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.

Modeling load distribution for rural photovoltaic grid areas using image recognition

Expanding photovoltaic(PV)resources in rural-grid areas is an essential means to augment the share of solar energy in the energy landscape,aligning with the“carbon peaking and carbon neutrality”objectives.However,rural power grids often lack digitalization;thus,the load distribution within these areas is not fully known.This hinders the calculation of the available PV capacity and deduction of node voltages.This study proposes a load-distribution modeling approach based on remote-sensing image recognition in pursuit of a scientific framework for developing distributed PV resources in rural grid areas.First,houses in remote-sensing images are accurately recognized using deep-learning techniques based on the YOLOv5 model.The distribution of the houses is then used to estimate the load distribution in the grid area.Next,equally spaced and clustered distribution models are used to adaptively determine the location of the nodes and load power in the distribution lines.Finally,by calculating the connectivity matrix of the nodes,a minimum spanning tree is extracted,the topology of the network is constructed,and the node parameters of the load-distribution model are calculated.The proposed scheme is implemented in a software package and its efficacy is demonstrated by analyzing typical remote-sensing images of rural grid areas.The results underscore the ability of the proposed approach to effectively discern the distribution-line structure and compute the node parameters,thereby offering vital support for determining PV access capability.

Modelling the nesting-habitat of threatened vulture species in the caucasus:An ecosystem approach to formalising environmental factors in species distribution models

Abiotic factors play an important role in species localisation,but biotic and anthropogenic predictors must also be considered in distribution modelling for models to be biologically meaningful.In this study,we formalised the biotic predictors of nesting sites for four threatened Caucasian vultures by including species distribution models(wild ungulates,nesting tree species)as biotic layers in the vulture Maxent models.Maxent was applied in the R dismo package and the best set of the model parameters were defined in the R ENMeval package.Performance metrics were continuous Boyce index,Akaike's information criterion,the area under receiver operating curve and true skill statistics.We also calculated and evaluated the null models.Kernel density estimation method was applied to assess the overlap of vulture ecological niches in the environmental space.The accessibility of anthropogenic food resources was estimated using the Path Distance measure that considers elevation gradient.The availability of pine forests(Scots Pine)and wild ungulates(Alpine Chamois and Caucasian Goat)contributed the most(29.6%and 34.3%)to Cinereous Vulture(Aegypius monachus)nesting site model.Wild ungulate distribution also contributed significantly(about 46%)to the Bearded Vulture(Gypaetus barbatus)model.This scavenger nests in the highlands of the Caucasus at a minimum distance of 5–10 km from anthropogenic facilities.In contrast,livestock as a food source was most important in colony distribution of Griffon Vulture(Gyps fulvus).The contribution of distances to settlements and agricultural facilities to the model was 45%.The optimal distance from Egyptian Vulture(Neophron percnopterus)nesting sites to settlements was only 3–10 km,to livestock facilities no more than 15 km with the factor contribution of about 57%.Excluding the wild ungulate availability,the ecological niches of studied vultures overlapped significantly.Despite similar foraging and nesting requirements,Caucasian vultures are not pronounced nesting and trophic competitors due to the abundance of nesting sites,anthropogenic food sources and successful niche sharing.

Common Species Distribution Models in Biodiversity Analysis and Their Challenges and Prospects in Application

Species distribution models have been widely used to explore suitable habitats of species,the impact of climate change on the distribution of suitable habitats of species,and the construction of ecological reserves.This paper introduced species distribution models commonly used in biodiversity analysis,as well as model performance evaluation indexes,challenges in the application of species distribution models,and finally prospected the development trend of research on species distribution models.

Autonomous Vehicle Platoons In Urban Road Networks:A Joint Distributed Reinforcement Learning and Model Predictive Control Approach

