Optimal insurance risk allocation with steepest ascent and genetic algorithms

Supervised Neural Network (NN) learning is a process in which input patterns and known targets are presented to a NN while it learns to recognize (classify, map, fit, etc.) them as desired. The learning is mathematically defined as an optimisation problem, i.e., an error function representing the differences between the desired and actual output, is being minimized (Bishop, 1995; Haykin, 1999). Because the most popular supervised learning techniques are gradient based (Backpropagation BP), they suffer from the so-called Local Minima Problem (Bishop, 1995). This has motivated the employment of Global Optimisation (GO) methods for the supervised NN learning. Stochastic and heuristic GO approaches including Evolutionary Algorithms (EA) demonstrated promising performance over the last decades (Smagt, 1994; Sexton et al., 1998; Jordanov & Georgieva, 2007; etc.). EA appeared more powerful than BP and its modifications (Sexton et al., 1998; Alba & Chicone 2004), but hybrid methods that combine the advantages of one or more GO techniques and local searches were proven to be even better (Yao, 1999; Rocha et al., 2003; Alba & Chicano, 2004; Ludemir et al., 2006). Hybrid methods were promoted over local searches and simple population based techniques in Alba & Chicone (2004). The authors compared five methods: two BP implementations (gradient descent and Levenberg-Marquardt), Genetic Algorithms (GA), and two hybrid methods, combining GA with different local methods. The methods were used for NN learning applied to problems arising in medicine. Ludemir et al. (2006) optimized simultaneously NN weights and topology with a hybrid method combining Simulated Annealing (SA), Tabu Search (TS) and BP. A set of new solutions was generated on each iteration by TS rules, but the best solution was only accepted according to the probability distribution as in conventional SA. Meanwhile, the topology of the NN was also optimized and the best solution was kept. Finally, BP was used to train the best NN topology found in the previous stages. The new methodology compared favorably with SA and TS on four classification and one prediction problems. Plaginakos et al. (2001) performed several experiments to evaluate various training methods – six Differential Evolution (DE) implementations (with different mutation operators), BP, BPD (BP with deflection), SA, hybridization of BP and SA (BPSA), and GA. They reported O pe n A cc es s D at ab as e w w w .in te ch w eb .o rg

This article builds on Froot and Stein in developing a framework for analyzing the risk allocation, capital budgeting, and capital structure decisions facing insurers and reinsurers. The model incorporates three key features: (i) value‐maximizing insurers and reinsurers face product‐market as well as capital‐market imperfections that give rise to well‐founded concerns with risk management and capital allocation; (ii) some, but not all, of the risks they face can be frictionlessly hedged in the capital market; and (iii) the distribution of their cash flows may be asymmetric, which alters the demand for underwriting and hedging. We show these features result in a three‐factor model that determines the optimal pricing and allocation of risk and capital structure of the firm. This approach allows us to integrate these features into: (i) the pricing of risky investment, underwriting, reinsurance, and hedging; and (ii) the allocation of risk across all of these opportunities, and the optimal amount of surplus capital held by the firm.

The paper describes optimization Non-GEO satellite constellation design with satellite diversity using genetic algorithm. This algorithm is called as genetic satellite constellation algorithm. A rank fitness assignment method of multiobjective optimization is proposed to capture a number of solutions simultaneously. The non-dominated in GA’s is sorted to find the multiple Pareto optimal points. The genetic satellite constellation algorithm is investigated to produce the optimum satellite constellation design with the minimum number of satellites while minimizing the maximum orbit’s altitude, to minimize the angle shift between satellites, to maximize the angle between planes and the inclination angle with satellite diversity. Illustrative results of satellite constellation components and dual satellite diversity statistics for LEO and MEO constellation are presented. The simulation results are compared to ICO for MEO constellation and to Globalstar for LEO constellation. Using genetic algorithm, satellite constellation design can be achieved the visibility of dual satellite diversity statistics whilst reducing the total number of satellites. Genetic algorithm is a robust method ideally suited for satellite constellation design.

Optimization and design of an aircraft’s morphing wing-tip demonstrator for drag reduction at low speed, Part I – Aerodynamic optimization using genetic, bee colony and gradient descent algorithms

Summary form only given. Increased flexibility in design of antennas such as cylindrical horns and planar microstrip antennas is possible by allowing the layout in the selected geometry to be chosen as part of meeting the performance objectives. An example is the design of the profiles of electromagnetic horns and dielectric rod antennas. Starting from an initial cross-section geometry the horn profiles are chosen so that desired performance objectives are ultimately achieved. The profiles are obtained by means of standard optimization methods with constraints in order to satisfy the design specifications. Implicit in this is the availability of accurate and fast method for analysing the basic horn structure. These can range from standard computer packages to dedicated packages that are often faster and more accurate. A further example of the approach is the design of planar microstrip patch antennas for a desired bandwidth and radiation performance. The initial antenna is chosen as the driven element and small patches are introduced as part of the optimization process. These patches can be parasitically excited or directly linked to the feed through a power divider. In this paper the type of geometric constraints are discussed and some optimization methods are compared for handling problems that involve the variation of the geometry, and also bounds that may be required to eliminate overlap and abrupt changes in geometry. Most conventional numerical optimization methods can be used for the design. These range from evolutionary methods such as the Genetic Algorithm, and swarm methods such as the Particle Swarm Optimization. These mentioned techniques are especially valuable in a wide domain search as they balance the fittest location uncovered by an element against others uncovered by neighbours, all of which are subject to constraints. Conventional gradient methods, such as steepest descent, can be advantageous in reaching an optimum solution in fewer function evaluations though once a broad coverage technique has reached a sub-optimum solution. The paper will review some techniques that have been employed to design a range of antennas by optimization of geometry.

