Feedback Control for Online Training of Neural Networks
This addresses the problem of inefficient online training for neural networks in image classification, offering a robust method for practitioners, though it is incremental as it builds on existing control theory concepts.
The paper tackles the challenge of adapting neural network training to data flows by proposing a performance-based learning rate strategy called E/PD-Control, which combines feedback control with exponential signals. Results show faster accuracy growth and higher performance levels on CIFAR-10 and Fashion-MNIST datasets compared to related works.
Convolutional neural networks (CNNs) are commonly used for image classification tasks, raising the challenge of their application on data flows. During their training, adaptation is often performed by tuning the learning rate. Usual learning rate strategies are time-based i.e. monotonously decreasing. In this paper, we advocate switching to a performance-based adaptation, in order to improve the learning efficiency. We present E (Exponential)/PD (Proportional Derivative)-Control, a conditional learning rate strategy that combines a feedback PD controller based on the CNN loss function, with an exponential control signal to smartly boost the learning and adapt the PD parameters. Stability proof is provided as well as an experimental evaluation using two state of the art image datasets (CIFAR-10 and Fashion-MNIST). Results show better performances than the related works (faster network accuracy growth reaching higher levels) and robustness of the E/PD-Control regarding its parametrization.