Incremental and Adaptive Feature Exploration over Time Series Stream

Abstract

Over past years, various attempts have been made at analysing Time Series (TS) which has been raising great interest of Data Mining community due to its special data format and broad application scenarios. An important aspect in TS analysis is Time Series Classification (TSC), which has been applied in medical diagnosis, human activity recognition, industrial troubleshooting, etc. Typically, all TSC work trains a stable model from an off-line TS dataset, without considering potential Concept Drift in streaming context. Domains like healthcare look to enrich the database gradually with more medical cases, or in astronomy, with human's growing knowledge about the universe, the theoretical basis for labelling data will change. The techniques applied in a stable TS dataset are then not adaptable in such dynamic scenarios (i.e. streaming context). Classical data stream analysis are biased towards vector or row data, where each attribute is independent to train an adaptive learning model, but rarely considers Time Series as a stream instance. Processing such type of data requires combining techniques in both communities of Time Series (TS) and Data Streams. To this end, by adopting the concepts of Shapelet and Matrix Profile, we conduct the first attempt to extract the adaptive features from Time Series Stream based on the Test-then-Train strategy, which is applicable in both contexts: a) under stable concept, learning model will be updated incrementally; b) for data source with Concept Drift, previous concepts that do not represent the current stream behavior will be discarded from the model.

On Handling Catastrophic Forgetting for Incremental Learning of Human Physical Activity on the Edge

“…More advanced work has been proposed in the Time Series Classification domain, where researchers aim to build general ML models covering various application domains [34], including HAR tasks. For instance, authors in [33] extract interpretable Shapelet features from time series, and combine with a kNN classifier; The recent end-to-end models [34] have shown promising results on HAR tasks, that generally rely on automatic feature extraction and selection [1,8] with regard to specific learning tasks, e.g., classification or forecasting.…”

Multi‐output regression for imbalanced data stream

“…As described in Zuo et al (2019), we have to accommodate two concepts: data streams and time series. Hence, to better cope with the situation, we define the concept of Streaming Time Series (Definition 2.1) in adapting the definition of Time Series Stream provided in Zuo et al (2019). Definition [Streaming Time Series (STS)] is a continuous (unbounded) flow of input data where each instance is a vector of real values: , where , is the timestamp of the last value entered, increases over time and is the length of the vector that is greater than or equal to 1.…”

“…Recall that, in our use case scenario, we are dealing with data (values) that are continuously emitted by sensors. As described in Zuo et al (2019), we have to accommodate two concepts: data streams and time series. Hence, to better cope with the situation, we define the concept of Streaming Time Series (Definition 2.1) in adapting the definition of Time Series Stream provided in Zuo et al (2019).…”

Learning the micro-environment from rich trajectories in the context of mobile crowd sensing

“…Nevertheless, the kNN-DTW is considered as the baseline for MTS classification and is widely outpaced by the advanced approaches such as Shapelets [49,59,60] or the frequent patterns [32]. Essentially, the kNN-DTW captures the global feature based on the distance measure between the entire sequences, while the local features (e.g., the frequent patterns [32], the interval features [12], Shapelets [49], etc.)…”