The numerical solution of a ordinary differential equation (ODE) basically describes any physical phenomenon of interest. Traditionally, the numerical solution of ODE is solved by a discretization process using the classical finite difference approximation. Recently, many studies have used deep neural networks (DNN) to approximate ODE solutions without discretization. However, without the optimization process, DNN cannot run independently. In this work, we proposed an alternative method that combines DNN with a genetic algorithm (GA) as an optimization method. GA initiates a population of DNN trainable parameters, making the DNN training process more streamlined on a single-generation GA. We focused on DNN-GA hybrid solution for linear and nonlinear ordinary differential equations (ODE‘s). Our method is achieved by solving several test problems and demonstrating a comparison with stochastic gradient descent (SGD) optimizers. In most ODE problems, using the same number of iterations and architecture, DNN-GA produces a better loss function than using DNN-SGD.

With the development of science and technology, there has been growing the grid size, the growing demand for electricity and the extent of increasing the electricity market.In this paper, a mathematical morphology and genetic algorithm based dynamic reactive power optimization method for distribution network is proposed.By means of constructing mathematical morphology filter, the binary image composed by allelic genes of chromosome is filtered and the problem that deals with the action times of voltage regulating devices for compensation is turned into filtering problem of discrete binary image.With further automate scheduling, distribution automation and unmanned substation reactive power optimization of distribution line running made an urgent request.Reactive power optimization can improve the voltage passing rate, increasing the reliability of the system, the power quality, and the system runs the security and economy of the perfect combination, very broad application prospects.

Application of Monte Carlo stochastic optimization (MOST) to deep learning

This article extends the optimal covariance steering (CS) problem for linear stochastic systems subject to chance constraints so as to account for an optimal allocation of the risk. Previous works have assumed a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">uniform</i> risk allocation in order to cast the optimal control problem as a semidefinite program, which can be solved efficiently using standard SDP solvers. An iterative risk allocation (IRA) formalism is used to solve the optimal risk allocation problem for CS using a two-stage approach. The upper-stage of IRA optimizes the risk, which is a convex problem, while the lower-stage optimizes the controller with the new constraints. The process is applied iteratively until the optimal risk allocation that achieves the lowest total cost is obtained. The proposed framework results in solutions that tend to maximize the terminal covariance, while still satisfying the chance constraints, thus leading to less conservative solutions than previous methodologies. In this article, we consider both polyhedral and cone state chance constraints. Finally, we demonstrate the approach to a spacecraft rendezvous problem and compare the results with other competing approaches.

In order to study the prediction and optimization of borehole stability parameters in deepwater drilling based on genetic algorithm. First, a genetic hybrid algorithm based on pattern search is proposed. Then, based on the adaptive genetic algorithm, the evolutionary population is searched for patterns, which makes the hybrid algorithm not only has a strong global search ability but also improves the local optimization accuracy. Finally, the unit footage cost in the drilling process is taken as the objective function, and the algorithm is verified by taking the drilling in Karamay area as an example. The calculation results show that if the bit wear reaches 0.8-0.9 and then the bit is pulled out, the utilization rate of the bit can be increased, the design efficiency and accuracy can be improved, and the drilling cost can be reduced. The wear amount of the optimized bit is higher than that of the actual bit. Increasing the utilization rate of the bit can reduce the cost of drilling meters to a certain extent and improve the economic benefits of drilling. The objective function and constraint conditions for the optimization of drilling parameters are determined, and the algorithm is verified with the drilling data of Karamay Oilfield. The results show that the algorithm improves the stability and speed of iterative convergence and improves the reliability of data analysis results. Based on the regional three-dimensional formation rock parameter data volume, the optimization method can be used to optimize the drilling parameters before drilling and provide a basis for formulating the drilling design scheme.

