Learning convolutional neural network to maximize Pos@Top performance measure

TL;DR

This paper introduces a CNN-based approach to directly optimize the Pos@Top performance measure, effectively ranking positive data points at the top, and demonstrates superior results over existing methods.

Contribution

It proposes a novel CNN training framework that maximizes Pos@Top by jointly learning filters and classifier parameters through an iterative optimization process.

Findings

Outperforms state-of-the-art Pos@Top maximization methods

Effective in ranking positive instances at top positions

Applicable to benchmark datasets

Abstract

In the machine learning problems, the performance measure is used to evaluate the machine learning models. Recently, the number positive data points ranked at the top positions (Pos@Top) has been a popular performance measure in the machine learning community. In this paper, we propose to learn a convolutional neural network (CNN) model to maximize the Pos@Top performance measure. The CNN model is used to represent the multi-instance data point, and a classifier function is used to predict the label from the its CNN representation. We propose to minimize the loss function of Pos@Top over a training set to learn the filters of CNN and the classifier parameter. The classifier parameter vector is solved by the Lagrange multiplier method, and the filters are updated by the gradient descent method alternately in an iterative algorithm. Experiments over benchmark data sets show that the proposed method outperforms the state-of-the-art Pos@Top maximization methods.