Separate Estimates Research Articles

РОЗРОБКА ЕФЕКТИВНОГО ТА ТОЧНОГО МЕТОДУ ВІДСТЕЖЕННЯ ОБʼЄКТІВ НА ВІДЕО НА МОБІЛЬНИХ ПРИСТРОЯХ

One of the top research topics in computer vision area is visual object tracking. Main goal is to obtain the target object’s location in the first video frame of a given video sequence. The recent innovations of deep neural networks, specifically Siamese networks has significant impact on visual object tracking. In spite of high accuracy and high results in academic benchmarks, there are drawbacks in current state-of-the-art approaches in particular compute-intensive and large memory footprint that cannot satisfy the performance requirements of real-world applications. The aim of this paper is to design a new lightweight framework for resource-efficient and accurate visual object tracking. To add, a new tracker of efficiency benchmark and protocol were introduced. Efficiency is defined in terms of both energy consumption and execution speed on edge devices. New dual template representation for object model adaptation was developed. The first template, static, fixes the original visible appearance and thus prevents deviation and, as a result, failures caused by adaptation. The other is dynamic; the state reflects the current conditions of assembly and the appearance of its object. Unlike STARK, which incorporates additional timing information by introducing a separate estimation prediction head, we introduce parameter-free module similarity as a template update rule optimized from the latest network. We show that the learned convex combination of two patterns is effective for tracking on multiple tests. A lightweight tracker was proposed, which includes functions, dual representation of patterns and pixel-by-pixel merged blocks in its compact network. The resulting FEAR-XS tracker runs at 205 FPS on the iPhone 11, which is 4.2 times faster than LightTrack and 26.6 times faster than Ocean, with high accuracy on many tests – no state-of-the-art tracker is more accurate and faster than any FEAR tracker. In addition, the algorithm is highly energy efficient.

Radial evolution of the accuracy of ballistic solar wind backmapping

Context. Solar wind backmapping is a technique employed to connect in situ measurements of heliospheric plasma structures to their origin near the Sun. The most widely used method is ballistic mapping, which neglects the effects of solar wind acceleration and corotation and instead models the solar wind as a constant radial outflow whose speed is determined by measurements in the heliosphere. This results in plasma parcel streamlines that form an Archimedean spiral (the Parker spiral) when viewed in the solar corotating frame. This simplified approach assumes that the effects of solar wind acceleration and corotation compensate for each other in the deviation of the source longitude. Most backmapping techniques so far considered magnetic connectivity from a heliocentric distance of 1 au to the Sun. Aims. We quantify the angular deviation between different backmapping methods that depends on the location of the radial probe and on the variation in the solar wind speed with radial distance. We assess these differences depending on source longitude and solar wind propagation time. Methods. We estimated backmapping source longitudes and travel times using (1) the ballistic approximation (constant speed), (2) a physically justified method using the empirically constrained acceleration profile Iso-poly, derived from Parker solar wind equations and also a model of solar wind tangential flows that accounts for corotational effects. We compared the differences across mapped heliocentric distances and for different asymptotic solar wind speeds. Results. The ballistic method results in a Carrington longitude of the source with a maximum deviation of 4″ below 3 au compared to the physics-based mapping method taken as reference. However, the travel time especially for the slow solar wind could be underestimated by 1.5 days at 1 au compared to non-constant speed profile. This time latency may lead to an association of incorrect solar source regions with given in situ measurements. Neglecting corotational effects and accounting for acceleration alone causes a large systematic shift in the backmapped source longitude. Conclusions. Incorporating both acceleration and corotational effects leads to a more physics-based representation of the plasma trajectories through the heliosphere compared to the ballistic assumption. This approach accurately assesses the travel time and provides a more realistic estimate of the longitudinal separation between a plasma parcel measured in situ and its source region. Nonetheless, it requires knowledge of the radial density and Alfvén speed profiles to compute the tangential flow. Therefore, we propose a compromise for computing the source longitude that employs the commonly used ballistic approach and the travel times computed from the derived radial acceleration speed profile. Moreover, we conclude that this approach remains valid at all radial distances we studied and is not limited to data obtained at 1 au.

Optimizing Binding among Bimolecular Tethered Complexes.