The article develops an effective method for solving optimization problems based on the synthesis of evolution ary mechanisms of nature’s development, which are based on the principles of genetic search for the best solu tions. It analyzes evolutionary concepts of biological species development by such well-known scientists as Ch. Darwin, J. Lamarck, Hugo de Vries, K. Popper, and M. Kimura. The key mechanisms for the emergence of new individuals with better adaptation (the best solutions) are identified. Based on the known relevant concepts, the main provisions of the theory of evolution are developed. Corresponding models are constructed, which in turn become computational analogies for the development of evolutionary methods for solving optimization prob lems. An optimization method has been developed based on a genetic algorithm, which is based on the basic operators of evolution: reproduction (selection), crossover, and mutation. A distinctive feature of the proposed approach is the hybridization of the classical genetic algorithm with adaptive mechanisms for parameter tuning and local improvement of solutions. The genetic algorithm used reproduction operators (tournament and roulette wheel selection), single- and double-point crossover, and mutation. This allows for an increase in the efficiency of global search in terms of convergence (number of computational iterations) and solution accuracy (average absolute error of the solution at 100 runs), as well as avoiding the stopping of the computational process at local extrema. Based on the developed optimization method, a genetic algorithm has been created that incorporates all the mechanisms of evolutionary computation. A genetic algorithm has been developed that contains all the mech anisms of evolutionary computation. Based on the genetic algorithm, model problems of multi-criteria optimi sation were calculated in Python in binary coding of the optimal solution (OneMax, LeadingOnes) and in coding of the solution using real numbers (two-extreme function). In the corresponding test problems, the stable achieve ment of the global extremum of the objective function and the stability of the algorithm were recorded. This allows us to conclude that the proposed method of optimizing multi-criteria functions based on genetic algo rithms is effective.

COVID-19 is a type of an infectious disease that is caused by the new coronavirus. The spread of COVID-19 needs to be suppressed because COVID-19 can cause death, especially for sufferers with congenital diseases and a weak immune system. COVID-19 spreads through direct contact, wherein the infected individual spreads the COVID-19 virus through cough, sneeze, or close contacts. Predicting the number of COVID-19 sufferers becomes an important task in the effort to curb the spread of COVID-19. Artificial neural network (ANN) is the prediction method that delivers effective results in doing this job. Backpropagation, a type of ANN algorithm, offers predictive problem solving with good performance. However, its performance depends on the optimization method applied during the training process. In general, the optimization method in ANN is the gradient descent method, which is known to have a slow convergence rate. Meanwhile, the Fletcher–Reeves method has a faster convergence rate than the gradient descent method. Based on this hypothesis, this paper proposes a prediction model for the number of COVID-19 sufferers in Malang using the Backpropagation neural network with the Fletcher–Reeves method. The experimental results show that the Backpropagation neural network with the Fletcher–Reeves method has a better performance than the Backpropagation neural network with the gradient descent method. This is shown by the Means Square Error (MSE) resulting from the proposed method which is smaller than the MSE resulting from the Backpropagation neural network with the gradient descent method.

An optimization method of adjoint assimilation based on the cumulative gradient: An application for correction of the bottom friction coefficient in the tidal wave model

PurposeLarge infrastructure projects involve many different stakeholders. Risk allocation is negotiated between the public sector and the special purpose vehicle (SPV)/consortium. Moreover, risk allocation is negotiated between the shareholders of the SPV and between the SPV and its contractors, financiers/lenders and insuring companies. Since the individual understanding of risks associated with the project differs amongst the contract partners, negotiation is often very time consuming and may not lead to the best possible outcome. The variability of risk management approaches and a lack of coherent practices often prove to be problematic. The purpose of this paper is to address these issues.Design/methodology/approachThe study has been conducted using a questionnaire and expert interviews to identify the current state of public‐private partnership (PPP) risk management. As a result, a process model for risk management for each stakeholder has been developed.FindingsBy developing an integrated risk management system (IRMS), it is possible to meet particular project requirements, such as reducing complexity, establishing risk awareness, covering the different perspectives of the parties involved, building on a common understanding regarding the nature of risks and opportunities providing step‐by‐step contingency instructions, and enhancing the available information.Originality/valueThe provision of an IRMS in PPP will induce transparency by utilizing better sources of data and information distribution. Moreover, the application of sophisticated risk management techniques will enhance the capabilities and accuracy of the decision‐making processes. As a result, credibility and a common understanding of key data, ratios and reports for all contract partners is achieved.

As a breakthrough of the additive manufacturing technology being achieved, many fields have broadly applied laser cladding due to its unique advantages. But the surface characteristics of the cladding layer are not frequently aligned with the standards necessary for industrial use. Consequently, with a particular focus on refining its surface roughness, it has emerged as a significant area of interest among numerous investigators. This paper reviews a variety of methods for optimizing the surface roughness of laser cladding, covering from deterministic algorithms such as Taguchi's method, orthogonal experimental method, gradient descent method, to stochastic algorithms including neural network, genetic algorithm, Gray Wolf algorithm, and even hybrid algorithms combining multiple algorithms like neural network genetic algorithm, adaptive neural fuzzy reasoning algorithm, and improved genetic algorithms for response surface analysis, and so on. Through comparative analysis, it is found that the hybrid algorithms can quickly generate the optimal optimization parameters for the sake of achieving the optimal surface quality since they may combine the accuracy of deterministic algorithms and the robustness of stochastic algorithms. In addition, this paper also looks forward to the future development direction of surface quality optimization methods for laser cladding, aiming at laying a foundation for the research work of high-quality coating preparation.