Influence Of Training Data Research Articles

Sea surface reconstruction from marine radar images using deep convolutional neural networks

The sea surface reconstructed from radar images provides valuable information for marine operations and maritime transport. The standard reconstruction method relies on the three-dimensional fast Fourier transform (3D-FFT), which introduces empirical parameters and modulation transfer function (MTF) to correct the modulation effects that may cause errors. In light of the convolutional neural networks’ (CNN) success in computer vision tasks, this paper proposes a novel sea surface reconstruction method from marine radar images based on an end-to-end CNN model with the U-Net architecture. Synthetic radar images and sea surface elevation maps were used for training and testing. Compared to the standard reconstruction method, the CNN-based model achieved higher accuracy on the same data set, with an improved correlation coefficient between reconstructed and actual wave fields of up to 0.96-0.97, and a decreased non-dimensional root mean square error (NDRMSE) of around 0.06. The influence of training data on the deep learning model was also studied. Additionally, the impact of the significant wave height and peak period on the CNN model’s accuracy was investigated. It has been demonstrated that the accuracy will fluctuate as the wave steepness increases, but the correlation coefficient remains above 0.90, and the NDRMSE remains less than 0.11.

Morphological Analyzer using the BILSTM Model only for Japanese Hiragana Sentences

This study proposes a method to develop neural models of the morphological analyzer for Japanese Hiragana sentences using the Bi-LSTM CRF model. Morphological analysis is a technique that divides text data into words and assigns information such as parts of speech. In Japanese natural language processing systems, this technique plays an essential role in downstream applications because the Japanese language does not have word delimiters between words. Hiragana is a type of Japanese phonogramic characters, which is used for texts for children or people who cannot read Chinese characters. Morphological analysis of Hiragana sentences is more difficult than that of ordinary Japanese sentences because there is less information for dividing. For morphological analysis of Hiragana sentences, we demonstrated the effectiveness of fine-tuning using a model based on ordinary Japanese text and examined the influence of training data on texts of various genres.

Analysis of Learning Influence of Training Data Selected by Distribution Consistency

This study suggests a method to select core data that will be helpful for machine learning. Specifically, we form a two-dimensional distribution based on the similarity of the training data and compose grids with fixed ratios on the distribution. In each grid, we select data based on the distribution consistency (DC) of the target class data and examine how it affects the classifier. We use CIFAR-10 for the experiment and set various grid ratios from 0.5 to 0.005. The influences of these variables were analyzed with the use of different training data sizes selected based on high-DC, low-DC (inverse of high DC), and random (no criteria) selections. As a result, the average point accuracy at 0.95% (±0.65) and the point accuracy at 1.54% (±0.59) improved for the grid configurations of 0.008 and 0.005, respectively. These outcomes justify an improved performance compared with that of the existing approach (data distribution search). In this study, we confirmed that the learning performance improved when the training data were selected for very small grid and high-DC settings.