Tethered motion is ubiquitous in nature, offering controlled movement and spatial constraints to otherwise chaotic systems. The enhanced functionality and practical utility of tethers has been exploited in biotechnology, catalyzing the design of novel biosensors and molecular assembly techniques. While notable technological advances incorporating tethered motifs have been made, a theoretical gap persists within the paradigm, hindering a comprehensive understanding of tethered-based technologies. In this work, we focus on the characterization of the binding kinetics of two tethered molecules functionalized to a hard surface. Using a mean-field approximation, the binding time of such bimolecular system is determined analytically. Furthermore, estimates of the grafting site separation and polymer lengths which expedite binding are provided. These estimates, along with the analytical theories and frameworks established here, have the potential to improve efficacy in self-assembly methods in DNA nanotechnology and can be extended to more biologically specific endeavors including targeted drug-delivery and molecular sensing.

New Sorption Isotherms Derived from a Gamma Distribution of Binding Constants.

New sorption isotherms for heterogeneous sorbents are derived by combining a Gamma distribution of binding constants with a local isotherm defined by a Langmuir or Hill equation. The new "Gamma isotherms" are expressed as Stieltjes transforms of the distribution and involve generalized exponential integrals. The related energy distributions are asymmetric and present a peak corresponding to the mean binding constant. The advantages of the new isotherms are (1) at low pressures or concentrations, with a Langmuir local isotherm, the global "Gamma-Langmuir" isotherm retrieves Henry's law; (2) contrary to the power Freundlich or hypergeometric Freundlich global isotherms, these Gamma isotherms do not need a redefinition of the standard state; (3) with a Hill local isotherm, the global "Gamma-Hill" isotherm allows a separate estimation of the cooperativity and heterogeneity parameters; and (4) the condensation approximation is a good approximation if the local isotherm is Hill and displays a high degree of cooperativity. The Gamma-Langmuir model is applied to three examples from the literature, with rather different Gamma distributions.

Combining prior and post‐release data while accounting for dispersal to improve predictions for reintroduction populations

AbstractAttempts to reintroduce species to managed areas may be compromised by dispersal into the surrounding landscape. Therefore, decisions regarding the selection and ongoing management of reintroduction areas require predicting dispersal as well as the survival and reproduction rates of the species to be reintroduced. Dispersal can potentially be measured directly by tracking animals, but this is often impractical. However, dispersal can also be inferred from re‐sighting surveys done within reintroduction areas if such data are available from multiple areas with varying connectivity to the surrounding landscape, allowing apparent survival and recruitment to be modelled as a function of connectivity metrics. Here, we show how data from 10 previous reintroductions of a New Zealand passerine, the toutouwai (Petroica longipes), were used to predict population dynamics at a predator‐controlled reintroduction area with high connectivity, and predictions then updated using post‐release data. Bayesian hierarchical modelling of the previous data produced prior distributions for productivity, adult survival and apparent juvenile survival rates that accounted for random variation among areas as well as rat density and connectivity. The modelling of apparent juvenile survival as a function of connectivity allowed it to be partitioned into estimates of survival and fidelity. Bayesian updating based on post‐release data produced posterior distributions for parameters that were consistent with the priors but much more precise. The prior data also allowed the recruitment rate estimated in the new area to be partitioned into separate estimates for productivity, juvenile survival and juvenile fidelity. Consequently, it was possible to not only estimate population growth under current management, but also predict the consequences of reducing the scale or intensity of predator control, facilitating adaptive management. The updated model could then be used to predict population growth as a function of the connectivity and predator control regime at proposed reintroduction areas while accounting for random variation among areas.

Development and Validation of the Jawline Subject Satisfaction Scale

Introduction: Research in aesthetic medicine commonly includes evaluations of subject satisfaction with treatment results. However, conventional analytic methods typically generate statistically imprecise ordinal scores. To overcome this limitation, researchers have begun employing the Rasch model, an analytical framework grounded in item response theory. The Rasch model permits scale modifications capable of enhancing measurement accuracy. This study focuses on using the Rasch model to evaluate a scale measuring subject satisfaction following aesthetic treatments to the jawline. Objective: To develop and validate a multiitem, self-administered questionnaire measuring patient satisfaction with aesthetic treatment of the jawline. Methods: A 10-item questionnaire [The Jawline Subject Satisfaction Scale (JS3)] was devised to measure subject satisfaction following aesthetic treatments of the jawline. Each question was responded to using a 5-point Likert scale, with response selections ranging from “very much satisfied” to “very much dissatisfied” or “strongly agree” to “strongly disagree.” The scale's psychometric properties (reliability and separation for items and persons, item and person fit statistics, and unidimensionality and local independence) were validated using a Rasch model based on a dataset collected from a sample of forty subjects. Results: The results of the Rasch analysis revealed high internal consistency of the JS3, with a person reliability estimate of 0.86 and an item reliability estimate of 0.96. The separation estimates for persons and items were 2.50 and 4.72, respectively, demonstrating the scale's ability to differentiate between high and low responders and validating the instrument's construct. All infit and outfit values fell within the established range (0.5-1.5), and the data fit the model of unidimensionality and local independence. Raw score transformations into logits were conducted, which were then converted to Rasch measurements. These measurements are available for use in practice for conducting standard statistical analyses evaluating treatment and/or group effects. Conclusions: The application of the Rasch model produced a valid and reliable scale (ie, JS3) for measuring satisfaction with the appearance of the jawline following aesthetic treatments.

