Recently, the attention mechanism combining spatial and channel information has been widely used in various deep convolutional neural networks (CNNs), proving its great potential in improving model performance. However, this usually uses 2D global pooling operations to compress spatial information or scaling methods to reduce the computational overhead in channel attention. These methods will result in severe information loss. Therefore, we propose a Spatial channel attention mechanism that captures cross-dimensional interaction, which does not involve dimensionality reduction and brings significant performance improvement with negligible computational overhead. The proposed attention mechanism can be seamlessly integrated into any convolutional neural network since it is a lightweight general module. Our method achieves a performance improvement of 2.08% on ResNet and 1.02% on MobileNetV2 in top-one error rate on the ImageNet dataset.

Diabetic Retinopathy (DR) may lead to blindness in diabetic patients, which is one of the most severe eye diseases. Therefore, using automatical technology to detect DR at the early phase has very vital clinical significance. In order to detect the microaneurysms (MAs) and hard exudates (HEs) of DR, a novel detection method based on deep symmetric convolutional neural network is proposed in this paper. The symmetric convolutional structure is used to improve the effectiveness of feature extraction. The proposed method also can overcome the imbalance of positive and negative samples to avoid overfitting by increasing the width and depth of the network. Furthermore, different network structures (convolution, pooling) are used to achieve different feature filtering in the stage of feature extractions. According to the experimental results, the proposed method is superior to the state-of-the-art approach on the public dataset DIARETDB1 (DB1). The detection accuracy of the objects is 92.0%, 93.2%, 93.6%, when using different filtering structures (convolution, max-pooling, ave-pooling) respectively. The detection of microaneurysms is much improved by using ave-pooling layer for feature filtering, and the max-pooling layer can improve the detection of hard exudates.

Dissolved oxygen content is a key indicator of water quality in aquaculture environment. Because of its nonlinearity, dynamics, and complexity, which makes traditional methods face challenges in the accuracy and speed of dissolved oxygen content prediction. As a solution to these issues, this study introduces a hybrid model consisting of the Light Gradient Boosting Machine (LightGBM) and the Bidirectional Simple Recurrent Unit (BiSRU). Firstly, Linear interpolation and smoothing were used to identify significant parameters. LightGBM algorithm then determines the significance of dissolved oxygen by eliminating irrelevant variables and predicting dissolved oxygen in intensive aquaculture. Finally, the attention method was implemented to map the weighting and learning parameter matrices, so enabling the BiSRU’s hidden states to be assigned different weights. The findings shown that the presented prediction model can accurately anticipate the fluctuating trend of dissolved oxygen over a 10-day period in just 122 seconds, and the accuracy rate reached 96.28%. Comparing the model effects of LightGBM -BiSRU, LightGBM - GRU, LightGBM-LSTM, and BiSRU - Attention takes the least time. Its higher prediction accuracy can provide an essential reference for intensive aquaculture water quality regulation.

In a mobile blockchain network, many mobile devices have insufficient computational capacity to execute computation-intensive tasks locally. To tackle this problem, blockchain tasks can be offloaded to edge servers with the aid of auction. However, most auction mechanisms on mobile blockchain ignore the automatic parallel execution and long-term performance. This paper aims to solve the problem of computation offloading in a mobile blockchain network. We transform this problem into a multi-choice multi-dimensional knapsack problem which is NP-hard. To improve the total utility of auction participants, this paper proposes a smart-contract-based double auction mechanism, named long-term auction for mobile blockchain (LAMB). The subtasks can be offloaded from one mobile device to heterogeneous edge servers. Also, LAMB satisfies the economic properties of an auction mechanism. Experimental results demonstrate that, the utility and utilization ratio can be achieved by 130.55% higher and 138.64% higher, respectively, in comparison to the existing auction algorithm WBD. Furthermore, the proposed LAMB can guarantee long-term performance for task offloading, and it can achieve automatic execution in an autonomous and secure environment.

Accurate feeding control in aquaculture using fish feeding behavior state is key to improving feed utilization efficiency and reducing water pollution. A hybrid method based on ResNet34-CBAM and migration learning is proposed in this paper to achieve tilapia feeding behavior classification. In order to enhance the diversity of samples, pre-processing, such as scaling and cropping, and adding noise, was performed on the images. An underwater camera was used for taking pictures of tilapia behavior, followed by feature extraction using a pre-trained ResNet34 residual network, training with an improved classification for identifying feeding behavior. The testing findings demonstrate that the suggested technique in this article is 99.72% accurate in an actual environment. The model also outperforms other recognition models based on MobileNetV2, AlexNet, VGG11, ShuffleNet_v2_x0_5, and ResNet regarding accuracy, precision, and recall compared with other convolutional neural networks. Compared to the original model, the method shows a 7.84 percent improvement, indicating that it can correctly identify the feeding behavior of real fish in aquaculture and then scientifically determine the amount of baiting, which can improve feed utilization, increase aquaculture production, and decrease aquaculture costs.

Aspect-level sentiment classification aims to predict the sentiment polarities towards the target aspects given in sentences. To address the issues of insufficient semantic information extraction and high computational complexity of attention mechanisms in existing aspect-level sentiment classification models based on deep learning, a contextual graph attention network (CGAT) is proposed. The proposed model adopts two graph attention networks to aggregate syntactic structure information into target aspects and employs a contextual attention network to extract semantic information in sentence-aspect sequences, aiming to generate aspect-sensitive text features. In addition, a syntactic attention mechanism based on syntactic relative distance is proposed, and the Gaussian function is cleverly introduced as a syntactic weight function, which can reduce computational complexities and effectively highlight the words related to aspects in syntax. Experiments on three public sentiment datasets show that the proposed model can make better use of semantic information and syntactic structure information to improve the accuracy of sentiment classification.