Two phase cooperative learning for supervised dimensionality reduction

Ioannis A. Nellas, Sotiris K. Tasoulis, Spiros V. Georgakopoulos, Vassilis P. Plagianakos

Abstract: Highlights•A novel cooperative framework for supervised dimensionality reduction is proposed.•Minimal parameter tuning is required, while enabling direct error minimization.•Boost classifier performance by utilizing its optimized latent representations.•Explainability, image generation, and classification boundaries are studied.•Thorough experimental analysis is performed to showcase applicability.