Federated Learning Model for Auto Insurance Rate Setting Based on Tweedie Distribution

In the assessment of car insurance claims,the claim rate for car insurance presents a highly skewed probability distribution,which is typically modeled using Tweedie distribution.The traditional approach to obtaining the Tweedie regression model involves training on a centralized dataset,when the data is provided by multiple parties,training a privacy-preserving Tweedie regression model without exchanging raw data becomes a challenge.To address this issue,this study introduces a novel vertical federated learning-based Tweedie regression algorithm for multi-party auto insurance rate setting in data silos.The algorithm can keep sensitive data locally and uses privacy-preserving techniques to achieve intersection operations between the two parties holding the data.After determining which entities are shared,the participants train the model locally using the shared entity data to obtain the local generalized linear model intermediate parameters.The homomorphic encryption algorithms are introduced to interact with and update the model intermediate parameters to collaboratively complete the joint training of the car insurance rate-setting model.Performance tests on two publicly available datasets show that the proposed federated Tweedie regression algorithm can effectively generate Tweedie regression models that leverage the value of data fromboth partieswithout exchanging data.The assessment results of the scheme approach those of the Tweedie regressionmodel learned fromcentralized data,and outperformthe Tweedie regressionmodel learned independently by a single party.

Theoretical analysis of the spatio-temporal evolution of the bulk-strain field based on a rheologic inclusion model

Based on the theoretical expression of the three-dimension rheologic inclusion model, we analyze in detail the spatio-temporal changes on the ground of the bulk-strain produced by a spherical rheologic inclusion in a semi-infinite rheologic medium. The results show that the spatio-temporal change of bulk-strain produced by the hard inclusion has three stages of different characteristics, which are similar to most of those geodetic deformation curves, but those by a soft inclusion do not. The α-stage is a long stage in which the precursors in both the near source region and the far field develop from the focal region to the periphery. The β-stage indicates a very rapid propagation of the precursors, so that they almost appear everywhere. During the γ-stage, the precursors in the far-field converge from the periphery, and the precursors in the near source region develop outwards. The theoretical results have been used to explain tentatively the stage characteristics of the spatio-temporal change of earthquake precursors.

Grouping tree species to estimate basal area increment in temperate multispecies forests in Durango,Mexico

Multispecies forests have received increased scientific attention,driven by the hypothesis that biodiversity improves ecological resilience.However,a greater species diversity presents challenges for forest management and research.Our study aims to develop basal area growth models for tree species cohorts.The analysis is based on a dataset of 423 permanent plots(2,500 m^(2))located in temperate forests in Durango,Mexico.First,we define tree species cohorts based on individual and neighborhood-based variables using a combination of principal component and cluster analyses.Then,we estimate the basal area increment of each cohort through the generalized additive model to describe the effect of tree size,competition,stand density and site quality.The principal component and cluster analyses assign a total of 37 tree species to eight cohorts that differed primarily with regard to the distribution of tree size and vertical position within the community.The generalized additive models provide satisfactory estimates of tree growth for the species cohorts,explaining between 19 and 53 percent of the total variation of basal area increment,and highlight the following results:i)most cohorts show a"rise-and-fall"effect of tree size on tree growth;ii)surprisingly,the competition index"basal area of larger trees"had showed a positive effect in four of the eight cohorts;iii)stand density had a negative effect on basal area increment,though the effect was minor in medium-and high-density stands,and iv)basal area growth was positively correlated with site quality except for an oak cohort.The developed species cohorts and growth models provide insight into their particular ecological features and growth patterns that may support the development of sustainable management strategies for temperate multispecies forests.

Prediction and driving factors of forest fire occurrence in Jilin Province,China

Forest fires are natural disasters that can occur suddenly and can be very damaging,burning thousands of square kilometers.Prevention is better than suppression and prediction models of forest fire occurrence have developed from the logistic regression model,the geographical weighted logistic regression model,the Lasso regression model,the random forest model,and the support vector machine model based on historical forest fire data from 2000 to 2019 in Jilin Province.The models,along with a distribution map are presented in this paper to provide a theoretical basis for forest fire management in this area.Existing studies show that the prediction accuracies of the two machine learning models are higher than those of the three generalized linear regression models.The accuracies of the random forest model,the support vector machine model,geographical weighted logistic regression model,the Lasso regression model,and logistic model were 88.7%,87.7%,86.0%,85.0%and 84.6%,respectively.Weather is the main factor affecting forest fires,while the impacts of topography factors,human and social-economic factors on fire occurrence were similar.

