论文标题
与邻居一致性学习嘈杂标签
Learning with Neighbor Consistency for Noisy Labels
论文作者
论文摘要
深度学习的最新进展依赖于大型，标记的数据集来培训大容量模型。但是，以时间和成本效益的方式收集大型数据集通常会导致标签噪声。我们提出了一种从嘈杂的标签中学习的方法，该方法利用特征空间中的训练示例之间的相似性，鼓励每个示例的预测与其最近的邻居相似。与使用多个模型或不同阶段的训练算法相比，我们的方法采用了简单的附加正规化项的形式。它可以解释为经典的，跨导式标签传播算法的归纳版。我们在数据集上彻底评估我们的方法评估合成（CIFAR-10，CIFAR-100）和现实（Mini-Webvision，Webvision，Clotsing1m，Mini-Imagenet-Red）噪声，并实现所有这些噪声，并实现所有这些噪声的竞争或最先进的精确度。
Recent advances in deep learning have relied on large, labelled datasets to train high-capacity models. However, collecting large datasets in a time- and cost-efficient manner often results in label noise. We present a method for learning from noisy labels that leverages similarities between training examples in feature space, encouraging the prediction of each example to be similar to its nearest neighbours. Compared to training algorithms that use multiple models or distinct stages, our approach takes the form of a simple, additional regularization term. It can be interpreted as an inductive version of the classical, transductive label propagation algorithm. We thoroughly evaluate our method on datasets evaluating both synthetic (CIFAR-10, CIFAR-100) and realistic (mini-WebVision, WebVision, Clothing1M, mini-ImageNet-Red) noise, and achieve competitive or state-of-the-art accuracies across all of them.
