Quantizing Time-Series Models As Dynamical Systems: Trajectory-Based Quantization Sensitivity Score
This work addresses the need for efficient quantization of time-series models, particularly for black-box or compiled networks, by providing a calibration-free sensitivity metric.
The authors introduce the Trajectory-based Quantization Sensitivity Score (TQS), a metric that reframes post-training quantization through dynamical-systems stability, enabling a priori sensitivity estimation without calibration data. TQS-PTQ achieves robust mixed-precision quantization for time-series models, outperforming existing methods in resource-constrained settings.
We introduce the Trajectory-based Quantization Sensitivity Score (TQS), a metric that reframes post-training quantization (PTQ) through the lens of dynamical-systems stability. By modeling the network's rollout as a discrete-time dynamical system, TQS characterizes how quantization-induced errors propagate and amplify over the rollout horizon. Unlike conventional PTQ methods, where sensitivity analysis is often coupled to the quantization procedure, TQS enables a priori sensitivity estimation decoupled from quantizer selection and bit-width assignment. This separation allows for quantization budget planning even for black-box or compiled networks with fused operators. Building on this, we present TQS-PTQ, a flexible mixed-precision framework that requires no calibration data or costly second-order approximations. Our experiments show that a dynamical-systems perspective provides a robust, high-performing pathway for low-precision deployment in resource-constrained settings.