[论文解读] A Primer on Domain Adaptation
本文通过使用直观的几何与数学框架,严谨分析了四种关键情形——先验偏移、协变量偏移、概念偏移和子空间映射——对领域自适应(DA)进行了教学性综述。它推导出一种基于期望最大化(EM)的算法用于先验偏移自适应,表明可通过贝叶斯法则对源模型预测进行迭代重加权,从而估计目标域的类别先验。
Standard supervised machine learning assumes that the distribution of the source samples used to train an algorithm is the same as the one of the target samples on which it is supposed to make predictions. However, as any data scientist will confirm, this is hardly ever the case in practice. The set of statistical and numerical methods that deal with such situations is known as domain adaptation, a field with a long and rich history. The myriad of methods available and the unfortunate lack of a clear and universally accepted terminology can however make the topic rather daunting for the newcomer. Therefore, rather than aiming at completeness, which leads to exhibiting a tedious catalog of methods, this pedagogical review aims at a coherent presentation of four important special cases: (1) prior shift, a situation in which training samples were selected according to their labels without any knowledge of their actual distribution in the target, (2) covariate shift which deals with a situation where training examples were picked according to their features but with some selection bias, (3) concept shift where the dependence of the labels on the features defers between the source and the target, and last but not least (4) subspace mapping which deals with a situation where features in the target have been subjected to an unknown distortion with respect to the source features. In each case we first build an intuition, next we provide the appropriate mathematical framework and eventually we describe a practical application.
研究动机与目标
- 解决监督机器学习中训练数据(源域)与测试数据(目标域)之间分布偏移的问题,该问题破坏了标准机器学习的基本假设。
- 解决领域自适应(DA)中因术语不一致和概念重叠而引起的混淆,特别是与迁移学习(TL)的混淆。
- 为四种主要的DA场景提供一个连贯且概念一致的框架,强调清晰性而非对方法的 exhaustive 列举。
- 通过连接直觉、数学与实际应用,弥合理论统计学习与实际实现之间的差距。
提出的方法
- 使用期望最大化(EM)算法估计先验偏移场景下未知的目标类别先验,将标签分配视为隐变量。
- 通过假设类别条件特征密度在源域与目标域之间相同,推导出EM更新规则,以最大化联合分布的对数似然。
- 应用贝叶斯法则,将目标后验概率表示为对源后验的重加权,权重为目标与源类别先验的比值。
- 实施迭代更新:首先从源预测估计目标后验(E步),然后通过经验平均更新目标先验(M步)。
- 在优化过程中引入拉格朗日乘子,以确保概率单纯形约束(先验之和为1)得到满足。
- 推导出目标先验的闭式更新公式:$\theta_k^{(s+1)} = \frac{1}{m}\sum_{i=1}^{m} \hat{p}_T^{(s)}(\omega_k|\mathbf{x}_i)$,该公式源自EM过程。
实验结果
研究问题
- RQ1如何将一个在标签分布有偏的源域上训练的分类器,适应到目标域,其中目标域的类别先验未知但不同?
- RQ2何种数学框架可实现仅基于源数据与特征时对目标先验的一致估计?
- RQ3EM算法如何在存在隐式标签分配的情况下,实现对先验估计的迭代优化?
- RQ4目标与源类别先验的比值在何种方式下可校正分层抽样中的选择偏差?
- RQ5能否从基本原理出发,推导出一种有原则的、基于概率的领域自适应方法,而无需依赖临时的启发式方法?
主要发现
- EM算法通过结合贝叶斯法则与经验平均,迭代优化后验概率,成功估计了目标类别先验。
- 推导出的更新公式 $\theta_k^{(s+1)} = \frac{1}{m}\sum_{i=1}^{m} \hat{p}_T^{(s)}(\omega_k|\mathbf{x}_i)$ 在先验偏移假设下收敛至目标先验的最大似然估计。
- 该方法在EM框架下具有理论基础,在标准正则性条件下保证收敛。
- 该方法通过使用未知的目标先验比值对源后验进行重加权,正确地校正了分层抽样中的选择偏差。
- 推导表明,$\theta_k$ 的M步更新与当前估计下被分配至类别 $k$ 的样本的期望数量成正比,从而确保了概率归一化。
- 该方法为先验偏移场景下提供了有原则的替代方案,优于启发式重加权或校准技术。
更好的研究,从现在开始
从阅读论文到最终审阅,大幅缩短您的研究时间。
无需绑定信用卡
本解读由 AI 生成,并经人工编辑审核。