In this paper, platoons of autonomous vehicles operating in urban road networks are considered. From a methodological point of view, the problem of interest consists of formally characterizing vehicle state trajectory tubes by means of routing decisions complying with traffic congestion criteria. To this end, a novel distributed control architecture is conceived by taking advantage of two methodologies: deep reinforcement learning and model predictive control. On one hand, the routing decisions are obtained by using a distributed reinforcement learning algorithm that exploits available traffic data at each road junction. On the other hand, a bank of model predictive controllers is in charge of computing the more adequate control action for each involved vehicle. Such tasks are here combined into a single framework:the deep reinforcement learning output(action) is translated into a set-point to be tracked by the model predictive controller;conversely, the current vehicle position, resulting from the application of the control move, is exploited by the deep reinforcement learning unit for improving its reliability. The main novelty of the proposed solution lies in its hybrid nature: on one hand it fully exploits deep reinforcement learning capabilities for decisionmaking purposes;on the other hand, time-varying hard constraints are always satisfied during the dynamical platoon evolution imposed by the computed routing decisions. To efficiently evaluate the performance of the proposed control architecture, a co-design procedure, involving the SUMO and MATLAB platforms, is implemented so that complex operating environments can be used, and the information coming from road maps(links,junctions, obstacles, semaphores, etc.) and vehicle state trajectories can be shared and exchanged. Finally by considering as operating scenario a real entire city block and a platoon of eleven vehicles described by double-integrator models, several simulations have been performed with the aim to put in light the main f eatures of the proposed approach. Moreover, it is important to underline that in different operating scenarios the proposed reinforcement learning scheme is capable of significantly reducing traffic congestion phenomena when compared with well-reputed competitors.

Probability Distribution Characteristics of Strong Nonlinear Waves Under Typhoon Conditions in the Northern South China Sea

The generation and propagation mechanism of strong nonlinear waves in the South China Sea is an essential research area. In this study, the third-generation wave model WAVEWATCH Ⅲ is employed to simulate wave fields under extreme sea states. The model, integrating the ST6 source term, is validated against observed data, demonstrating its credibility. The spatial distribution of the occurrence probability of strong nonlinear waves during typhoons is shown, and the waves in the straits and the northeastern part of the South China Sea show strong nonlinear characteristics. The high-order spectral model HOS-ocean is employed to simulate the random wave surface series beneath five different platform areas. The waves during the typhoon exhibit strong nonlinear characteristics, and freak waves exist. The space-varying probability model is established to describe the short-term probability distribution of nonlinear wave series. The exceedance probability distributions of the wave surface beneath different platform areas are compared and analyzed. The results show that with an increase in the platform area, the probability of a strong nonlinear wave beneath the platform increases.

Data-driven Wasserstein distributionally robust chance-constrained optimization for crude oil scheduling under uncertainty

Crude oil scheduling optimization is an effective method to enhance the economic benefits of oil refining.But uncertainties,including uncertain demands of crude distillation units(CDUs),might make the production plans made by the traditional deterministic optimization models infeasible.A data-driven Wasserstein distributionally robust chance-constrained(WDRCC)optimization approach is proposed in this paper to deal with demand uncertainty in crude oil scheduling.First,a new deterministic crude oil scheduling optimization model is developed as the basis of this approach.The Wasserstein distance is then used to build ambiguity sets from historical data to describe the possible realizations of probability distributions of uncertain demands.A cross-validation method is advanced to choose suitable radii for these ambiguity sets.The deterministic model is reformulated as a WDRCC optimization model for crude oil scheduling to guarantee the demand constraints hold with a desired high probability even in the worst situation in ambiguity sets.The proposed WDRCC model is transferred into an equivalent conditional value-at-risk representation and further derived as a mixed-integer nonlinear programming counterpart.Industrial case studies from a real-world refinery are conducted to show the effectiveness of the proposed method.Out-of-sample tests demonstrate that the solution of the WDRCC model is more robust than those of the deterministic model and the chance-constrained model.

Asymptotic normality of error density estimator in stationary and explosive autoregressive models

In this paper,we consider the limit distribution of the error density function estima-tor in the rst-order autoregressive models with negatively associated and positively associated random errors.Under mild regularity assumptions,some asymptotic normality results of the residual density estimator are obtained when the autoregressive models are stationary process and explosive process.In order to illustrate these results,some simulations such as con dence intervals and mean integrated square errors are provided in this paper.It shows that the residual density estimator can replace the density\estimator"which contains errors.

Slip distribution inversion of seismic sub-fault dip iteration using gradient based optimizer algorithm

This paper considers setting different dips for different sub-faults to fit the actual rupture situation based on the fault rupture of the 2013 Lushan M_(S)7.0 earthquake.Meanwhile,combined with the coseismic GNSS data of the Lushan earthquake,the source parameters and sliding distribution of the Lushan earthquake fault are inversed.Firstly,we use the gradient based optimizer(GBO)in nonlinear inversion to obtain the source parameters of this seismic fault.The inversion results indicate that the strike of the fault is 206.52°,the dip is 44.10°,the length is 21.92 km,and the depth is 12.79 km.To refine the sliding distribution of the seismic fault,the seismic fault is divided into 3×3 sub-faults.Then,we fix the central sub-fault dip of 44.10°;the dip of other sub-faults is obtained by iteration.After that,the model is further divided into a fault layer model composed of 23×19 sub fault slices,and using the Matlab fitting function is used to fit the dip of the 23×19 sub faults.Finally,the Lushan seismic fault plane is established as a shovel structure with steep upper and gentle lower,steep south and gentle north.The slip distribution inversion results indicate that the depth of the slip peak is 13 km,the corresponding maximum slip momentum is 0.67 m,the seismic moment is 1.10×10^(19)N·m and the corresponding moment magnitude is MW6.66.The results above are consistent with the research results of seismology.

A unified Minorization-Maximization approach for estimation of general mixture models

The mixed distribution model is often used to extract information from heteroge-neous data and perform modeling analysis.When the density function of mixed distribution is complicated or the variable dimension is high,it usually brings challenges to the parameter es-timation of the mixed distribution model.The application of MM algorithm can avoid complex expectation calculations,and can also solve the problem of high-dimensional optimization by decomposing the objective function.In this paper,MM algorithm is applied to the parameter estimation problem of mixed distribution model.The method of assembly and decomposition is used to construct the substitute function with separable parameters,which avoids the problems of complex expectation calculations and the inversion of high-dimensional matrices.

A possible probe to neutron-skin thickness by fragment parallel momentum distribution in projectile fragmentation reactions

Neutron-skin thickness is a key parameter for a neutron-rich nucleus;however,it is difficult to determine.In the framework of the Lanzhou Quantum Molecular Dynamics(LQMD)model,a possible probe for the neutron-skin thickness(δ_(np))of neutron-rich ^(48)Ca was studied in the 140A MeV ^(48)Ca+^(9)Be projectile fragmentation reaction based on the parallel momentum distribution(p∥)of the residual fragments.A Fermi-type density distribution was employed to initiate the neutron density distributions in the LQMD simulations.A combined Gaussian function with different width parameters for the left side(Γ_(L))and the right side(Γ_(R))in the distribution was used to describe the p∥of the residual fragments.Taking neutron-rich sulfur isotopes as examples,Γ_(L) shows a sensitive correlation withδ_(np) of ^(48)Ca,and is proposed as a probe for determining the neutron skin thickness of the projectile nucleus.

Control of the Dust Vertical Distribution over Western Africa by Convection and Scavenging

Saharan dust represents more than 50%of the total desert dust emitted around the globe and its radiative effect significantly affects the atmospheric circulation at a continental scale.Previous studies on dust vertical distribution and the Saharan Air Layer(SAL)showed some shortcomings that could be attributed to imperfect representation of the effects of deep convection and scavenging.The authors investigate here the role of deep convective transport and scavenging on the vertical distribution of mineral dust over Western Africa.Using multi-year(2006-2010)simulations performed with the variable-resolution(zoomed)version of the LMDZ climate model.Simulations are compared with aerosol amounts recorded by the Aerosol Robotic Network(AERONET)and with vertical profiles of the Cloud-Aerosol Lidar with Orthogonal Polarization(CALIOP)measurements.LMDZ allows a thorough examination of the respective roles of deep convective transport,convective and stratiform scavenging,boundary layer transport,and advection processes on the vertical mineral dust distribution over Western Africa.The comparison of simulated dust Aerosol Optical Depth(AOD)and distribution with measurements suggest that scavenging in deep convection and subsequent re-evaporation of dusty rainfall in the lower troposphere are critical processes for explaining the vertical distribution of desert dust.These processes play a key role in maintaining a well-defined dust layer with a sharp transition at the top of the SAL and in establishing the seasonal cycle of dust distribution.This vertical distribution is further reshaped offshore in the Inter-Tropical Convergence Zone(ITCZ)over the Atlantic Ocean by marine boundary layer turbulent and convective transport and wet deposition at the surface.

Numerical Analysis on Temperature Distribution in a Single Cell of PEFC Operated at Higher Temperature by1D Heat Transfer Model and 3D Multi-Physics Simulation Model

This study is to understand the impact of operating conditions, especially initial operation temperature (Tini) which is set in a high temperature range, on the temperature profile of the interface between the polymer electrolyte membrane (PEM) and the catalyst layer at the cathode (i.e., the reaction surface) in a single cell of polymer electrolyte fuel cell (PEFC). A 1D multi-plate heat transfer model based on the temperature data of the separator measured using the thermograph in a power generation experiment was developed to evaluate the reaction surface temperature (Treact). In addition, to validate the proposed heat transfer model, Treact obtained from the model was compared with that from the 3D numerical simulation using CFD software COMSOL Multiphysics which solves the continuity equation, Brinkman equation, Maxwell-Stefan equation, Butler-Volmer equation as well as heat transfer equation. As a result, the temperature gap between the results obtained by 1D heat transfer model and those obtained by 3D numerical simulation is below approximately 0.5 K. The simulation results show the change in the molar concentration of O2 and H2O from the inlet to the outlet is more even with the increase in Tini due to the lower performance of O2 reduction reaction. The change in the current density from the inlet to the outlet is more even with the increase in Tini and the value of current density is smaller with the increase in Tini due to the increase in ohmic over-potential and concentration over-potential. It is revealed that the change in Treact from the inlet to the outlet is more even with the increase in Tini irrespective of heat transfer model. This is because the generated heat from the power generation is lower with the increase in Tini due to the lower performance of O2 reduction reaction.

Potential global distribution of the guava root-knot nematode Meloidogyne enterolobii under different climate change scenarios using MaxEnt ecological niche modeling

In recent years,Meloidogyne enterolobii has emerged as a major parasitic nematode infesting many plants in tropical or subtropical areas.However,the regions of potential distribution and the main contributing environmental variables for this nematode are unclear.Under the current climate scenario,we predicted the potential geographic distributions of M.enterolobii worldwide and in China using a Maximum Entropy(MaxEnt)model with the occurrence data of this species.Furthermore,the potential distributions of M.enterolobii were projected under three future climate scenarios(BCC-CSM2-MR,CanESM5 and CNRM-CM6-1)for the periods 2050s and 2090s.Changes in the potential distribution were also predicted under different climate conditions.The results showed that highly suitable regions for M.enterolobii were concentrated in Africa,South America,Asia,and North America between latitudes 30°S to 30°N.Bio16(precipitation of the wettest quarter),bio10(mean temperature of the warmest quarter),and bio11(mean temperature of the coldest quarter)were the variables contributing most in predicting potential distributions of M.enterolobii.In addition,the potential suitable areas for M.enterolobii will shift toward higher latitudes under future climate scenarios.This study provides a theoretical basis for controlling and managing this nematode.

Stiffness and Shear Stress Distribution of Glulam Beams in Elastic-Plastic Stage:Theory,Experiments and Numerical Modelling

Traditional methods focus on the ultimate bending moment of glulam beams and the fracture failure of materials with defects,which usually depends on empirical parameters.There is no systematic theoretical method to predict the stiffness and shear distribution of glulam beams in elastic-plastic stage,and consequently,the failure of such glulam beams cannot be predicted effectively.To address these issues,an analytical method considering material nonlinearity was proposed for glulam beams,and the calculating equations of deflection and shear stress distribution for different failure modes were established.The proposed method was verified by experiments and numerical models under the corresponding conditions.Results showed that the theoretical calculations were in good agreement with experimental and numerical results,indicating that the equations proposed in this paper were reliable and accurate for such glulam beams with wood material in the elastic-plastic stage ignoring the influence of mechanic properties in radial and tangential directions of wood.Furthermore,the experimental results reported by the previous studies indicated that the method was applicable and could be used as a theoretical reference for predicting the failure of glulam beams.

Micro sliding friction model considering periodic variation stress distribution of contact surface and experimental verification

Micro sliding phenomenon widely exists in the operation process of mechanical systems,and the micro sliding friction mechanism is always a research hotspot.In this work,based on the total reflection method,a measuring device for interface contact behavior under two-dimensional(2D)vibration is built.The stress distribution is characterized by the light intensity distribution of the contact image,and the interface contact behavior in the 2D vibration process is studied.It is found that the vibration angle of the normal direction of the contact surface and its fluctuation affect the interface friction coefficient,the tangential stiffness,and the fluctuation amplitude of the stress distribution.Then they will affect the change of friction state and energy dissipation in the process of micro sliding.Further,an improved micro sliding friction model is proposed based on the experimental analysis,with the nonlinear change of contact parameters caused by the normal contact stress distribution fluctuation taken into account.This model considers the interface tangential stiffness fluctuation,friction coefficient hysteresis,and stress distribution fluctuation,whose simulation results are consistent well with the experimental results.It is found that considering the nonlinear effect of a certain contact parameter alone may bring a greater error to the prediction of friction behavior.Only by integrating multiple contact parameters can the accuracy of friction prediction is improved.

