This library provides differentiable GPU-capable solvers for controlled differential equations (CDEs). Backpropagation through the solver or via the adjoint method is supported; the latter allows for improved memory efficiency.
In particular this allows for building Neural Controlled Differential Equation models, which are state-of-the-art models for (arbitrarily irregular!) time series. Neural CDEs can be thought of as a "continuous time RNN".
Powered by the
Don't forget to tag @patrick-kidger in your comment, otherwise they may not be notified.