Flower image classification based on an improved lightweight neural network with multi-scale feature fusion and attention mechanism.

In order to solve the problem that deep learning-based flower image classification methods lose more feature information in the early feature extraction process, and the model takes up more storage space, a new lightweight neural network model based on multi-scale feature fusion and attention mechanism is proposed in this paper. First, the AlexNet model is chosen as the basic framework. Second, a multi-scale feature fusion module (MFFM) is used to replace the shallow single-scale convolution. MFFM, which contains three depthwise separable convolution branches with different sizes, can fuse features with different scales and reduce the feature loss caused by single-scale convolution. Third, two layers of improved Inception module are first added to enhance the extraction of deep features, and a layer of hybrid attention module is added to strengthen the focus of the model on key information at a later stage. Finally, the flower image classification is completed using a combination of global average pooling and fully connected layers. The experimental results demonstrate that our lightweight model has fewer parameters, takes up less storage space and has higher classification accuracy than the baseline model, which helps to achieve more accurate flower image recognition on mobile devices.

A Spatio-Temporal Attention Mechanism Based Approach for Remaining Useful Life Prediction of Turbofan Engine.

The time-series data generated by turbofan engines has a great degree of complexity and dynamics. At present, recurrent neural networks are commonly used to model and forecast the remaining useful life (RUL). The relationship of the sample data is not taken into account, and there are issues such as gradient explosion. In view of this, a spatio-temporal attention model is proposed, which comprehensively relates to the temporal association of data features and the hidden state of data features in space. At the same time, position coding is performed on the temporal relationship, avoiding the use of recurrent neural networks. Experimental results show that by combining the two dimensions, the predictive performance of the model is significantly improved. Compared with different methods on the four data sets of the commercial modular aerospace propulsion system simulation (C-MAPSS), the stability and prediction accuracy of the spatio-temporal attention model are better than that of alternative methods.