Learning Classifiers from Synthetic Data Using a Multichannel Autoencoder

TL;DR

This paper introduces a Multichannel Autoencoder (MCAE) to effectively learn classifiers from synthetic data by bridging the distribution gap between synthetic and real data, improving feature representation for classification.

Contribution

The paper presents a novel Multichannel Autoencoder (MCAE) that jointly learns from synthetic and real data to enhance classifier performance.

Findings

MCAE improves feature representation for classification.

Experimental results validate the effectiveness of MCAE on two datasets.

Synthetic data generation combined with MCAE enhances classifier accuracy.

Abstract

We propose a method for using synthetic data to help learning classifiers. Synthetic data, even is generated based on real data, normally results in a shift from the distribution of real data in feature space. To bridge the gap between the real and synthetic data, and jointly learn from synthetic and real data, this paper proposes a Multichannel Autoencoder(MCAE). We show that by suing MCAE, it is possible to learn a better feature representation for classification. To evaluate the proposed approach, we conduct experiments on two types of datasets. Experimental results on two datasets validate the efficiency of our MCAE model and our methodology of generating synthetic data.