Time-Varying GPS Displacement Network Modeling by Sequential Monte Carlo.

Geodetic observations through high-rate GPS time-series data allow the precise modeling of slow ground deformation at the millimeter level. However, significant attention has been devoted to utilizing these data for various earth science applications, including to determine crustal velocity fields and to detect significant displacement from earthquakes. The relationships inherent in these GPS displacement observations have not been fully explored. This study employs the sequential Monte Carlo method, specifically particle filtering (PF), to develop a time-varying analysis of the relationships among GPS displacement time-series within a network, with the aim of uncovering network dynamics. Additionally, we introduce a proposed graph representation to enhance the understanding of these relationships. Using the 1-Hz GEONET GNSS network data of the Tohoku-Oki Mw9.0 2011 as a demonstration, the results demonstrate successful parameter tracking that clarifies the observations' underlying dynamics. These findings have potential applications in detecting anomalous displacements in the future.

Non-homogeneous Poisson and renewal processes as spatial models for cancer mutation.

Advances in sequencing technology assisted biologists in revealing signatures of DNA cancer mutation process and in demonstrating the mutagenesis behind. However, most of these signatures proposed by majority of work focus only on the type and frequency of mutations, without considering spatial information which is non-negligible in exploring mechanisms of mutation occurrence, e.g., Kataegis. Statistical characterization of the distance between consecutive mutations can give us relative spatial information; however, it ignores location information which is as important as distance information. In this work, we assume that DNA cancer mutations are location-dependent and that integrating the two variables, location and inter-distance, is beneficial to study DNA cancer mutation processes more accurately. Particularly, instead of following a specific distribution over the whole DNA sequence, we found out that the distribution of distance between successive mutations alternates between exponential and power-law distributions. Apart from this, the parameters of either of the two distributions vary with DNA locations. The cancers with kataegis phenomenon, a specific mutation pattern caused by abnormal activity of APOBEC protein family, are more likely to be accompanied by higher parameter values of distance distribution, implying higher occurrence rate of mutation. Therefore, we propose non-homogeneous Poisson and Renewal processes to spatially model DNA cancer mutations and to describe mutation patterns quantitatively and more accurately through a statistical perspective.

Beyond Photometric Consistency: Geometry-Based Occlusion-Aware Unsupervised Light Field Disparity Estimation.

Although learning-based light field disparity estimation has achieved great progress in the most recent years, the performance of unsupervised light field learning is still hindered by occlusions and noises. By analyzing the overall strategy underlying the unsupervised methodology and the light field geometry implied in epipolar plane images (EPIs), we look beyond the photometric consistency assumption, and design an occlusion-aware unsupervised framework to deal with the situations of photometric consistency conflict. Specifically, we present a geometry-based light field occlusion modeling, which predicts a group of visibility masks and occlusion maps, respectively, by forward warping and backward EPI-line tracing. In order to learn better the noise-and occlusion-invariant representations of the light field, we propose two occlusion-aware unsupervised losses: occlusion-aware SSIM and statistics-based EPI loss. Experiment results demonstrate that our method can improve the estimation accuracy of light field depth over the occluded and noisy regions, and preserve the occlusion boundaries better.

Modeling SARS-CoV-2 nucleotide mutations as a stochastic process.

This study analyzes the SARS-CoV-2 genome sequence mutations by modeling its nucleotide mutations as a stochastic process in both the time-series and spatial domain of the gene sequence. In the time-series model, a Markov Chain embedded Poisson random process characterizes the mutation rate matrix, while the spatial gene sequence model delineates the distribution of mutation inter-occurrence distances. Our experiment focuses on five key variants of concern that had become a global concern due to their high transmissibility and virulence. The time-series results reveal distinct asymmetries in mutation rate and propensities among different nucleotides and across different strains, with a mean mutation rate of approximately 2 mutations per month. In particular, our spatial gene sequence results reveal some novel biological insights on the characteristic distribution of mutation inter-occurrence distances, which display a notable pattern similar to other natural diseases. Our findings contribute interesting insights to the underlying biological mechanism of SARS-CoV-2 mutations, bringing us one step closer to improving the accuracy of existing mutation prediction models. This research could also potentially pave the way for future work in adopting similar spatial random process models and advanced spatial pattern recognition algorithms in order to characterize mutations on other different kinds of virus families.

Neural Network Structure Optimization by Simulated Annealing.

A critical problem in large neural networks is over parameterization with a large number of weight parameters, which limits their use on edge devices due to prohibitive computational power and memory/storage requirements. To make neural networks more practical on edge devices and real-time industrial applications, they need to be compressed in advance. Since edge devices cannot train or access trained networks when internet resources are scarce, the preloading of smaller networks is essential. Various works in the literature have shown that the redundant branches can be pruned strategically in a fully connected network without sacrificing the performance significantly. However, majority of these methodologies need high computational resources to integrate weight training via the back-propagation algorithm during the process of network compression. In this work, we draw attention to the optimization of the network structure for preserving performance despite compression by pruning aggressively. The structure optimization is performed using the simulated annealing algorithm only, without utilizing back-propagation for branch weight training. Being a heuristic-based, non-convex optimization method, simulated annealing provides a globally near-optimal solution to this NP-hard problem for a given percentage of branch pruning. Our simulation results have shown that simulated annealing can significantly reduce the complexity of a fully connected network while maintaining the performance without the help of back-propagation.