General limited information diffusion method of small-sample information analysis in insurance

When analyzing and evaluating risks in insurance, people are often confronted with the situation of incomplete information and insufficient data, which is known as a small-sample problem. In this paper, a one-dimensional small-sample problem in insurance was investigated using the kernel density estimation method （KerM） and general limited information diffusion method （GIDM）. In particular, MacCormack technique was applied to get the solutions of GIDM equations and then the optimal diffusion solution was acquired based on the two optimization principles. Finally, the analysis introduced in this paper was verified by treating some examples and satisfying results were obtained.

Analysis method on shoot precision of weapon in small-sample case

Because of limits of cost, in general, the test data of weapons are shortness. It is always an important topic that to gain scientific results of weapon performance analyses in small-sample case. Based on the analysis of distribution function characteristics and grey mathematics, a weighting grey method in small-sample case is presented. According to the analysis of test data of a weapon, it is proved that the method is a good method to deal with data in the small-sample case and has a high value in the analysis of weapon performance.

Gyroscope Dynamic Balance Counterweight Prediction Based on Multi-Head ResGAT Networks

The dynamic balance assessment during the assembly of the coordinator gyroscope significantly impacts the guidance accuracy of precision-guided equipment.In dynamic balance debugging,reliance on rudimentary counterweight empirical formulas persists,resulting in suboptimal debugging accuracy and an increased repetition rate.To mitigate this challenge,we present a multi-head residual graph attention network(ResGAT)model,designed to predict dynamic balance counterweights with high precision.In this research,we employ graph neural networks for interaction feature extraction from assembly graph data.An SDAE-GPC model is designed for the assembly condition classification to derive graph data inputs for the ResGAT regression model,which is capable of predicting gyroscope counterweights under small-sample conditions.The results of our experiments demonstrate the effectiveness of the proposed approach in predicting dynamic gyroscope counterweight in its assembly process.Our approach surpasses current methods in mitigating repetition rates and enhancing the assembly efficiency of gyroscopes.

DCGAN Based Spectrum Sensing Data Enhancement for Behavior Recognition in Self-Organized Communication Network

Communication behavior recognition is an issue with increasingly importance in the antiterrorism and national defense area.However,the sensing data obtained in actual environment is often not sufficient to accurately analyze the communication behavior.Traditional means can hardly utilize the scarce and crude spectrum sensing data captured in a real scene.Thus,communication behavior recognition using raw sensing data under smallsample condition has become a new challenge.In this paper,a data enhanced communication behavior recognition(DECBR)scheme is proposed to meet this challenge.Firstly,a preprocessing method is designed to make the raw spectrum data suitable for the proposed scheme.Then,an adaptive convolutional neural network structure is exploited to carry out communication behavior recognition.Moreover,DCGAN is applied to support data enhancement,which realize communication behavior recognition under small-sample condition.Finally,the scheme is verified by experiments under different data size.The results show that the DECBR scheme can greatly improve the accuracy and efficiency of behavior recognition under smallsample condition.

An Improved Algorithm for Imbalanced Data and Small Sample Size Classification

Traditional classification algorithms perform not very well on imbalanced data sets and small sample size. To deal with the problem, a novel method is proposed to change the class distribution through adding virtual samples, which are generated by the windowed regression over-sampling (WRO) method. The proposed method WRO not only reflects the additive effects but also reflects the multiplicative effect between samples. A comparative study between the proposed method and other over-sampling methods such as synthetic minority over-sampling technique (SMOTE) and borderline over-sampling (BOS) on UCI datasets and Fourier transform infrared spectroscopy (FTIR) data set is provided. Experimental results show that the WRO method can achieve better performance than other methods.

Sealing reliability modeling of aviation seal based on interval uncertainty method and multidimensional response surface

Many variables affect the sealing performance, and their distribution characteristics are difficult to obtain with probabilistic methods owing to the high cost involved. Numerous problems in engineering are similar due to the appearance of small-sample parameters. In this study, the sealing reliability of an aviation seal was defined as the research object, and an interval uncertainty method and multidimensional response surface were proposed to calculate the sealing reliability.Based on this, we first analyzed the failure mechanism of the aviation seal and established a leakage rate model. Then, based on the non-probabilistic interval model, an interval uncertainty method was proposed to construct the analytical model. With reference to the limit state equation from the structural reliability theory, the multidimensional response surface was used for fast calculation.Then, we chose the single-cylinder gas steering gear used in aircraft as the case study, its sealing reliability in working and non-working statuses were calculated, and the results were verified with the actual maintenance records. By analyzing the sensitivity of some variables, we can improve the sealing reliability of the aviation seal by improving the surface roughness only if the cost allows.Finally, we consider that the method proposed in this study realizes the application of smallsample uncertainty analysis in reliability analysis, and could provide a feasible way to solve the similar problems in engineering with multidimensional and small-sample parameters.

A generic and extensible model for the martensite start temperature incorporating thermodynamic data mining and deep learning framework

The martensite start temperature is a critical parameter for steels with metastable austenite.Although numerous models have been developed to predict the martensite start(Ms)temperature,the complexity of the martensitic transformation greatly limits their performance and extensibility.In this work,we apply deep data mining of thermodynamic calculations and deep learning to develop a generic model for Msprediction.Deep data mining was used to establish a hierarchical database with three levels of information.Then,a convolutional neural network model,which can accurately treat the hierarchical data structure,was used to obtain the final model.By integrating thermodynamic calculations,traditional machine learning and deep learning modeling,the final predictor model shows excellent generalizability and extensibility,i.e.model performance both within and beyond the composition range of the original database.The effects of 15 alloying elements were considered successfully using the proposed methodology.The work suggests that,with the help of deep data mining considering the physical mechanisms,deep learning methods can partially mitigate the challenge with limited data in materials science and provide a means for solving complex problems with small databases.