Deep Canonical Time Warping for Simultaneous Alignment and Representation Learning of Sequences

Abstract

Abstract-Machine learning algorithms for the analysis of time-series often depend on the assumption that utilised data are temporally aligned. Any temporal discrepancies arising in the data is certain to lead to ill-generalisable models, which in turn fail to correctly capture properties of the task at hand. The temporal alignment of time-series is thus a crucial challenge manifesting in a multitude of applications. Nevertheless, the vast majority of algorithms oriented towards temporal alignment are either applied directly on the observation space or simply utilise linear projections -thus failing to capture complex, hierarchical non-linear representations that may prove beneficial, especially when dealing with multi-modal data (e.g., visual and acoustic information). To this end, we present Deep Canonical Time Warping (DCTW), a method that automatically learns non-linear representations of multiple time-series that are (i) maximally correlated in a shared subspace, and (ii) temporally aligned. Furthermore, we extend DCTW to a supervised setting, where during training, available labels can be utilised towards enhancing the alignment process. By means of experiments on four datasets, we show that the representations learnt significantly outperform state-of-the-art methods in temporal alignment, elegantly handling scenarios with heterogeneous feature sets, such as the temporal alignment of acoustic and visual information.

Human Motion Analysis with Deep Metric Learning

“…We compare our MMD-NCA loss against the methods from DTW [42], MDDTW [25], CTW [47] and GDTW [48], as well as four state-of-the-art deep metric learning approaches: DCTW [41], triplet [33], triplet+GOR [45], and the N -Pairs deep metric loss [14]. Primarily, these methods are evaluated through action recognition task in Sec.…”

“…Since one of the datasets [1] labeled the actions with their corresponding subjects, we also investigate the possibility of performing a person identification task wherein, instead of measuring the similarity of the pose, we intend to measure the similarity the actors themselves based on their movement. To have a fair comparison, we only used our attention based LSTM architecture for all methods and only changed the loss function except the DCTW [41]. Prosed loss function in DCTW [41] requires the two sequences, therefore we remove the attention layer and use only our LSTM model.…”

“…To have a fair comparison, we only used our attention based LSTM architecture for all methods and only changed the loss function except the DCTW [41]. Prosed loss function in DCTW [41] requires the two sequences, therefore we remove the attention layer and use only our LSTM model. Notably, all deep metric learning methods are evaluated and trained with the same data splits.…”

“…The FPR at different percentages on CMU mocap and Human3.6M are reported in Table 1. With a true positive rate of 70%, the learning approaches [33,41,45,35] Query Cartwheel motion -1st- Fig. 4.…”

Temporal Transformer Networks: Joint Learning of Invariant and Discriminative Time Warping

“…There are also several works on Canonical Time Warping (CTW) for multimodal data where the time series from different streams are projected to a common space (similar to canonical correlation analysis) before aligning them [43]. Deep learning versions of CTW have also been proposed recently [36,35,12].…”