Split learning (SL) has been proposed to train deep learning models in a decentralized manner. For decentralized healthcare applications with vertical data partitioning, SL can be beneficial as it allows institutes with complementary features or images for a shared set of patients to jointly develop more robust and generalizable models. In this work, we propose "Split-U-Net" and successfully apply SL for collaborative biomedical image segmentation. Nonetheless, SL requires the exchanging of intermediate activation maps and gradients to allow training models across different feature spaces, which might leak data and raise privacy concerns. Therefore, we also quantify the amount of data leakage in common SL scenarios for biomedical image segmentation and provide ways to counteract such leakage by applying appropriate defense strategies.
翻译:提议以分散方式进行分解学习(SL),以培训深层次学习模式。对于具有纵向数据分割的分散式保健应用,SL可能是有益的,因为它允许具有互补特征或图像的研究所为一组共同患者联合开发更稳健、更通用的模式。在这项工作中,我们建议“Split-U-Net”并成功应用SL,用于合作生物医学图像分割。然而,SL要求交换中间活化地图和梯度,以便在不同特征空间进行训练,从而有可能泄漏数据并引起隐私问题。因此,我们还量化生物医学图像分割共同的SL情景中数据泄漏的数量,并通过应用适当的防御战略来防止这种泄漏。