让每个样本都达标：一种新训练范式提升模型在弱样本上的表现。
Feasible Learning
- 以每个样本的损失上限为约束，动态调整权重训练模型。
- 在图像分类等任务中，尾部性能显著提升，平均性能几乎不变。
- 适合关注少数类或异常样本表现的场景，如医疗诊断、长尾分布。
我们提出可行学习（Feasible Learning, FL），一种以样本为中心的学习范式，模型通过求解一个约束每个训练样本损失上限的可行性问题进行训练。与普遍采用的经验风险最小化（ERM）框架——优化平均性能不同，FL 要求在每一个数据点上都达到可接受的性能。由于只要满足预设性能阈值的模型即为有效解，优化算法及其动态过程对最终解的特性具有决定性影响。我们研究了一种原始-对偶方法，该方法在训练过程中动态重加权每个样本的重要性。为解决实际中设定合理阈值的挑战，我们引入了包含最小范数松弛变量的FL松弛形式。在图像分类、年龄回归及大语言模型偏好优化等多个任务上的实证分析表明，使用FL训练的模型在学习数据的同时，相较于ERM展现出更优的尾部行为，且对平均性能仅造成微小影响。
原文摘要 · Abstract (English)
We introduce Feasible Learning (FL), a sample-centric learning paradigm where models are trained by solving a feasibility problem that bounds the loss for each training sample. In contrast to the ubiquitous Empirical Risk Minimization (ERM) framework, which optimizes for average performance, FL demands satisfactory performance on every individual data point. Since any model that meets the prescribed performance threshold is a valid FL solution, the choice of optimization algorithm and its dynamics play a crucial role in shaping the properties of the resulting solutions. In particular, we study a primal-dual approach which dynamically re-weights the importance of each sample during training. To address the challenge of setting a meaningful threshold in practice, we introduce a relaxation of FL that incorporates slack variables of minimal norm. Our empirical analysis, spanning image classification, age regression, and preference optimization in large language models, demonstrates that models trained via FL can learn from data while displaying improved tail behavior compared to ERM, with only a marginal impact on average performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。