Learning to Warm-Start Fixed-Point Optimization Algorithms
This work addresses the computational bottleneck in optimization for various domains, though it is incremental as it builds on existing fixed-point methods with a learned warm-start approach.
The authors tackled the problem of accelerating fixed-point optimization algorithms by introducing a machine-learning framework that predicts warm starts, which significantly reduces the number of iterations and solution time in applications like control, statistics, and signal processing.
We introduce a machine-learning framework to warm-start fixed-point optimization algorithms. Our architecture consists of a neural network mapping problem parameters to warm starts, followed by a predefined number of fixed-point iterations. We propose two loss functions designed to either minimize the fixed-point residual or the distance to a ground truth solution. In this way, the neural network predicts warm starts with the end-to-end goal of minimizing the downstream loss. An important feature of our architecture is its flexibility, in that it can predict a warm start for fixed-point algorithms run for any number of steps, without being limited to the number of steps it has been trained on. We provide PAC-Bayes generalization bounds on unseen data for common classes of fixed-point operators: contractive, linearly convergent, and averaged. Applying this framework to well-known applications in control, statistics, and signal processing, we observe a significant reduction in the number of iterations and solution time required to solve these problems, through learned warm starts.