Deep Learning for Irregularly and Regularly Missing 3-D Data Reconstruction

Reconstruction of Missing Samples in Antepartum and Intrapartum FHR Measurements Via Mini-Batch-Based Minimized Sparse Dictionary Learning

“…2) Second, find an iterative solution of the independent variable W: Let w (k) i be the update result of the kth iteration of the ith sparse coefficient vector and δ (k) be the regularization term error of the kth iteration. If one marks formula (12) as f (W, δ), it can also be expressed as follows:…”

“…2) This paper focuses on addressing the problem of online mode, i.e timely data recovery for both antepartum and intrapartum FHR measurements. On the one hand, most previous approaches are actually designed for an offline mode since the reconstruction relies on tens of thousands of training samples and is time-consuming and sometimes quite repetitive [10]- [12]. But on the other, traditional dictionary learning is usually computationally expensive to train as well as to use [18], [19], especially for an inpainting problem with long missing sample length.…”

Consistent convolution kernel design for missing shots interpolation using an improved U‐net

“…U-net is distinguished by its U-shaped symmetrical structure (Fig. 2a), which is composed primarily of three parts, namely, the encoder, decoder and channel concatenation, and more information can be found in Wang et al (2021). Furthermore, we employ the nonlinear LeakyReLU (He et al, 2015) activation function to approach seismic data more accurately.…”

“…proposed using a convolutional autoencoder for randomly missing seismic trace interpolation, as well as a transfer learning strategy to reduce reliance on a large volume of labelled data in field data interpolation applications. Chai et al (2021) used a 3D U-net to reconstruct 3D seismic data from regularly and irregularly sampled seismic data with a large amount of training data. Zhang and van der Baan (2020) proposed an unsupervised algorithm for seismic data recovery that built a dictionary using the Indian buffet process.…”

“…Chai et al . (2021) used a 3D U‐net to reconstruct 3D seismic data from regularly and irregularly sampled seismic data with a large amount of training data. Zhang and van der Baan (2020) proposed an unsupervised algorithm for seismic data recovery that built a dictionary using the Indian buffet process.…”