... reservoir characterization attenuation training data machine learning upstream oil & gas applied geoscience identification application activation function exploration geophysicist 10 loss function noise attenuation model-driven processing one-click processing seismic processing One...

.... The network is trained on random input points in space and a variance of the Helmholtz equation for the scattered wavefield is used as the loss function to update the network parameters. In spite of the methods flexibility, like handling irregular surfaces and complex media, and its potential for velocity...

... deep learning applied geoscience linear extrapolation artificial intelligence upstream oil & gas autoencoder co 2 softmax extrapolation network latent space sleipner area loss function encoder network exploration geophysicist 10 seismic survey reservoir characterization neural...

... in the deblending task also intensifies the overfitting problem. We propose a self-supervised learning method for seismic data deblending and a flexible deblending algorithm for speed-accuracy tradeoff. Using a novel blind-trace network and blending loss function, self-supervised training is deployed directly...

... multiple Gulf of Mexico datasets to train and test the 2D neural network. Further, to preserve the timing of events, RMO curvatures, frequencies, and amplitudes, we propose a composite loss function based on the short-time Fourier transform. Experimental results show that the designed loss helps...

..., which made industry-scale application possible in terms of efficiency. This study improves the picking algorithm in three ways: (1) it is more stable by using a new loss function which combines the conventional wavenumber differences and the wavenumber slope differences; (2) it is more efficient because...

... is used as the prior dis- implemented to construct the loss function. We can estimate the data uncertainty from the variance in the logit space and tribution due to its interpretability. For example, a Bernoulli the mode uncertainty using entropy. distribution with a mean of 0.8 can be interpreted...

... learning artificial intelligence applied geoscience automatic stratigraphic correlation neural network stratigraphic correlation correlation segnet geophysical prospecting application segnet model loss function training data stratigraphic interpretation frequency signal reservoir...

... at every location. To address the challenge, we proposed a semi-supervised learning approach with a hybrid optimizer, i.e., Genetic-evolutionary adaptive moment estimation (G-ADAM) for acoustic impedance estimation from seismic data. In this approach, reflectivity and seismic loss functions are used...

... that the model was not overfitting on training data. Figure 5 shows the After each forward pass of images through the network, a calculated loss function over the 30 epochs. The loss loss function is calculated. A low value for the loss function decreased by approximately 60% for both the training and mean ha he...

... exposition imaginary part exploration geophysicist 10 loss function erence real part Machine learned Green s functions that approximately satisfy the wave equation Tariq Alkhalifah, Chao Song, KAUST, Umair bin Waheed, KFUPM SUMMARY inversion (Pratt, 1999). Such solutions are obtained by inverting...

... are known to suffer from instability and increased computational cost compared to the isotropic case. Here, we employ the emerging paradigm of physics-informed neural networks to solve the anisotropic qP-wave eikonal equation. By minimizing a loss function formed by imposing the validity of the eikonal...

... (PINNs) to solve potential features. WRI is often implemented in the frequency partial differential equations (PDEs). In PINNs, they use phys- domain, and thus, requires expensive matrix inversions to ical laws in the loss function to reconstruct NN based func- reconstruct the wave eld. A recently...

...( t2 / 2 2), t R , and is the bandwidth with 0 . st 2 sup 2 s R , Advantages of CIM for high-amplitude noise g (t) (s), (6) suppression where the supremum is reached at s g (t) . Then there The CIM-based loss functions with different bandwidth holds and least-squares based...

... for approximating solu- tions to wave equations include nite difference, (e.g. Virieux then the loss function of the neural network, L takes the form (1986 nite element, (e.g. Komatitsch et al. (2010 and spectral element methods, (e.g. Komatitsch and Tromp (1999 Xn (3) L := || f ||p + ai||up, Physics Informed...

... of the event; A is de ned by the energy recorded; Figure 1. Several convolutional layers are stacked together as N is determined by how many events appear in the record. This one convolutional feature (or block). Inside the convolutional loss function is back-propagated to train the neural network in layers...

... . A general additive Gaussian noise T Another loss function that has been widely used in geophysics Given this operator, we then de ne the Wasserstein loss by is closely related to the integral wave elds mis t (Huang et al., 1 W2 ((P G (ux, (P y)(x, 2 (dx). 2014; Liu et al., 2012). Mathematically...

... seismic images with different resolutions and noise levels to serve as training data sets. To improve the perception quality, we design a novel loss function that combines the l 1 loss and multi-scale structural similarity loss. Extensive experimental results on both synthetic and field seismic images...

... Type: Oral reservoir characterization upstream oil & gas colombo neural network inversion workflow machine learning deep learning sandoval-curiel artificial intelligence objective function joint inversion procedure operator exploration geophysicist 10 loss function dl inversion...

... inversion fcn dataset regular fcn inverse fcn loss function CycleFCN: A Physics-Informed Data-driven Seismic Waveform Inversion Method Peng Jin, Shihang Feng, Youzuo Lin, Brendt Wohlberg, and David Moulton, Los Alamos National Laboratory; Erol Cromwell and Xingyuan Chen, Paci c Northwest National...