Generating Multi-type Temporal Sequences to Mitigate Class-imbalanced Problem

TL;DR

This paper proposes two novel GAN training methods, RL-based and Gumbel-Softmax, to generate multi-type temporal sequences, effectively addressing class imbalance issues in user activity modeling for ad networks.

Contribution

It introduces two multi-type training approaches for GANs, improving sequence generation for imbalanced classes in user activity analysis.

Findings

Generated sequences meet multiple desired criteria

GANs trained with proposed methods produce realistic multi-type sequences

Approaches effectively mitigate class imbalance in synthetic data

Abstract

From the ad network standpoint, a user's activity is a multi-type sequence of temporal events consisting of event types and time intervals. Understanding user patterns in ad networks has received increasing attention from the machine learning community. Particularly, the problems of fraud detection, Conversion Rate (CVR), and Click-Through Rate (CTR) prediction are of interest. However, the class imbalance between major and minor classes in these tasks can bias a machine learning model leading to poor performance. This study proposes using two multi-type (continuous and discrete) training approaches for GANs to deal with the limitations of traditional GANs in passing the gradient updates for discrete tokens. First, we used the Reinforcement Learning (RL)-based training approach and then, an approximation of the multinomial distribution parameterized in terms of the softmax function (Gumble-Softmax). Our extensive experiments based on synthetic data have shown the trained generator can generate sequences with desired properties measured by multiple criteria.