Polyhedral Conic Classifier for CTR Prediction

TL;DR

This paper proposes a polyhedral conic neural network classifier for CTR prediction that effectively handles dataset imbalance and distribution asymmetry, outperforming traditional methods across multiple datasets.

Contribution

Introduces a novel polyhedral conic neural network classifier specifically designed for CTR prediction, addressing imbalance and asymmetry challenges with superior performance.

Findings

Outperforms BCE loss in CTR tasks

Effective on multiple public datasets

Handles data imbalance and geometric asymmetry

Abstract

This paper introduces a novel approach for click-through rate (CTR) prediction within industrial recommender systems, addressing the inherent challenges of numerical imbalance and geometric asymmetry. These challenges stem from imbalanced datasets, where positive (click) instances occur less frequently than negatives (non-clicks), and geometrically asymmetric distributions, where positive samples exhibit visually coherent patterns while negatives demonstrate greater diversity. To address these challenges, we have used a deep neural network classifier that uses the polyhedral conic functions. This classifier is similar to the one-class classifiers in spirit and it returns compact polyhedral acceptance regions to separate the positive class samples from the negative samples that have diverse distributions. Extensive experiments have been conducted to test the proposed approach using state-of-the-art (SOTA) CTR prediction models on four public datasets, namely Criteo, Avazu, MovieLens and Frappe. The experimental evaluations highlight the superiority of our proposed approach over Binary Cross Entropy (BCE) Loss, which is widely used in CTR prediction tasks.