Sequence Pattern Extraction by Segmenting Time Series Data Using GP-HSMM with Hierarchical Dirichlet Process

Nagano, Masatoshi; Nagaoka, Tomoaki; Nagai, Takayuki; Mochihashi, Daichi; Kobayashi, Ichiro; Kaneko, Masahide

doi:10.1109/iros.2018.8594029

Cited by 11 publications

(15 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, in HDP-GP-HSMM ( Figure 2C ), the trajectory can be represented continuously using two Gaussian processes (GPs). We confirmed that our GP-based model can estimate segments more accurately than HMM-based methods (Nagano et al, 2018 ).…”

Section: Introductionsupporting

confidence: 84%

“…The process by which the probability distribution is constructed through a two-phase Dirichlet process is called a hierarchical Dirichlet process (HDP) (Teh et al, 2006). HDP is described in detail in Nagano et al (2018). The class c j of the j-th segment is determined by the class of the (j − 1)-th segment and transition probability π c .…”

Section: Hierarchical Dirichlet Process-variational Autoencoder-gaussmentioning

confidence: 99%

“…used identical to the latent variables of VAE. In the case of PCA and ICA, the min-max normalization, in which the values are normalized to a range from -1 to 1, was applied for the sequence of latent variables, as with Nagano et al (2018). Additionally, we described the results of HVGH and VAE+HDP-GP-HSMM for comparison.…”

Section: A B Cmentioning

confidence: 99%

Section: Hierarchical Dirichlet Process–variational Autoencoder–gamentioning

confidence: 99%

“…Additionally, we have proposed a method to extract the segments by capturing such a nature stochastically. One of our previous methods is the hierarchical Dirichlet process–Gaussian process–hidden semi-Markov model (HDP-GP-HSMM) (Nagano et al, 2018 ). HDP-GP-HSMM is a non-parametric Bayesian model that is a hidden semi-Markov model, the emission distributions of which are Gaussian processes (MacKay, 1998 ), and it facilitates the segmentation of time-series data in an unsupervised manner.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

HVGH: Unsupervised Segmentation for High-Dimensional Time Series Using Deep Neural Compression and Statistical Generative Model

et al. 2019

Self Cite

View full text Add to dashboard Cite

Humans perceive continuous high-dimensional information by dividing it into meaningful segments, such as words and units of motion. We believe that such unsupervised segmentation is also important for robots to learn topics such as language and motion. To this end, we previously proposed a hierarchical Dirichlet process-Gaussian process-hidden semi-Markov model (HDP-GP-HSMM). However, an important drawback of this model is that it cannot divide high-dimensional time-series data. Furthermore, low-dimensional features must be extracted in advance. Segmentation largely depends on the design of features, and it is difficult to design effective features, especially in the case of high-dimensional data. To overcome this problem, this study proposes a hierarchical Dirichlet process-variational autoencoder-Gaussian process-hidden semi-Markov model (HVGH). The parameters of the proposed HVGH are estimated through a mutual learning loop of the variational autoencoder and our previously proposed HDP-GP-HSMM. Hence, HVGH can extract features from high-dimensional time-series data while simultaneously dividing it into segments in an unsupervised manner. In an experiment, we used various motion-capture data to demonstrate that our proposed model estimates the correct number of classes and more accurate segments than baseline methods. Moreover, we show that the proposed method can learn latent space suitable for segmentation.

show abstract

Section: Introductionsupporting

confidence: 84%

Section: Hierarchical Dirichlet Process-variational Autoencoder-gaussmentioning

confidence: 99%

Section: A B Cmentioning

confidence: 99%

Section: Hierarchical Dirichlet Process–variational Autoencoder–gamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations