Isotonic CCA for sequence alignment and activity recognition

Abstract

This paper presents an approach for sequence alignment based on canonical correlation analysis(CCA). We show that a novel set of constraints imposed on traditional CCA leads to canonical solutions with the time warping property, i.e., non-decreasing monotonicity in time. This formulation generalizes the more traditional dynamic time warping (DTW) solutions to cases where the alignment is accomplished on arbitrary subsequence segments, optimally determined from data, instead on individual sequence samples. We then introduce a robust and efficient algorithm to find such alignments using non-negative least squares reductions. Experimental results show that this new method, when applied to MOCAP activity recognition problems, can yield improved recognition accuracy.

Robust time-series retrieval using probabilistic adaptive segmental alignment

“…Maximum segment length is set to 10. We compare our method against DTW, canonical time warping (CTW) Zhou and de la Torre (2009)) and IsoCCA Shariat and Pavlovic (2011)). SPHMM achieved the highest accuracy, 85.5(±6.18).…”

“…We also consider the dataset proposed in Shariat and Pavlovic (2011)), where the authors developed an alternative approach to segmental alignment. The dataset consists of sinusoidal and rectangular signals that are embedded into Gaussian noise such that the placement of the signal is also random.…”

“…In Shariat and Pavlovic (2011)) the authors proposed a segmental alignment method based on CCA, i.e., IsoCCA. Despite promising results, the proposed framework does not provide a proper metric between the segments.…”

“…One consequence of DTW's essential reliance on comparison of pairs of individual time-series samples is its, as well as many of its derivatives', sensitivity to noise Shariat and Pavlovic (2011)); Ye and Keogh (2009)); Vlachos et al (2002)); Zakaria et al (2015)). We have observed that in the presence of significant noise, edit-distance based methods outperforms DTW.…”

“…In Shariat and Pavlovic (2011)) the authors proposed an approach, based on canonical correlation analysis (CCA), to handle this segmental alignment. The objective function (IsoCCA) is constrained properly to impose time monotonicity over segments.…”