On the Stability and Realizability of Recurrent Polynomial Surrogate Ternary Logic Gate Networks

ArXi:2605.24649v1 Announce Type: cross Recurrent Neural Networks (RNNs) can learn to predict Signal Temporal Logic (STL) verdicts online from partial trajectories, but deploying them as runtime monitors in safety-critical systems demands than predictive accuracy. Standard RNN architectures offer no structural guarantee that outputs degrade gracefully under sensor degradation; a dropped input can silently flip a verdict from safe to unsafe. We