A Generalized Layered Radiative Transfer Model in the Vegetation Canopy

In this paper, a generalized layered model for radiation transfer in canopy with high vertical resolution is developed. Differing from the two-stream approximate radiation transfer model commonly used in the land surface models, the generalized model takes into account the effect of complicated canopy morphology and inhomogeneous optical properties of leaves on radiation transfer within the canopy. In the model, the total leaf area index （LAI） of the canopy is divided into many layers. At a given layer, the influences of diffuse radiation angle distributions and leaf angle distributions on radiation transfer within the canopy are considered. The derivation of equations serving the model are described in detail, and these can deal with various diffuse radiation transfers in quite broad categories of canopy with quite inhomogeneons vertical structures and uneven leaves with substantially different optical properties of adaxial and abaxial faces of the leaves. The model is used to simulate the radiation transfer for canopies with horizontal leaves to validate the generalized model. Results from the model are compared with those from the two-stream scheme, and differences between these two models are discussed.

Coverless Steganography for Digital Images Based on a Generative Model

In this paper,we propose a novel coverless image steganographic scheme based on a generative model.In our scheme,the secret image is first fed to the generative model database,to generate a meaning-normal and independent image different from the secret image.The generated image is then transmitted to the receiver and fed to the generative model database to generate another image visually the same as the secret image.Thus,we only need to transmit the meaning-normal image which is not related to the secret image,and we can achieve the same effect as the transmission of the secret image.This is the first time to propose the coverless image information steganographic scheme based on generative model,compared with the traditional image steganography.The transmitted image is not embedded with any information of the secret image in this method,therefore,can effectively resist steganalysis tools.Experimental results show that our scheme has high capacity,security and reliability.

THEORETICAL MODEL OF EFFECTIVE STRESS COEFFICIENT FOR ROCK/SOIL-LIKE POROUS MATERIALS

Physical mechanisms and influencing factors on the effective stress coefficient for rock/soil-like porous materials are investigated, based on which equivalent connectivity index is proposed. The equivalent connectivity index, relying on the meso-scale structure of porous material and the property of liquid, denotes the connectivity of pores in Representative Element Area （REA）. If the conductivity of the porous material is anisotropic, the equivalent connectivity index is a second order tensor. Based on the basic theories of continuous mechanics and tensor analysis, relationship between area porosity and volumetric porosity of porous materials is deduced. Then a generalized expression, describing the relation between effective stress coefficient tensor and equivalent connectivity tensor of pores, is proposed, and the expression can be applied to isotropic media and also to anisotropic materials. Furthermore, evolution of porosity and equivalent connectivity index of the pore are studied in the strain space, and the method to determine the corresponding functions in expressions above is proposed using genetic algorithm and genetic programming. Two applications show that the results obtained by the method in this paper perfectly agree with the test data. This paper provides an important theoretical support to the coupled hydro-mechanical research.

Optimization of environmental variables in habitat suitability modeling for mantis shrimp Oratosquilla oratoria in the Haizhou Bay and adjacent waters

Habitat suitability index(HSI)models have been widely used to analyze the relationship between species abundance and environmental factors,and ultimately inform management of marine species.The response of species abundance to each environmental variable is different and habitat requirements may change over life history stages and seasons.Therefore,it is necessary to determine the optimal combination of environmental variables in HSI modelling.In this study,generalized additive models(GAMs)were used to determine which environmental variables to be included in the HSI models.Significant variables were retained and weighted in the HSI model according to their relative contribution(%)to the total deviation explained by the boosted regression tree(BRT).The HSI models were applied to evaluate the habitat suitability of mantis shrimp Oratosquilla oratoria in the Haizhou Bay and adjacent areas in 2011 and 2013–2017.Ontogenetic and seasonal variations in HSI models of mantis shrimp were also examined.Among the four models(non-optimized model,BRT informed HSI model,GAM informed HSI model,and both BRT and GAM informed HSI model),both BRT and GAM informed HSI model showed the best performance.Four environmental variables(bottom temperature,depth,distance offshore and sediment type)were selected in the HSI models for four groups(spring-juvenile,spring-adult,falljuvenile and fall-adult)of mantis shrimp.The distribution of habitat suitability showed similar patterns between juveniles and adults,but obvious seasonal variations were observed.This study suggests that the process of optimizing environmental variables in HSI models improves the performance of HSI models,and this optimization strategy could be extended to other marine organisms to enhance the understanding of the habitat suitability of target species.

