Implicit Regularization and Momentum Algorithms in Nonlinearly Parameterized Adaptive Control and Prediction
arXiv:1912.13154 · doi:10.1162/neco_a_01360
Abstract
Stable concurrent learning and control of dynamical systems is the subject of adaptive control. Despite being an established field with many practical applications and a rich theory, much of the development in adaptive control for nonlinear systems revolves around a few key algorithms. By exploiting strong connections between classical adaptive nonlinear control techniques and recent progress in optimization and machine learning, we show that there exists considerable untapped potential in algorithm development for both adaptive nonlinear control and adaptive dynamics prediction. We begin by introducing first-order adaptation laws inspired by natural gradient descent and mirror descent. We prove that when there are multiple dynamics consistent with the data, these non-Euclidean adaptation laws implicitly regularize the learned model. Local geometry imposed during learning thus may be used to select parameter vectors -- out of the many that will achieve perfect tracking or prediction -- for desired properties such as sparsity. We apply this result to regularized dynamics predictor and observer design, and as concrete examples, we consider Hamiltonian systems, Lagrangian systems, and recurrent neural networks. We subsequently develop a variational formalism based on the Bregman Lagrangian. We show that its Euler Lagrange equations lead to natural gradient and mirror descent-like adaptation laws with momentum, and we recover their first-order analogues in the infinite friction limit. We illustrate our analyses with simulations demonstrating our theoretical results.
arXiv commentsync title and abstract with journal version, minor cosmetic re-arrangements to text
References in corpus (20)
- Reconciling modern machine learning practice and the bias-variance trade-off
- Data-driven discovery of coordinates and governing equations
- On the Complexity of Best Arm Identification in Multi-Armed Bandit Models
- Benign Overfitting in Linear Regression
- A Differential Equation for Modeling Nesterov's Accelerated Gradient Method: Theory and Insights
- A Variational Perspective on Accelerated Methods in Optimization
- Characterizing Implicit Bias in Terms of Optimization Geometry
- Linear Matrix Inequalities for Physically-Consistent Inertial Parameter Identification: A Statistical Perspective on the Mass Distribution
- A Lyapunov Analysis of Momentum Methods in Optimization
- Predicting non-linear dynamics by stable local learning in a recurrent spiking neural network
- Symplectic Recurrent Neural Networks
- A Dynamical Systems Perspective on Nesterov Acceleration
- Learning One Convolutional Layer with Overlapping Patches
- Beyond Convexity -- Contraction and Global Convergence of Gradient Descent
- Hamiltonian Descent Methods
- Stochastic Mirror Descent on Overparameterized Nonlinear Models: Convergence, Implicit Regularization, and Generalization
- Learning Stability Certificates from Data
- Learning nonlinear dynamical systems from a single trajectory
- Contraction Metrics in Adaptive Nonlinear Control
- A continuous-time analysis of distributed stochastic gradient