Method and system for time series representation learning via dynamic time warping
Techniques that facilitate time series analysis using machine learning are provided. In one example, a system includes a matrix generation component, a matrix factorization component and a machine learning component. The matrix generation component converts at least a first stream of time series data and a second stream of time series data (e.g., raw time series data) into a data matrix (e.g., a partially-observed similarity matrix) that comprises void data and numerical data associated with the first stream of time series data and the second stream of time series data. The matrix factorization component factorizes the data matrix into a first factorization data matrix and a second factorization data matrix. The machine learning component processes a machine learning model based on first matrix data associated with the first factorization data matrix and second matrix data associated with the second factorization data matrix.
1. A system, comprising:
a memory; and
a processor, operably coupled to the memory, and wherein the processor:
receives a first stream of time series data comprising first data elements describing a first event over time and a second stream of time series data comprising second data elements describing a second event over time, wherein the first stream is based on partial observation of the first event over time with the first stream comprising a void data element where no observation of the first event was made,
generates an n by n partially-observed similarity matrix by randomly sampling O(n log(n)) pairs of time series data between the first stream of time series data and the second stream of time series data based on a dynamic time warping process that pairs elements from different times based on similarity, wherein n is the total number of time series pairs between the first stream of time series data and the second stream of time series data,
based on a symmetric matrix factorization process, factors the partially-observed similarity matrix into a first factorization data matrix and a second factorization data matrix,
trains a machine learning model based on first matrix data associated with the first factorization data matrix and second matrix data associated with the second factorization data matrix wherein the first matrix data comprises void data based on the void data element, and
based on the machine learning model, generates a prediction for a future occurrence of the first event by employing a machine learning process.
2. The system of claim 1 , wherein the first stream of time series data comprises a first time span and the second stream of time series data comprises a second time span, wherein the second time span is different than the first time span, and wherein the dynamic time warping process pairs the elements based on aligning elements of the first time span and the second time span.
3. The system of claim 2 , wherein the aligning of the elements of the first time span and the second time span comprises modifying time data associated with the elements of the first time span and the second time span.
4. The system of claim 1 , wherein the machine learning model comprises a neural network.
5. The system of claim 1 , wherein the processor trains the machine learning model based on the void data of the first matrix data by cyclically analyzing the void data based on a cyclic coordinate descent process.
6. The system of claim 1 , wherein the processor maintains temporal data of the first stream of time series data and the second stream of time series data.
7. The system of claim 1 , wherein the processor trains the machine learning model based on the first matrix data and the second matrix data to facilitate a reduced amount of time to generate machine learning output via the machine learning model.
8. The system of claim 1 , wherein the processor further generates, based on the machine learning model, machine learning output to facilitate classification or grouping of data associated with at least one of an image processing system, a video processing system, a signal processing system, and a data analytics system.
9. A computer program product for machine learning, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
receive a first stream of time series data comprising first data elements describing a first event over time and a second stream of time series data comprising second data elements describing a second event over time, wherein the first stream is based on partial observation of the first event over time with the first stream comprising a void data element where no observation of the first event was made,
generates an n by n partially-observed similarity matrix by randomly sampling O(n log(n)) pairs of time series data between the first stream of time series data and the second stream of time series data based on a dynamic time warping process that pairs elements from different times based on similarity, wherein n is the total number of time series pairs between the first stream of time series data and the second stream of time series data,
based on a symmetric matrix factorization process, factor the partially-observed similarity matrix into a first factorization data matrix and a second factorization data matrix,
train a machine learning model based on first matrix data associated with the first factorization data matrix and second matrix data associated with the second factorization data matrix wherein the first matrix data comprises void data based on the void data element, and
based on the machine learning model, generate a prediction for a future occurrence of the first event by employing a machine learning process.
10. The computer program product of claim 9 , wherein the dynamic time warping process pairs the elements based on aligning elements of the first time span and the second time span.
11. The computer program product of claim 10 , wherein the dynamic time warping process is performed repeatedly for a defined amount of time.
12. The computer program product of claim 9 , wherein the first stream of time series data comprises a first time span and the second stream of time series data comprises a second time span, wherein the second time span is different than the first time span.
13. The computer program product of claim 9 , wherein the program instructions are further executable by the processor to cause the processor to:
maintain temporal data of the first stream of time series data and the second stream of time series data.
14. The computer program product of claim 9 , wherein the program instructions are further executable by the processor to cause the processor to:
train the machine learning model based on the first factorization matrix data and the second factorization matrix data to facilitate a reduced amount of time to generate machine learning output via the machine learning model.
15. The computer program product of claim 9 , wherein the program instructions are further executable by the processor to cause the processor to:
generate, based on the machine learning model, output to facilitate classification or grouping of data associated with at least one of an image processing system, a video processing system, a signal processing system, and a data analytics system.