Multi-source information fused generative adversarial network model and data assimilation based history matching for reservoir with complex geologies

For reservoirs with complex non-Gaussian geological characteristics,such as carbonate reservoirs or reservoirs with sedimentary facies distribution,it is difficult to implement history matching directly,especially for the ensemble-based data assimilation methods.In this paper,we propose a multi-source information fused generative adversarial network(MSIGAN)model,which is used for parameterization of the complex geologies.In MSIGAN,various information such as facies distribution,microseismic,and inter-well connectivity,can be integrated to learn the geological features.And two major generative models in deep learning,variational autoencoder(VAE)and generative adversarial network(GAN)are combined in our model.Then the proposed MSIGAN model is integrated into the ensemble smoother with multiple data assimilation(ESMDA)method to conduct history matching.We tested the proposed method on two reservoir models with fluvial facies.The experimental results show that the proposed MSIGAN model can effectively learn the complex geological features,which can promote the accuracy of history matching.

Baroclinic Instability in the Generalized Phillips' Model Part Ⅱ:Three-layer Model

The nonlinear stability of the three-layer generalized Phillips model, for which the velocity in each layeris constant and the top and bottom surfaces are either rigid or free, is studied by employing Arnol'd'svariational principle and a prior estimate method. The nonlinear stability criteria are established. For comparison, the linear instability criteria are also obtained by using normal mode method. and the influences ofthe free parameter, β parameter and curvature in vertical profile of the horizontal velocity on the linear instability are discussed by use of the growth rate curves. The comparison between the nonlinear stability criterion and the linear one is made. It is shown that insome cases the two criteria are exactly the same in form, but in other cases, they are different. This phenomenon, which reveals the nonlinear property of the linear instability features. is explained by the explosiveresonant interaction (ERI). When there exists the ERI, i.e., the nonlinear mechanisms play a leading role inthe dynamical system. the nonlinear stability criterion is different from the linear one, on the other hand.when there does not exist the ERI. the nonlinear stability criterion is the same as the linear one in form.

Blind source separation of ship-radiated noise based on generalized Gaussian model

When the distribution of the sources cannot be estimated accurately, the ICA algorithms failed to separate the mixtures blindly. The generalized Gaussian model （GGM） is presented in ICA algorithm since it can model non- Ganssian statistical structure of different source signals easily. By inferring only one parameter, a wide class of statistical distributions can be characterized. By using maximum likelihood （ML） approach and natural gradient descent, the learning rules of blind source separation （BSS） based on GGM are presented. The experiment of the ship-radiated noise demonstrates that the GGM can model the distributions of the ship-radiated noise and sea noise efficiently, and the learning rules based on GGM gives more successful separation results after comparing it with several conventional methods such as high order cumnlants and Gaussian mixture density function.

Generalized Internal Model Robust Control for Active Front Steering Intervention

Because of the tire nonlinearity and vehicle＇s parameters＇uncertainties,robust control methods based on the worst cases,such as H_∞,μsynthesis,have been widely used in active front steering control,however,in order to guarantee the stability of active front steering system（AFS）controller,the robust control is at the cost of performance so that the robust controller is a little conservative and has low performance for AFS control.In this paper,a generalized internal model robust control（GIMC）that can overcome the contradiction between performance and stability is used in the AFS control.In GIMC,the Youla parameterization is used in an improved way.And GIMC controller includes two sections：a high performance controller designed for the nominal vehicle model and a robust controller compensating the vehicle parameters＇uncertainties and some external disturbances.Simulations of double lane change（DLC）maneuver and that of braking on split-μroad are conducted to compare the performance and stability of the GIMC control,the nominal performance PID controller and the H_∞controller.Simulation results show that the high nominal performance PID controller will be unstable under some extreme situations because of large vehicle＇s parameters variations,H_∞controller is conservative so that the performance is a little low,and only the GIMC controller overcomes the contradiction between performance and robustness,which can both ensure the stability of the AFS controller and guarantee the high performance of the AFS controller.Therefore,the GIMC method proposed for AFS can overcome some disadvantages of control methods used by current AFS system,that is,can solve the instability of PID or LQP control methods and the low performance of the standard H_∞controller.

