arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.02745cs.LGstat.ML
收藏

基于规范化流的半监督学习预测分布的学习

Learning the Prediction Distribution for Semi-Supervised Learning with Normalising Flows

University of Edinburgh(爱丁堡大学);Samsung AI Centre Cambridge(三星剑桥人工智能中心)

Ivana Balažević, Carl Allen, Timothy Hospedales

摘要:

随着数据量的持续增长,标记过程日益成为瓶颈,从而产生了对利用未标记数据中信息的方法的需求。在图像分类的半监督学习(SSL)中已经取得了令人印象深刻的结果,接近完全监督的性能,只有一小部分数据被标记。在这项工作中,我们提出了一个概率原则的一般方法SSL,认为分布在标签的预测,标签的不同复杂性,从“一个热”向量的二进制向量和图像。我们的方法正则化了一个底层的监督模型,使用一个标准化流来学习标记数据预测的后验分布,作为未标记数据预测的先验。我们证明了这种方法在一系列具有不同输出复杂度的计算机视觉任务上的普遍适用性:分类,属性预测和图像到图像的翻译。

英文摘要:

As data volumes continue to grow, the labelling process increasingly becomes a bottleneck, creating demand for methods that leverage information from unlabelled data. Impressive results have been achieved in semi-supervised learning (SSL) for image classification, nearing fully supervised performance, with only a fraction of the data labelled. In this work, we propose a probabilistically principled general approach to SSL that considers the distribution over label predictions, for labels of different complexity, from "one-hot" vectors to binary vectors and images. Our method regularises an underlying supervised model, using a normalising flow that learns the posterior distribution over predictions for labelled data, to serve as a prior over the predictions on unlabelled data. We demonstrate the general applicability of this approach on a range of computer vision tasks with varying output complexity: classification, attribute prediction and image-to-image translation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