Spatially Encoding Temporal Correlations to Classify Temporal Data Using Convolutional Neural Networks
This work addresses the challenge of applying convolutional neural networks to temporal data classification, which is incremental as it adapts existing image-based methods to time series.
The authors tackled the problem of classifying temporal data by spatially encoding temporal patterns as images, enabling the use of computer vision techniques. Their approach achieved competitive classification results with state-of-the-art methods on 12 benchmark time series datasets and two real spatial-temporal trajectory datasets.
We propose an off-line approach to explicitly encode temporal patterns spatially as different types of images, namely, Gramian Angular Fields and Markov Transition Fields. This enables the use of techniques from computer vision for feature learning and classification. We used Tiled Convolutional Neural Networks to learn high-level features from individual GAF, MTF, and GAF-MTF images on 12 benchmark time series datasets and two real spatial-temporal trajectory datasets. The classification results of our approach are competitive with state-of-the-art approaches on both types of data. An analysis of the features and weights learned by the CNNs explains why the approach works.