Additions of landscape metrics improve predictions of occurrence of species distribution models

Species distribution models are used to aid our understanding of the processes driving the spatial patterns of species’ habitats. This approach has received criticism, however, largely because it neglects landscape metrics. We examined the relative impacts of landscape predictors on the accuracy of habitat models by constructing distribution models at regional scales incorporating environmental variables (climate, topography, vegetation, and soil types) and secondary species occurrence data, and using them to predict the occurrence of 36 species in 15 forest fragments where we conducted rapid surveys. We then selected six landscape predictors at the landscape scale and ran general linear models of species presence/absence with either a single scale predictor (the probabilities of occurrence of the distribution models or landscape variables) or multiple scale predictors (distribution models?+?one landscape variable). Our results indicated that distribution models alone had poor predictive abilities but were improved when landscape predictors were added; the species responses were not, however, similar to the multiple scale predictors. Our study thus highlights the importance of considering landscape metrics to generate more accurate habitat suitability models.

Establishment of a multi-cycle generalized Weng model and its application in forecasts of global oil supply

Low oil prices under the influence of economic structure transformation and slow economic growth have hit the existing markets of traditional big oil suppliers and upgraded the conflict of oil production capacity and interest between OPEC producers and other big oil supplier countries such as the USA and Russia. Forecasting global oil production is significant for all countries for energy strategy planning, although many past forecasts have later been proved to be very seriously incorrect. In this paper,the original generalized Weng model is expanded to a multi-cycle generalized Weng model to better reflect the multi-cycle phenomena caused by political, economic and technological factors. This is used to forecast global oil production based on parameter selection from a large sample, depletion rate of remaining resources, constraints on oil reserves and cycle number determination. This research suggests that the world will reach its peak oil production in 2022, at about 4340×10~6 tonnes. China needs to plan for oil import diversity, a domestic oil production structure based on the supply pattern of large oil suppliers worldwide and the oil demand for China＇s own development.

An exact solution for flow past an accelerated horizontal plate in a rotating fluid with the generalized Oldroyd-B model

In this paper, the generalized Oldroyd-B with fractional calculus approach is used. An exact solution in terms of Fox-H function for flow past an accelerated horizontal plate in a rotating fluid is obtained by using discrete Laplace transform method. A comparison among the influence of various parameters in the Oldroyd-B model and the angular velocity of the fluid on the velocity profiles is made through numerical method in graphic form.

THEORETICAL BASIS AND GENERAL OPTIMAL FORMULATIONS OF ISOPARAMETRIC GENERALIZED HYBRID/MIXED ELEMENT MODEL FOR IMPROVED STRESS ANALYSIS

By the modified three-field Hu-Washizu principle, this paper establishes a theoretical founda- tion and general convenient formulations to generate convergent stable generalized hybrid/mixed cle- ment (GH/ME) model which is invariant with respect to coordinate, insensitive to geometric distortion and suitable for improved stress computation. In the two proposed formulations, the stress equilibrium and orthogonality constraints are imposed through incompatible displacement and internal strain modes respectively. The proposed model by the general formulations in this paper is characterized by including as- sumed stress/strain, assumed stress, variable-node, singular, compatible and incompatible GH/ME models. When using regular meshes or the constant values of the isoparametric Jacobian Det in the assumed strain in- terpolation, the incompatible GH/ME model degenerates to the hybrid/mixed element model. Both general and concrete guidelines for the optimal selection of element shape functions are suggested. By means of the GH/ME theory in this paper, a family of new GH/ME can be and have been easily constructed. The software can also be developed conveniently because all the standard subroutines for the corresponding isoparametric displacement elements can be utilized directly.

Generalized Pareto Distribution Model and Its Application to Hydrocarbon Resource Structure Prediction of the Huanghua Depression

The generalized Pareto distribution model is a kind of hydrocarbon pool size probability statistical method for resource assessment. By introducing the time variable, resource conversion rate and the geological variable, resource density, such model can describe not only different types of basins, but also any exploration samples at different phases of exploration, up to the parent population. It is a dynamic distribution model with profound geological significance and wide applicability. Its basic principle and the process of resource assessment are described in this paper. The petroleum accumulation system is an appropriate assessment unit for such method. The hydrocarbon resource structure of the Huanghua Depression in Bohai Bay Basin was predicted by using this model. The prediction results accord with the knowledge of exploration in the Huanghua Depression, and point out the remaining resources potential and structure of different petroleum accumulation systems, which are of great significance for guiding future exploration in the Huanghua Depression.

