the premise for the application of high-voltage technologies, such as power transmission, emergent electrified transportations, and advanced propulsion systems. With ...the current insulation strategy, the credibility of insulation system depends on the sufficient insulation materials usage, suffering from bulky structure, expensive production costs and environmental pollution. Here, we propose a strategy that applies a lightweight multiple-barrier skeleton in gas insulation to tailor electron free path and restrain electron avalanche, thus realizing a high insulation strength with less material usage. This strategy is verified by the composite insulation of porous polyimide aerogel and various types of insulating gases. It achieves a comparable insulation strength to conventional polymers by less than 10% materials usage. With above insulation strategy, a 152.5% improvement in power density and a 54.6% weight reduction are acquired in the cable system for more or all electric aircraft. This strategy inspires the design of efficient insulation solution with both high breakdown strength and lightweight.
•High insulated and lightweight gas solid composite insulation.•Electron multiplication restraining mechanism of nanopore structure is demonstrated.•Equivalent insulation strength of polymer can be achieved with 10% material usage.•Lightweight aircraft cable system with high power-density is designed.
In this study, different amounts of cinnamomum essential oil (CEO) were encapsulated in chitosan nanoparticles (NPs) (CS-NPs) through oil-in-water emulsification and ionic gelation. An ...ultraviolet-visible spectrophotometer, Fourier-transform infrared spectroscopy, synchronous thermal analysis, and X-ray diffraction were employed to analyze the CEO encapsulation. As observed by field-emission scanning electron microscopy, NP size analysis and zeta potential, the prepared CS-NPs, containing CEO (CS-CEO), were spherical with uniformly distributed sizes (diameters: 190–340 nm). The ranges of encapsulation efficiency (EE) and loading capacity (LC) were 4.6–32.9% and 0.9–10.4%, with variations in the starting weight ratio of CEO to CS from 0.11 to 0.53 (w/w). It was also found that the antioxidant activity of the CS-NPs loaded with CEO increased as the EE increased. The active ingredients of the CEO were prevented from being volatilized, significantly improving the chemical stability. The antioxidant activity of CS-CEO was higher than that of the free CEO. These results indicate the promising potential of CS-CEO as an antioxidant for food processing, and packaging applications.