Requirements
Can anyone recommend papers of the below time-series classification?
Background
I have tried implementing a decision-making program.
The input variables include:
(1) time-series data measured by wearable sensors (numerical data)
(2) human data such as gender, birth, blood type, etc (categorical data).
The (1) was collected every 10ms and (2) was obtained every session.
(A session consists of 30 minutes)
The output variables are some gestures (multiple classes).
This problem has been defined as time-series classification.
I've read some time-series classification articles, but most classifiers deal with numerical time-series data only. (e.g. sensor data input, multiple classes output)
Have you thought of perhaps defining this as perhaps panel data. Then looking at some references in the z-library.. this might help with the different lengths of time that you are trying to put together. Best wishes, David Booth
For the (1) numerical time series data, I recommend our paper entitled TSFEL: Time Series Feature Extraction Library, which was written primarily inspired by features for human activity recognition.
Paper: https://www.sciencedirect.com/science/article/pii/S2352711020300017
Open Source code: https://github.com/fraunhoferportugal/tsfel
Our lab has some work published on the topic of combining (1) and (2). I recommend this article: https://ieeexplore.ieee.org/abstract/document/8894463).
Dear David Eugene Booth , Andreas Theissler , Duarte Folgado ,
Thank you very much for your sincere answer.
The articles and methods will be of great help to the ongoing research.
I hope you're always happy.
Best regards, Minsu Chang
Hi Minsu,
I can recommend a paper by my colleague Stylianos Paraschiakos: "Activity recognition using wearable sensors for tracking the elderly" (Article Activity recognition using wearable sensors for tracking the elderly
). Also check out his other relevant work: https://scholar.google.com.ph/citations?user=wkfTmRwAAAAJ&hl=enGood luck!
- Badges
- Science topic
- Similar topics
- Papers