Spatial variation of body wave attenuation in Garhwal-Kumaun Himalaya region, India

We have investigated the spatial variation of body wave attenuation in Garhwal-Kumaun Himalaya region from the datasets of 465 well located earthquakes, recorded at 52 broadband stations between 2017 and 2020. The body wave attenuation parameters QP and QS were estimated at each station by applying the extended coda normalization method for five different central frequencies ranging from 1.5 to 18 Hz. Strong frequency-dependent body wave attenuation was observed at each station over the whole region. Also, we found a significant variation of QP and QS from north to south which is consistent with geotectonic diversity beneath the study region. We have also made separate estimations of frequency-dependent relations for Garhwal and Kumaun Himalaya region in order to investigate the lateral variation in attenuation characteristics, and obtained the frequency-dependent relations as follows: QP=30±3f1.05±0.05, QS=143±20f0.88±0.07 for the Garhwal Himalaya region and QP=31±1f1.09±0.02, QS=121±11f1.00±0.04 for the Kuamun Himalaya region. Obtained Q values indicate strong body wave attenuation for both the region with no significant lateral variation. It may suggest the presence of crustal level folding, faulting, aqueous fluids, and metamorphic fluids below both the regions. Also, the ratios of QS/QP are high (>1) for the entire analyzed frequency range, suggesting a significant level of heterogeneity and tectonic complexities in the crust of the study region.

Utilizing experimental design and desirability function in optimizing RP-HPLC method for simultaneous determination of some skeletal muscle relaxants and analgesics

An experimental design and response surface methodologies using Plackett–Burman and Box-Behnken designs were applied for selecting and optimizing the most appropriate parameters which significantly affect the separation and quantitative estimation of five skeletal muscle relaxants and four analgesic drugs (baclofen, methocarbamol, dantrolene sodium, orphenadrine citrate, cyclobenzaprine hydrochloride, ketoprofen, etoricoxib, ibuprofen, and mefenamic acid) with a relatively short duration of analysis in a single run. For the separation of the nine drugs, an INERTSIL ODS-V3-5 µm C18 column (250 × 4.6 mm I.D.) was used with the optimum mobile phase conditions (45.15 mM ammonium acetate buffer pH 5.56 adjusted with acetic acid, acetonitrile, and methanol in a ratio of 30.5:29.5:40, v/v/v with a flow rate of 1.5 mL/min) and UV-detection at 220 nm. The optimized method was successfully subjected to the validation steps as described in ICH guidelines for linearity, precision, accuracy, robustness, and sensitivity. The optimized and validated method was effectively applied to determine the content of the studied drugs in their pharmaceutical preparations and to expand its applicability to the counterfeit estimation of etoricoxib in different brands of tablet dosage forms.

A Transaction Cost Model of Tax-Motivated Income Shifting into Dot-Sized Tax Havens and an Empirical Examination of E-Commerce Effects

ABSTRACT Our paper develops a new transaction cost model of tax-motivated income shifting. We analytically show that income shifted by U.S. multinational parent corporations into their tax haven subsidiaries differs from predictions based only on tax rate differences because of the transaction costs of shifting income into tax haven subsidiaries. Our model predicts that e-commerce facilitates greater income shifting by lowering the transaction costs of income-shifting operations, which theoretically supports our focus on actual income shifting instead of the traditional focus on effective tax rates alone. Empirically, we find that greater numbers of multinational corporations’ dot-sized tax havens significantly increase profits shifted out of their U.S. parent corporations and also show with a separate estimation that this amount is similar to amounts shifted into their tax haven subsidiaries, consistent with income shifting being a zero-sum strategy. This novel approach documents income shifting materially increasing with e-commerce activities that lower transaction costs. Data Availability: Data are available from the public sources cited in the text. JEL Classifications: H26; L81; F38; M48.

