Do Neural Networks Trained with Topological Features Learn Different Internal Representations?
Colorado State University(科罗拉多州立大学);University of Texas, El Paso(德克萨斯大学埃尔帕索分校);University of Washington(华盛顿大学);Michigan State University(密歇根州立大学);Pacific Northwest National Laboratory(太平洋西北国家实验室)
Sarah McGuire, Shane Jackson, Tegan Emerson, Henry Kvinge
备注:
To appear at NeurIPS 2022 Workshop on Symmetry and Geometry in Neural Representations (NeurReps)
There is a growing body of work that leverages features extracted via topological data analysis to train machine learning models. While this field, sometimes known as topological machine learning (TML), has seen some notable successes, an understanding of how the process of learning from topological features differs from the process of learning from raw data is still limited. In this work, we begin to address one component of this larger issue by asking whether a model trained with topological features learns internal representations of data that are fundamentally different than those learned by a model trained with the original raw data. To quantify ``different'', we exploit two popular metrics that can be used to measure the similarity of the hidden representations of data within neural networks, neural stitching and centered kernel alignment. From these we draw a range of conclusions about how training with topological features does and does not change the representations that a model learns. Perhaps unsurprisingly, we find that structurally, the hidden representations of models trained and evaluated on topological features differ substantially compared to those trained and evaluated on the corresponding raw data. On the other hand, our experiments show that in some cases, these representations can be reconciled (at least to the degree required to solve the corresponding task) using a simple affine transformation. We conjecture that this means that neural networks trained on raw data may extract some limited topological features in the process of making predictions.
Prediction of the motion of chest internal points using a recurrent neural network trained with real-time recurrent learning for latency compensation in lung cancer radiotherapy
The University of Tokyo(东京大学);National Institute for Quantum and Radiological Science and Technology(国立量子与放射科学技术研究所)
在肺癌患者的放射治疗期间,需要最小化递送到肿瘤周围的健康组织的辐射,由于呼吸运动和线性加速器系统的延迟,这是困难的。在这项研究中,我们首先使用Lucas-Kanade金字塔光流算法进行变形图像配准的胸部计算机断层扫描图像的四个肺癌患者。然后,我们根据先前计算的变形场跟踪靠近肺肿瘤的三个内部点,并使用使用实时递归学习(RTRL)和梯度裁剪训练的递归神经网络(RNN)预测它们的位置。呼吸数据是相当规则的,以大约2.5Hz采样,并且包括脊柱方向上的人工漂移。追踪点的运动幅度范围为12.0 mm ~ 22.7 mm。最后,我们提出了一种简单的方法,用于恢复和预测三维肿瘤图像的跟踪点和初始肿瘤图像的基础上的线性对应模型和Nadaraya-Watson非线性回归。测试集上RNN预测对应的均方根误差,最大误差和抖动小于线性预测和最小均方(LMS)获得的相同性能指标。特别地,与RNN相关联的最大预测误差等于1.51mm,分别比与线性预测和LMS相关联的最大误差低16.1%和5.0%。RTRL的每个时间步长的平均预测时间等于119 ms,这小于400 ms标记位置采样时间。预测图像中的肿瘤位置在视觉上看起来是正确的,这通过原始图像和预测图像之间的高平均互相关(等于0.955)来确认。