A q-deformed deterministic gradient framework generalizing affine projection adaptive filters
Juan Gerardo Avalos, Brayans Becerra, Erick Lee, Brandon Pineda, Giovanny Sánchez, Eliseo Sarmiento, Ángel VázquezPurpose
This work proposes a q-deformed deterministic gradient framework that generalizes affine projection adaptive filters by embedding a deformation of the classical gradient operator into a fully deterministic formulation.
Design/methodology/approach
By embedding the q-gradient into an affine projection–like structure, we derive a closed form update that preserves projection geometry while incorporating a feature-wise diagonal correction controlled by q.
Findings
The resulting deterministic q-APL recursion eliminates probabilistic assumptions and is specifically designed for coherent periodic disturbances.
Originality/value
We present a new deterministic q-APL algorithm for adaptive filtering. Experimental results on demonstrate that the proposed framework achieves strong attenuation of periodic interference, preserves waveform structure, and maintains convergence properties comparable to stochastic adaptive algorithms under both nominal and distorted interference conditions.