Mathematical model for footwear upper material consumption estimation

Appropriate material consumption estimation since the design phase for footwear fabrication is a vital issue as material costs account for a sizeable portion of the overall production cost of a pair of shoes. This paper presents a mathematical model to predict the calculation of footwear upper material consumption to improve the utilization ratio of materials through a suitable nesting map onto leather. The proposed model concentrates on the two-dimensional geometry of footwear components and the application of rich mathematical concepts. The model reflects the outlines of individual footwear components while determining the area using definite integral calculus. Five distinct rotational kinds are applied for component arrangements that correlate to the physiognomy of leather because nesting onto the leather is intractable. The simple concept of a minimal-area polygonal enclosure is applied to maximize material utilization with minimum waste. Finally, the model was verified for four consecutive Oxford footwear sizes by comparing actual upper material consumption with predicted upper material consumption. The noble contribution of this analysis is to use ImageJ software to compute upper material consumption in real case analysis through image processing techniques and separate estimation of wastes, especially the fourth waste. The results of the comparison study show that the proposed model can reduce average material requirements by 2.06 %. This minimization of waste could be beneficial in terms of economic and environmental sustainability. Therefore, this study can be applied to estimate more accurately the amount of upper material required for footwear fabrication and support better utilization of material in the footwear industry.

Agent expectations and news sentiment in the dynamics of price in a financial market

PurposeAlthough the effects of both news sentiment and expectations on price in financial markets have now been extensively demonstrated, the jointness that these predictors can have in their effects on price has not been well-defined. Investigating causal ordering in their effects on price can further our understanding of both direct and indirect effects in their relationship to market price.Design/methodology/approachWe use autoregressive distributed lag (ARDL) methodology to examine the relationship between agent expectations and news sentiment in predicting price in a financial market. The ARDL estimation is supplemented by Grainger causality testing.FindingsIn the ARDL models we implement, measures of expectations and news sentiment and their lags were confirmed to be significantly related to market price in separate estimates. Our results further indicate that in models of relationships between these predictors, news sentiment is a significant predictor of agent expectations, but agent expectations are not significant predictors of news sentiment. Granger-causality estimates confirmed the causal inferences from ARDL results.Research limitations/implicationsTaken together, the results extend our understanding of the dynamics of expectations and sentiment as exogenous information sources that relate to price in financial markets. They suggest that the extensively cited predictor of news sentiment can have both a direct effect on market price and an indirect effect on price through agent expectations.Practical implicationsEven traditional financial management firms now commonly track behavioral measures of expectations and market sentiment. More complete understanding of the relationship between these predictors of market price can further their representation in predictive models.Originality/valueThis article extends the frequently reported bivariate relationship of expectations and sentiment to market price to examine jointness in the relationship between these variables in predicting price. Inference from ARDL estimates is supported by Grainger-causality estimates.

Spiritually Based Interventions for High Blood Pressure: A Systematic Review and Meta-analysis.

This systematic review and meta-analysis aimed to evaluate the effectiveness of spiritually based interventions on blood pressure (BP) among adults. A systematic search was performed using the PubMed, Scopus, and Cochrane databases to identify studies evaluating spiritual interventions, including meditation, transcendental meditation, mindfulness meditation, and yoga, for high BP among adults up to January 1, 2022. The inclusion criteria were (a) randomized controlled trials (RCTs), (b) studies in English or Persian, (c) studies conducted among adults (≥ 18years), and (d) studies reporting systolic or diastolic BP. Given the high heterogeneity of these studies, a random effect model was used to calculate the effect sizes for the RCTs. In total, the systematic review included 24 studies and the meta-analysis included 23 studies. As some of studies reported two or more outcome measurements, separate estimates of each outcome were extracted for that study (24 datasets). Fifteen trials reported the mean (SD) systolic blood pressure (SBP), and 13 trials reported the mean (SD) diastolic blood pressure (DBP). In addition, 13 studies reported means (SDs) and six trials reported mean changes in DBP. A significant decrease was found in systolic BP following intervention ((WMD (weighted mean difference) = - 7.63 [- 9.61 to - 5.65; P < 0.001]). We observed significant heterogeneity among the studies (I2 = 96.9; P < 0.001). A significant decrease was observed in DBP following the interventions (WMD = - 4.75 [- 6.45 to - 3.05; P < 0.001]). Spiritually based interventions including meditation and yoga had beneficial effects in reducing both SBP and DBP. Reducing BP can be expected to reduce the risk of cardiovascular diseases.

