InceptionTime: Finding AlexNet for time series classification
- PDF / 1,720,298 Bytes
- 27 Pages / 439.37 x 666.142 pts Page_size
- 87 Downloads / 226 Views
InceptionTime: Finding AlexNet for time series classification Hassan Ismail Fawaz, et al. [full author details at the end of the article] Received: 11 September 2019 / Accepted: 27 July 2020 © The Author(s), under exclusive licence to Springer Science+Business Media LLC, part of Springer Nature 2020
Abstract This paper brings deep learning at the forefront of research into time series classification (TSC). TSC is the area of machine learning tasked with the categorization (or labelling) of time series. The last few decades of work in this area have led to significant progress in the accuracy of classifiers, with the state of the art now represented by the HIVE-COTE algorithm. While extremely accurate, HIVE-COTE cannot be applied to many real-world datasets because of its high training time complexity in O(N 2 · T 4 ) for a dataset with N time series of length T . For example, it takes HIVE-COTE more than 8 days to learn from a small dataset with N = 1500 time series of short length T = 46. Meanwhile deep learning has received enormous attention because of its high accuracy and scalability. Recent approaches to deep learning for TSC have been scalable, but less accurate than HIVE-COTE. We introduce InceptionTime—an ensemble of deep Convolutional Neural Network models, inspired by the Inception-v4 architecture. Our experiments show that InceptionTime is on par with HIVE-COTE in terms of accuracy while being much more scalable: not only can it learn from 1500 time series in one hour but it can also learn from 8M time series in 13 h, a quantity of data that is fully out of reach of HIVE-COTE. Keywords Time series classification · Deep learning · Scalable model · Inception
1 Introduction Recent times have seen an explosion in the magnitude and prevalence of time series data. Industries varying from health care (Forestier et al. 2018; Lee et al. 2018; Ismail Fawaz et al. 2019d) and social security (Yi et al. 2018) to human activity recognition (Yuan et al. 2018) and remote sensing (Pelletier et al. 2019), all now produce time series datasets of previously unseen scale—both in terms of time series
Responsible editor: Eamonn Keogh.
B
Hassan Ismail Fawaz [email protected]
Extended author information available on the last page of the article
123
H. I. Fawaz et al.
length and quantity. This growth also means an increased dependence on automatic classification of time series data, and ideally, algorithms with the ability to do this at scale. These problems, known as Time Series Classification (TSC), differ significantly to traditional supervised learning for structured data, in that the algorithms should be able to handle and harness the temporal information present in the signal (Bagnall et al. 2017). It is easy to draw parallels from this scenario to computer vision problems such as image classification and object localization, where successful algorithms learn from the spatial information contained in an image. Put simply, the time series problem is essentially the same class of problem, just with one les
Data Loading...
