The XXI International Scientific Conference of Young Scientists and Specialists (AYSS-2017)

Using neural networks to calculate integrals

Not scheduled

Oral Mathematical Modeling and Computational Physics

Speaker

Mr Dmitrii Melnitskii (SPBU)

Description

Nowadays we spend a lot of time for calculation n-dimensional integrals, using different methods, like Monte-Carlo and Gauss methods. Computations by these methods expend a lot of computer resources, which is leading to the problem of golden mean between computing speed and resource employment. In our research work we studied a new method for calculation n-dimensional integrals, which is based on neural networks. We tested different schemes of constructing neural network and compared them by the following parameters: • Learning speed • Required memory • Calculation of integral after learning • Usability Probably, in the future, method of calculation n-dimensional integrals with the help of neural networks will keep up with the other methods.