Analytical model of non-uniform charge distribution within the gated region of GaN HEMTs

A physics-based analytical expression that predicts the charge,electrical field and potential distributions along the gated region of the GaN HEMT channel has been developed.Unlike the gradual channel approximation(GCA),the proposed model considers the non-uniform variation of the concentration under the gated region as a function of terminal applied volt-ages.In addition,the model can capture the influence of mobility and channel temperature on the charge distribution trend.The comparison with the hydrodynamic(HD)numerical simulation showed a high agreement of the proposed model with numerical data for different bias conditions considering the self-heating and quantization of the electron concentration.The ana-lytical nature of the model allows us to reduce the computational and time cost of the simulation.Also,it can be used as a core expression to develop a complete physics-based transistorⅣmodel without GCA limitation.

Species abundance distribution models of Toona ciliata communities in Hubei Province,China

The study of plant species abundance distribution(SAD)in natural communities is of considerable importance to understand the processes and ecological rules of community assembly.With the distribution of tree,shrub and herb layers of eight natural communities of Toona ciliata as research targets,three diff erent ecological niche models were used:broken stick model,overlapping niche model and niche preemption model,as well as three statistical models:log-series distribution model,log-normal distribution model and Weibull distribution model,to fi t SAD of the diff erent vegetation layers based on data collected.Goodness-of-fi t was compared with Chi square test,Kolmogorov–Smirnov(K–S)test and Akaike Information Criterion(AIC).The results show:(1)based on the criteria of the lowest AIC value,Chi square value and K–S value with no signifi cant diff erence(p>0.05)between theoretic and observed SADs.The suitability and goodness-of-fi t of the broken stick model was the best of three ecological niche models.The log-series distribution model did not accept the fi tted results of most vegetation layers and had the lowest goodness-of-fi t.The Weibull distribution model had the best goodness-of-fi t for SADs.Overall,the statistical SADs performed better than the ecological ones.(2)T.ciliata was the dominant species in all the communities;species richness and diversity of herbs were the highest of the vegetation layers,while the diversities of the tree layers were slightly higher than the shrub layers;there were fewer common species and more rare species in the eight communities.The herb layers had the highest community evenness,followed by the shrub and the tree layers.Due to the complexity and habitat diversity of the diff erent T.ciliata communities,comprehensive analyses of a variety of SADs and tests for optimal models together with management,are practical steps to enhance understanding of ecological processes and mechanisms of T.ciliata communities,to detect disturbances,and to facilitate biodiversity and species conservation.

The biogeographical distribution of tree species-abundance and its relation to climatic factors in mass islands

Tree species-abundance in forests is a function of geographical area and climate, although it is not clear whether such relationships apply to mass islands. We examined the spatial pattern of tree species in mass islands along the coast of Zhejiang, East China Sea using the Preston model, to identify the relationships between tree communities and climatic conditions. The results show that：（1） the biogeographical distribution of tree species-abundance conformes to Preston＇s log-normal pattern, and is in accordance with the findings in both tropical rainforests and estuarine forests;（2） the climatic factors related to tree communities in mass islands are similar to that of the subtropical zone, including the major species of evergreen needle-leaf, broad-leaf and deciduous broad-leaf forests. We conclude that the Preston model can be applied to the trees of mass islands and thus facilitate the systematic ecological researches of vegetation species＇ composition in subtropical zone.

Potential distributional shifts in North America of allelopathic invasive plant species under climate change models

Predictive studies play a crucial role in the study of biological invasions of terrestrial plants under possible climate change scenarios.Invasive species are recognized for their ability to modify soil microbial communities and influence ecosystem dynamics.Here,we focused on six species of allelopathic flowering plants-Ailanthus altissima,Casuarina equisetifolia,Centaurea stoebe ssp.micranthos,Dioscorea bulbifera,Lantana camara,and Schinus terebinthifolia-Xhat are invasive in North America and examined their potential to spread further during projected climate change.We used Species Distribution Models(SDMs)to predict future suitable areas for these species in North America under several proposed future climate models.ENMEval and Maxent were used to develop SDMs,estimate current distributions,and predict future areas of suitable climate for each species.Areas with the greatest predicted suitable climate in the future include the northeastern and the coastal northwestern regions of North America.Range size estimations demonstrate the possibility of extreme range loss for these invasives in the southeastern United States,while new areas may become suitable in the northeastern United States and southeastern Canada.These findings show an overall northward shift of suitable climate during the next few decades,given projected changes in temperature and precipitation.Our results can be utilized to analyze potential shifts in the distribution of these invasive species and may aid in the development of conservation and management plans to target and control dissemination in areas at higher risk for potential future invasion by these allelopathic species.