Preference Neural Network
Ayman Elgharabawy, Mukesh Prasad, Chin‐Teng Lin · IEEE Transactions on Emerging Topics in Computational Intelligence · 2023
This paper proposes a novel label ranker network to learn the relationship between labels to solve ranking and classification problems. The Preference Neural Network (PNN) usesspearmancorrelation gradient ascent and two new activation functions, positive smooth staircase (PSS), and smooth staircase (SS) that accelerate the ranking by creating almost deterministic preference values.PNNis proposed in two forms, fully connected simple Three layers and Preference Net (PN), where the latter is the deep ranking form ofPNNto learning feature selection using ranking to solve images classification problem.PNuses a new type of ranker kernel to generate a feature map.PNNoutperforms five previously proposed methods for label ranking, obtaining state-of-the-art results on label ranking, andPNachieves promising results onCFAR-100with high computational efficiency.