标题： 基于子空间特征表示的医学影像推理少样本学习 显示英文标题

标题： Few-shot Learning for Inference in Medical Imaging with Subspace Feature Representations

摘要： 与视觉场景识别领域不同，由于可以使用非常大的数据集来训练深度神经网络，该领域取得了巨大的进展，而从医学图像进行推断通常会受到只有少量数据可用的限制。 当处理规模约为几百个数据项的非常小的数据集问题时，可以通过使用在自然图像上预训练的模型作为特征提取器，并在此特征空间中执行经典模式识别技术，从而仍可利用深度学习的强大功能，这被称为少样本学习问题。 在特征空间的维度与数据项的数量相当甚至更大的情况下，降维是必要的，通常通过主成分分析，即奇异值分解（SVD）来实现。 在本文中，注意到在这种情况下使用SVD不合适，我们引入并探索了基于判别分析和非负矩阵分解（NMF）的两种替代方法。 使用涵盖$11$种不同疾病类型的14个不同数据集，我们证明了低维的判别子空间在性能上显著优于基于SVD的子空间和原始特征空间。 我们还表明，在适度维度下，NMF在这一设置中是SVD的有竞争力的替代方法。 显示英文摘要

摘要： Unlike the field of visual scene recognition where tremendous advances have taken place due to the availability of very large datasets to train deep neural networks, inference from medical images is often hampered by the fact that only small amounts of data may be available. When working with very small dataset problems, of the order of a few hundred items of data, the power of deep learning may still be exploited by using a model pre-trained on natural images as a feature extractor and carrying out classic pattern recognition techniques in this feature space, the so-called few-shot learning problem. In regimes where the dimension of this feature space is comparable to or even larger than the number of items of data, dimensionality reduction is a necessity and is often achieved by principal component analysis, i.e., singular value decomposition (SVD). In this paper, noting the inappropriateness of using SVD for this setting, we usher in and explore two alternatives based on discriminant analysis and non-negative matrix factorization (NMF). Using 14 different datasets spanning $11$ distinct disease types, we demonstrate that discriminant subspaces at low dimensions achieve significant improvements over SVD-based subspaces and the original feature space. We also show that NMF at modest dimensions is a competitive alternative to SVD in this setting.