Optical Imaging Method of Synthetic-Aperture Radar for Moving Targets

Traditional synthetic-aperture radar (SAR) imaging, while adept at capturing stationary scenes with clarity, often results in the blurring of moving targets due to Doppler spectrum disparities. To overcome this, this study introduces an innovative optical approach for imaging moving targets. By employing a spatial light modulator (SLM), the phase information of SAR data can be modulated onto a light beam. Then, the light propagation path is meticulously designed to take advantage of the free propagation of light to compensate for the phase differences in SAR data, thereby achieving focal clarity. Simulations and experimental validations have demonstrated that this method surpasses traditional digital imaging techniques in terms of focusing precision. Additionally, the processing delay is only as long as the time taken for light propagation. The optical path is simple, avoiding complex assembly and alignment. This novel approach opens up new possibilities for the SAR imaging of moving targets, offering potential applications in moving target extraction, separation, and velocity estimation.

Integrated distance sampling models for simple point counts.

Point counts (PCs) are widely used in biodiversity surveys but, despite numerous advantages, simple PCs suffer from several problems: detectability, and therefore abundance, is unknown; systematic spatiotemporal variation in detectability yields biased inferences, and unknown survey area prevents formal density estimation and scaling-up to the landscape level. We introduce integrated distance sampling (IDS) models that combine distance sampling (DS) with simple PC or detection/nondetection (DND) data to capitalize on the strengths and mitigate the weaknesses of each data type. Key to IDS models is the view of simple PC and DND data as aggregations of latent DS surveys that observe the same underlying density process. This enables the estimation of separate detection functions, along with distinct covariate effects, for all data types. Additional information from repeat or time-removal surveys, or variable survey duration, enables the separate estimation of the availability and perceptibility components of detectability with DS and PC data. IDS models reconcile spatial and temporal mismatches among data sets and solve the above-mentioned problems of simple PC and DND data. To fit IDS models, we provide JAGS code and the new "IDS()" function in the R package unmarked. Extant citizen-science data generally lack the information necessary to adjust for detection biases, but IDS models address this shortcoming, thus greatly extending the utility and reach of these data. In addition, they enable formal density estimation in hybrid designs, which efficiently combine DS with distance-free, point-based PC or DND surveys. We believe that IDS models have considerable scope in ecology, management, and monitoring.

Can Climate Shocks Make Vulnerable Subjects More Willing to Take Risks?

While economists in the past tended to assume that individual preferences, including risk preferences, are stable over time, a recent literature has developed and indicates that risk preferences respond to shocks, with mixed evidence on the direction of the responses. This paper utilizes a natural experiment with covariate (drought) and idiosyncratic shocks in combination with an independent field risk experiment. The risk experiment uses a Certainty Equivalent-Multiple Choice List approach and is played 1–2 years after the subjects were (to a varying degree) exposed to a covariate drought shock or idiosyncratic shocks for a sample of resource-poor young adults living in a risky semi-arid rural environment in Sub-Saharan Africa. The experimental approach facilitates a comprehensive assessment of shock effects on experimental risk premiums for risky prospects with varying probabilities of good and bad outcomes. The experiment also facilitates the estimation of the utility curvature in an Expected Utility (EU) model and, alternatively, separate estimation of probability weighting and utility curvature in three different Rank Dependent Utility models with a two-parameter Prelec probability weighting function. Our study is the first to comprehensively test the theoretical predictions of Gollier and Pratt (Econom J Econom Soc 64:1109–1123, 1996) versus Quiggin (Econ Theor 22(3):607–611, 2003). Gollier and Pratt (1996) build on EU theory and state that an increase in background risk will make subjects more risk averse while Quiggin (2003) states that an increase in background risk can enhance risk-taking in certain types of non-EU models. We find strong evidence that such non-EU preferences dominate in our sample.

