Fig. 2

Download original image
Architecture for the CNN model. The input layers first receive the periodogram, which is then processed by an initial convolutional block (ConvBlock) with kernel sizes of 5 and 8. This is followed by three identical convolutional blocks, each with kernel sizes of 11 and 10. Each ConvBlock concludes with a MaxPooling layer to reduce the dimensionality of the data. All convolutional layers utilize the ReLU activation function. Subsequently, the data is flattened and passed through a compact fully connected neural network with dropout layers. Finally, the network predicts the three target parameters, which are normalized within the range of 0 to 1.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.