Constrained Adaptive Distillation Based on Topological Persistence for Wearable Sensor Data
Constrained Adaptive Distillation Based on Topological Persistence for Wearable Sensor Data
复制标题
DOI：
10.1109/tim.2023.3329818
复制
发表时间：
2023
影响因子：
5.6
通讯作者：
Eunyeong Jeon;Hongjun Choi;Ankita Shukla;Yuan Wang;M. Buman;Pavan Turaga
中科院分区：
文献类型：
--
作者：
Eunyeong Jeon;Hongjun Choi;Ankita Shukla;Yuan Wang;M. Buman;Pavan Turaga
可穿戴传感器数据分析与拓扑数据分析（TDA）产生的持久性特征，已取得了巨大的成功，在各种应用中，然而，它遭受了大量的计算和时间资源提取拓扑特征。在本文中，我们的方法利用了知识蒸馏（KD），其中涉及使用TDA生成的原始时间序列和持久性图像（PI）训练的多个教师网络。然而，从教师模型利用不同的特征作为输入到学生模型的知识的直接转移导致知识差距和有限的性能。为了解决这个问题，我们引入了一个强大的框架，集成了两个不同的教师的多模态功能，使学生能够有效地学习所需的知识。为了考虑多模态的统计差异，利用基于熵的约束自适应加权机制来自动平衡教师的影响，并鼓励学生模型充分采用来自两个教师的知识。为了同化由不同风格的蒸馏模型产生的不同结构信息，使用小批次内的批次和通道相似性。我们证明了该方法对可穿戴传感器数据的有效性。
Wearable sensor data analysis with persistence features generated by topological data analysis (TDA) has achieved great success in various applications, and however, it suffers from large computational and time resources for extracting topological features. In this article, our approach utilizes knowledge distillation (KD) that involves the use of multiple teacher networks trained with the raw time series and persistence images (PIs) generated by TDA. However, direct transfer of knowledge from the teacher models utilizing different characteristics as inputs to the student model results in a knowledge gap and limited performance. To address this problem, we introduce a robust framework that integrates multimodal features from two different teachers and enables a student to learn desirable knowledge effectively. To account for statistical differences in multimodalities, an entropy-based constrained adaptive weighting mechanism is leveraged to automatically balance the effects of teachers and encourage the student model to adequately adopt the knowledge from two teachers. To assimilate dissimilar structural information generated by different style models for distillation, batch and channel similarities within a mini-batch are used. We demonstrate the effectiveness of the proposed method on wearable sensor data.