A Simplified Scheme of the Generalized Layered Radiative Transfer Model

In this paper, firstly, a simplified version （SGRTM） of the generalized layered radiative transfer model （GRTM） within the canopy, developed by us, is presented. It reduces the information requirement of inputted sky diffuse radiation, as well as of canopy morphology, and in turn saves computer resources. Results from the SGRTM agree perfectly with those of the GRTM. Secondly, by applying the linear superposition principle of the optics and by using the basic solutions of the GRTM for radiative transfer within the canopy under the condition of assumed zero soil reflectance, two sets of explicit analytical solutions of radiative transfer within the canopy with any soil reflectance magnitude are derived： one for incident diffuse, and the other for direct beam radiation. The explicit analytical solutions need two sets of basic solutions of canopy reflectance and transmittance under zero soil reflectance, run by the model for both diffuse and direct beam radiation. One set of basic solutions is the canopy reflectance αf （written as α1 for direct beam radiation） and transmittance βf （written as β1 for direction beam radiation） with zero soil reflectance for the downward radiation from above the canopy （i.e. sky）, and the other set is the canopy reflectance （αb） and transmittance βb for the upward radiation from below the canopy （i.e., ground）. Under the condition of the same plant architecture in the vertical layers, and the same leaf adaxial and abaxial optical properties in the canopies for the uniform diffuse radiation, the explicit solutions need only one set of basic solutions, because under this condition the two basic solutions are equal, i.e., αf = αb and βf = βb. Using the explicit analytical solutions, the fractions of any kind of incident solar radiation reflected from （defined as surface albedo, or canopy reflectance）, transmitted through （defined as canopy transmittance）, and absorbed by （defined as canopy absorptance） the canopy and other properties pertinent to the radiative transfer within the canopy can be estimated easily on the ground surface below the canopy （soil or snow surface） with any reflectance magnitudes. The simplified transfer model is proven to have a similar accuracy compared to the detailed model, as well as very efficient computing.

Linearity extensions of the market model:a case of the top 10 cryptocurrency prices during the pre‑COVID‑19 and COVID‑19 periods

This research investigates the appropriateness of the linear specification of the market model for modeling and forecasting the cryptocurrency prices during the pre-COVID-19 and COVID-19 periods.Two extensions are offered to compare the performance of the linear specification of the market model(LMM),which allows for the measurement of the cryptocurrency price beta risk.The first is the generalized additive model,which permits flexibility in the rigid shape of the linearity of the LMM.The second is the time-varying linearity specification of the LMM(Tv-LMM),which is based on the state space model form via the Kalman filter,allowing for the measurement of the time-varying beta risk of the cryptocurrency price.The analysis is performed using daily data from both time periods on the top 10 cryptocurrencies by adjusted market capitalization,using the Crypto Currency Index 30(CCI30)as a market proxy and 1-day and 7-day forward predictions.Such a comparison of cryptocurrency prices has yet to be undertaken in the literature.The empirical findings favor the Tv-LMM,which outperforms the others in terms of modeling and forecasting performance.This result suggests that the relationship between each cryptocurrency price and the CCI30 index should be locally instead of globally linear,especially during the COVID-19 period.

Benchmarking deep learning-based models on nanophotonic inverse design problems

Photonic inverse design concerns the problem of finding photonic structures with target optical properties.However,traditional methods based on optimization algorithms are time-consuming and computationally expensive.Recently,deep learning-based approaches have been developed to tackle the problem of inverse design efficiently.Although most of these neural network models have demonstrated high accuracy in different inverse design problems,no previous study has examined the potential effects under given constraints in nanomanufacturing.Additionally,the relative strength of different deep learning-based inverse design approaches has not been fully investigated.Here,we benchmark three commonly used deep learning models in inverse design:Tandem networks,Variational Auto-Encoders,and Generative Adversarial Networks.We provide detailed comparisons in terms of their accuracy,diversity,and robustness.We find that tandem networks and Variational Auto-Encoders give the best accuracy,while Generative Adversarial Networks lead to the most diverse predictions.Our findings could serve as a guideline for researchers to select the model that can best suit their design criteria and fabrication considerations.In addition,our code and data are publicly available,which could be used for future inverse design model development and benchmarking.