遷移類神經樹:在異質領域適應的應用與延伸

Abstract

本論文提出了一異質領域適應(HeterogeneousDomainAdaptation, HDA)的演算法。異質領域適應旨在找出由不同特徵所描述的領域資 料間的關聯性。受最近蓬勃發展的類神經網路與深度學習的啟發， 我們提出了遷移類神經樹(TransferNeuralTrees,TNT)，將跨領域的特 徵投影、適應、以及辨識整合於一個類神經網路架構之中。在其中 的辨識層，我們提出了遷移學習版本的類神經森林(Transfer-Neural DecisionForest)，以機率剪枝(stochasticpruning)的技巧讓我們架構中的神經元能夠更加適應於領域的差異。而為了有效利用半監督式的異質領域適應問題內所擁有的資訊，我們提出了一個獨特的嵌入誤差函 數(embeddinglossterm)來保存有標記(labeled)與無標記(unlabeled)目標領域資料(targetdomaindata)間，預測結果與投影後結構的一致性。我們進一步將我們的演算法延伸至零樣本學習(zero-shotlearning)，透過找出影像與屬性資料間的關聯來得到良好的表現。最後，我們將會進行跨特徵、跨資料來源、跨型態的異質領域適應實驗，來證明我們所提出的遷移類神經樹的能力。

This thesis presents a novel algorithm for Heterogeneous domain adaptation (HDA). HDA addresses the task of associating data not only across dissimilar domains but also described by different types of features. Inspired by the recent advances of neural networks and deep learning, we propose a deep learning model of Transfer Neural Trees (TNT), which jointly solves cross-domain feature mapping, adaptation, and classification in a unified architecture. As the prediction layer in TNT, we introduce Transfer Neural Decision Forest (Transfer-NDF), which is able to learn the neurons in TNT for adaptation by stochastic pruning. In order to handle semi-supervised HDA, a unique embedding loss term is introduced to TNT for preserving prediction and structural consistency between labeled and unlabeled target-domain data. We further show that our TNT can be extended to zero shot learning for associating image and attribute data with promising performance. Finally, experiments on different classification tasks across features, datasets, and modalities would verify the effectiveness of our TNT.