Local mitigation of higher-order ionospheric effects in DFMC SBAS and system performance evaluation

Dual-frequency multi-constellation (DFMC) satellite-based augmentation system (SBAS) is a new SBAS standard for aeronautical navigation systems. It supports aircraft navigation from the enroute to approach phases via the L1 and L5 frequencies (1575.42 and 1176.45 MHz). Although the ionosphere-free (IF) combination in the DFMC SBAS operation removes the first-order ionospheric delays in the pseudorange measurement, remaining terms including the satellite-clock offset errors and higher-order ionospheric (HOI) delays are still unaccounted for. The DFMC SBAS accuracy and integrity can be affected by the HOI effects, especially during severe ionospheric disturbances. In this work, we present the local DFMC SBAS corrections with and without the mitigation of HOI delays. We first estimate the HOI delay terms using the received pseudorange followed by separate satellite and receiver bias estimations based on the minimum sum-variance technique. The integrity terms can then be obtained. The performances of DFMC SBAS using the global navigation satellite system (GNSS) data including GPS, Galileo, and QZSS are evaluated using obtained GNSS data at stations in Thailand on the ionospheric quiet and disturbed days. The results show that with the HOI mitigation, the vertical positioning errors (VPE) on the quiet and disturbed days can be improved by 12% and 9%, whereas the vertical protection levels (VPL) are improved by 16% and 21%, respectively. In addition, we perform a preliminary assessment of DFMC SBAS based on the International Civil Aviation Organization (ICAO) requirements of two categories: Localizer Performance with Vertical guidance (LPV-200) and Category I precision approach (CAT-I) showing promising results.

Ultra-sensitive separation estimation of optical sources

Historically, the resolution of optical imaging systems was dictated by diffraction, and the Rayleigh criterion was long considered an unsurpassable limit. In superresolution microscopy, this limit is overcome by manipulating the emission properties of the object. However, in passive imaging, when sources are uncontrolled, reaching sub-Rayleigh resolution remains a challenge. Here, we implement a quantum-metrology-inspired approach for estimating the separation between two incoherent sources, achieving a sensitivity five orders of magnitude beyond the Rayleigh limit. Using a spatial mode demultiplexer, we examine scenes with bright and faint sources, through intensity measurements in the Hermite–Gauss basis. Analyzing sensitivity and accuracy over an extensive range of separations, we demonstrate the remarkable effectiveness of demultiplexing for sub-Rayleigh separation estimation. These results effectively render the Rayleigh limit obsolete for passive imaging.

Constructing decision rules for multiproduct newsvendors: An integrated estimation-and-optimization framework

In this paper, we develop an integrated estimation-and-optimization framework for constructing decisions rules for the order quantities of multiple perishable products. The framework accounts for substitution in case of scarcity by providing models incorporating substitution rates, received as input. Specifically, it builds decision rules minimizing the empirical risk and accommodates multiple classes of functions, including those leveraging exogenous features. We provide a continuous reformulation of the empirical risk minimization problem that is quadratic in the decision rules. In addition, we show how to incorporate information from in-sample optimal decisions to derive a linear approximation to it. The dual of this approximation allows us to estimate decision rules that exploit nonlinearities and interactions of features. The numerical results on real and simulated datasets suggest that decision rules computed within the proposed integrated framework have the potential to outperform separated estimation and optimization methods. In particular, larger improvements are obtained under low service levels requirements or high volatility of the demands. Furthermore, we empirically observe that information provided by exogenous features may be particularly significant in the presence of large fluctuations of the demand.

DeepMulticut: Deep Learning of Multicut Problem for Neuron Segmentation from Electron Microscopy Volume.

Superpixel aggregation is a powerful tool for automated neuron segmentation from electron microscopy (EM) volume. However, existing graph partitioning methods for superpixel aggregation still involve two separate stages-model estimation and model solving, and therefore model error is inherent. To address this issue, we integrate the two stages and propose an end-to-end aggregation framework based on deep learning of the minimum cost multicut problem called DeepMulticut. The core challenge lies in differentiating the NPhard multicut problem, whose constraint number is exponential in the problem size. With this in mind, we resort to relaxing the combinatorial solver-the greedy additive edge contraction (GAEC)-to a continuous Soft-GAEC algorithm, whose limit is shown to be the vanilla GAEC. Such relaxation thus allows the DeepMulticut to integrate edge cost estimators, Edge-CNNs, into a differentiable multicut optimization system and allows a decision-oriented loss to feed decision quality back to the Edge-CNNs for adaptive discriminative feature learning. Hence, the model estimators, Edge-CNNs, can be trained to improve partitioning decisions directly while beyond the NP-hardness. Also, we explain the rationale behind the DeepMulticut framework from the perspective of bi-level optimization. Extensive experiments on three public EM datasets demonstrate the effectiveness of the proposed DeepMulticut.

Enhanced CRS-based Wavefield Separation and WNE-based Velocity Estimation for UAV-borne GPR Data Processing

Enhanced CRS-based Wavefield Separation and WNE-based Velocity Estimation for UAV-borne GPR Data Processing