Neural Flows: Efficient Alternative to Neural ODEs
This work addresses computational bottlenecks for researchers and practitioners using neural ODEs in sequential data applications, offering an efficient alternative with empirical gains.
The authors tackled the computational inefficiency of neural ODEs by proposing neural flows, which directly model solution curves to eliminate expensive numerical solvers, achieving improved computational efficiency and favorable generalization in tasks like time series modeling and density estimation.
Neural ordinary differential equations describe how values change in time. This is the reason why they gained importance in modeling sequential data, especially when the observations are made at irregular intervals. In this paper we propose an alternative by directly modeling the solution curves - the flow of an ODE - with a neural network. This immediately eliminates the need for expensive numerical solvers while still maintaining the modeling capability of neural ODEs. We propose several flow architectures suitable for different applications by establishing precise conditions on when a function defines a valid flow. Apart from computational efficiency, we also provide empirical evidence of favorable generalization performance via applications in time series modeling, forecasting, and density estimation.