Learned neural control improves prediction with hidden gain adaptation
History, Adaptation, and Hidden Gain: Two Prospective Tests of Learned Predictive Control in a Synthetic Neural Population
Machine Learning
Summary
Controlling systems by predicting their next step can be tricky when the system has hidden factors that change how it reacts to inputs. These authors studied a computer model of neurons whose sensitivity to light was scaled by a hidden number. They found that including some memory of past inputs helps control accuracy, especially when the hidden sensitivity is low or moderate. They showed that estimating the hidden gain value on the fly and adding it as a network input greatly improves predictions and control. However, when noise is high, these gains are much smaller.
What this means in practice
- •For control system engineers: Improve controller performance in systems with unknown or changing input gains by adding online gain estimation to predictive neural control models.
- •For signal processing teams: Use recursive adaptive models with online parameter updates to better predict outputs in dynamic systems affected by hidden sensitivities.
Tested on simulated data.
Authors
Mahesh Reddy Pagadala
Abstract
We study learned one-step predictive control of a synthetic population of 64 leaky integrate-and-fire neurons whose light sensitivity is scaled by a hidden factor s, using two prospective confirmations on plants simulated only after the analysis was frozen. In the first (50 plants, 11 conditions), removing the 200-ms input-output history substantially worsened a parameter-count-matched neural predictor at low and moderate gain (s = 0.6, 0.8; RMSE rose by 2.47 Hz at s = 0.6, 97.5% CI 2.43-2.51). In secondary comparisons, recursively adapted ARX had the lowest mean error of the deployable controllers in 10 of 11 conditions, tuned PI with anti-windup matched or beat the network for s >= 0.7, and training for 10,000 rather than 900 epochs lowered prediction error but worsened control. Development work suggested the network's learned action sensitivity did not track the hidden gain. A second protocol (50 new plants, 9 sensitivities, 2 noise levels) confirmed three pre-specified predictions at the primary noise level. Re-estimating a single multiplicative input gain recovered 87% (95% CI 86-88%) of the benefit of full adaptive ARX. Supplying that online estimate to the network as an extra input reduced high-gain tracking error by 1.84 Hz [1.81, 1.88]. Longer training worsened the plain network by 1.09 Hz but improved the gain-input network by 1.07 Hz, a differential of 2.16 Hz [2.13, 2.20]. Under high process noise (sigma = 2.0) the effects shrank to about 0.04 Hz or less and gain-only adaptation recovered only 24% of the adaptive-ARX benefit. In this synthetic plant, at the primary noise level, one-parameter online gain estimation accounts for most of the gap between a history-based network and full adaptive ARX, and for its prediction-control mismatch.