One-Shot Adaptation of Supervised Deep Convolutional Models
Judy Hoffman, Eric Tzeng, Jeff Donahue UC Berkeley, EECS、ICSI、Yangqing Jia Google Research \ANDKate Saenko UMass Lowell, CS、ICSI、Trevor Darrell UC Berkeley, EECS、ICSI This work was completed while Yangqing Jia was a graduate student at UC Berkeley
Abstract
Dataset bias remains a significant barrier towards solving real world computer vision tasks. Though deep convolutional networks have proven to be a competitive approach for image classification, a question remains: have these models have solved the dataset bias problem? In general, training or fine-tuning a state-of-the-art deep model on a new domain requires a significant amount of data, which for many applications is simply not available. Transfer of models directly to new domains without adaptation has historically led to poor recognition performance. In this paper, we pose the following question: is a single image dataset, much larger than previously explored for adaptation, comprehensive enough to learn general deep models that may be effectively applied to new image domains? In other words, are deep CNNs trained on large amounts of labeled data as susceptible to dataset bias as previous methods have been shown to be? We show that a generic supervised deep CNN model trained on a large dataset reduces, but does not remove, dataset bias. Furthermore, we propose several methods for adaptation with deep models that are able to operate with little (one example per category) or no l
中文速览
深度卷积神经网络（CNN）虽然在图像分类上表现出色，但换个拍摄场景或数据来源（即"域偏移"）后性能往往大幅下滑，这一"数据集偏差"问题至今悬而未决。研究者用包含120万张图片的大规模ImageNet数据集训练深度CNN，首次系统检验：用如此海量的源域数据训练出来的模型，是否还会受域偏差困扰？实验表明，大规模预训练确实能缓解偏差，但无法彻底消除它；直接微调（fine-tuning）在目标域标注极少时反而会降低性能。为此，作者提出一种简洁的适应框架：冻结预训练网络权重，只在其某层特征上训练一个轻量的域自适应分类器，并系统比较了多种有监督与无监督自适应算法。结果显示，即便每个类别只有一张目标域标注图片（one-shot），该框架也能将目标域准确率提升至与源域相当的水平，为实际场景中标注稀缺时的视觉域自适应提供了切实可行的解决思路。
原文 arXiv:1312.6204;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1312.6204v2