Approaches to assessing the effectiveness of earthquake predictions using the LURR (load-unload response ratio) method on Sakhalin

A retrospective analysis of the seismicity of Sakhalin from 1997 to 2019 was performed to demonstrate the possibilities of the LURR technique recently in previous our work. The following results were obtained: 84 % of earthquakes (16 out of 19, with M ≥ 5) are predicted, 25% alarms (4 out of 15 predicted areas) were false. This paper proposes an analytical dependence to describe the forecast effectiveness (Ke) for this research. The extremes of Ke were found at the value of the alarm period of 12 and 24 months. At the same time, Ke is significantly higher for the alarm period of 24 months and decreases after a two-year alarm period. Another way to prove the results obtained is the random spatio-temporal distributions of the predicted objects (19 earthquakes with M ≥ 5). 10 such random sets have been assigned to 15 predicted areas, the result shows a significant advantage of a real sample over random ones, and also practically confirms the reliability of the algorithm for using the LURR technique. The methodology and results of this work can serve as practical recommendations for working with the LURR method for seismologists.

Real Sample Consistency Regularization for GANs

Mode collapse has always been a fundamental problem in generative adversarial networks. The recently proposed Zero Gradient Penalty (0GP) regularization can alleviate the mode collapse, but it will exacerbate a discriminator’s misjudgment problem, that is the discriminator judges that some generated samples are more real than real samples. In actual training, the discriminator will direct the generated samples to point to samples with higher discriminator outputs. The serious misjudgment problem of the discriminator will cause the generator to generate unnatural images and reduce the quality of the generation. This paper proposes Real Sample Consistency (RSC) regularization. In the training process, we randomly divided the samples into two parts and minimized the loss of the discriminator’s outputs corresponding to these two parts, forcing the discriminator to output the same value for all real samples. We analyzed the effectiveness of our method. The experimental results showed that our method can alleviate the discriminator’s misjudgment and perform better with a more stable training process than 0GP regularization. Our real sample consistency regularization improved the FID score for the conditional generation of Fake-As-Real GAN (FARGAN) from 14.28 to 9.8 on CIFAR-10. Our RSC regularization improved the FID score from 23.42 to 17.14 on CIFAR-100 and from 53.79 to 46.92 on ImageNet2012. Our RSC regularization improved the average distance between the generated and real samples from 0.028 to 0.025 on synthetic data. The loss of the generator and discriminator in standard GAN with our regularization was close to the theoretical loss and kept stable during the training process.

Voltammetric Determination of Pd(II) Using 5-Hydroxyimino-4-imino-1,3-thiazolidin-2-one after a Thermal Activation of the Interaction

The interaction of Pd(II) with 5-hydroxyimino-4-imino-1,3-thiazolidin-2-one (HITO) was investigated over the pH range of 1.0–10.0. Optimal conditions for the voltammetric determination of Pd(II) were established. A new voltammetric method of Pd(II) determination by the peak of complex using NaCl as a supporting electrolyte at pH 2.0 was developed with the limit of detection 2.0∙10-7 M. The study of the method’s selectivity was presented in the work. The correctness of the method was confirmed by the analysis of model solutions and the real sample (resistor SP5-35B).

Super High-Quality SEM/FIB Imaging of Dentine Structures Without Collagen Fiber Loss Through a Metal Staining Process

Scanning Electron Microscope/Focused Ion Beam (SEM/FIB) system has become a valuable and popular tool for the analysis of biological materials such as dentine structures. According to physiological and anatomical studies, dentine structures are a complicated system containing collagen fibers, nanocrystalline hydroxyapatite, and numerous networks of tubular pores. During a routine FIB milling process, collagen fibers and other organic structures are vaporized, while the number of tubular pores remaining is increased. This causes the final cross-section to be more porous than the real sample. Unfortunately, little attention has been paid to the collagen fiber loss and how to preserve them during a FIB milling process. In this work, we present a novel and simple approach to preserve the organic portions of the dentine structure through metal staining. By using this method, the porosity of the dentine structure after the FIB milling process is significantly reduced similar to the real sample. This indicates that the organic portion of the dentine structure is well protected by the metal staining. This approach enables the SEM/FIB system to generate super-high quality SEM images with less ion beam damage; and the SEM images can better reflect the original condition of the dentine structure. Further, serial energy-dispersive X-ray spectroscopy (EDS) mapping of the stained dentine structure is achieved without an additional metal coating; and three-dimensional (3-D) elemental mapping of an occluded dentine is achieved with a significantly reduced data acquisition time.