單領域監督之跨領域深度解離特徵學習

Abstract

深度生成模型於電腦視覺與機器學習領域中,近期有顯著發展與影響。特徵解離主要在不可分析之潛在向量中分離出具有語義之個別特徵,傳統方法多為監督學習框架,少數非監督學習框架則無法確保解離語義之穩定性。本篇論文中,我們將提出深度生成類神經網路架構,在單邊領域監督之下,達成跨領域解離語義特徵學習,同時在本文中,我們應用非監督領域適應之概念,學習共同特徵解離與適應。藉由生成對抗學習架構,本文將出新式具特徵解離能力之深度學習架構,此架構將同時訓練於跨領域資料,學習出具有共同語義之分離特徵,進而在生成模型框架之下,完成單領域監督之跨領域深度解離特徵學習。本文實驗中,我們利用此深度生成架構,將原始輸入影像於潛在空間空改變屬性後,生成對應屬性之跨領域影像。同時也將呈現單邊監督情況之下,利用此深度網路架構,完成雙邊領域個別之影像分類,解決非監督領域適應影像分類問題。

The recent progress and development of deep generative models have led to remarkable improvements in research topics in computer vision and machine learning. In this article, the task of cross-domain feature disentanglement is addressed. This thesis advances the idea of unsupervised domain adaptation and propose to perform joint feature disentanglement and adaptation. Based on generative adversarial networks, a novel deep learning architecture with disentanglement ability is presented, which observes cross-domain image data and derives latent features with the underlying factors(e.g., attributes). As a result, our generative model is able to address cross-domain feature disentanglement with only the (attribute) supervision from the source-domain data (not the target-domain ones). In the experiments, the model is applied for generating and classifying images with particular attributes, and show that satisfactory results can be produced.