An Ode to an ODE
This addresses training instability in deep neural networks like Neural ODEs, offering a novel solution with broad applicability, though it builds on existing ODE methods.
The paper tackles the gradient vanishing-explosion problem in Neural ODEs by introducing ODEtoODE, a new paradigm where parameters evolve on the orthogonal group O(d), leading to stable training and improved downstream models, with empirical gains over SOTA baselines in reinforcement learning and supervised learning.
We present a new paradigm for Neural ODE algorithms, called ODEtoODE, where time-dependent parameters of the main flow evolve according to a matrix flow on the orthogonal group O(d). This nested system of two flows, where the parameter-flow is constrained to lie on the compact manifold, provides stability and effectiveness of training and provably solves the gradient vanishing-explosion problem which is intrinsically related to training deep neural network architectures such as Neural ODEs. Consequently, it leads to better downstream models, as we show on the example of training reinforcement learning policies with evolution strategies, and in the supervised learning setting, by comparing with previous SOTA baselines. We provide strong convergence results for our proposed mechanism that are independent of the depth of the network, supporting our empirical studies. Our results show an intriguing connection between the theory of deep neural networks and the field of matrix flows on compact manifolds.