Victor Quintanar-Zilinskas

This paper introduces an optimal-fitness framework to reconcile different models of perceptual compression by evaluating their evolutionary fitness in various environments.

This paper develops a neuromimetic network that simulates multi-environment learning with memory retention and plasticity, showing how reinforcement and pruning balance to enable stable, adaptive representations.

The paper shows that natural selection aligns with perceptual compression theories by identifying utility-weighted distortion minimization as the optimal adaptation to decoding noise.