Neural networks in solving differential equations
Viacheslav Gorikhovskii, T. O. Evdokimova, Viktor Poletansky
Abstract
Abstract The paper considers the possibility of using neural networks in solving differential equations and systems. The paper investigates the influence of various basic neural network topologies, various optimizers, activation functions, fast access links and some heuristic methods on the quality of neural network training and the accuracy of solving the differential equation.
Topics & Concepts
Artificial neural networkHeuristicDifferential equationComputer scienceNetwork topologyStochastic neural networkTopology (electrical circuits)MathematicsArtificial intelligenceTime delay neural networkMathematical analysisOperating systemCombinatoricsAdvanced Data Processing TechniquesReservoir Engineering and Simulation MethodsEngineering Diagnostics and Reliability