WALKING WALKing walking: Action Recognition from Action Echoes

Recognizing human actions represented by 3D trajectories of skeleton joints is a challenging machine learning task. In this paper, the 3D skeleton sequences are regarded as multivariate time series, and their dynamics and multiscale features are efficiently learned from action echo states. Specifically, first the skeleton data from the limbs and trunk are projected into five high dimensional nonlinear spaces, that are randomly generated by five dynamic, training-free recurrent networks, i.e., the reservoirs of echo state networks (ESNs). In this way, the history of the time series is represented as nonlinear echo states of actions. We then use a single multiscale convolutional layer to extract multiscale features from the echo states, and maintain multiscale temporal invariance by a max-over-time pooling layer. We propose two multi-step fusion strategies to integrate the spatial information over the five parts of the human physical structure. Finally, we learn the label distribution using softmax. With one training-free recurrent layer and only layer of convolution, our Convolutional Echo State Network (ConvESN) is a very efficient end-to-end model, and achieves state-of-the-art performance on four skeleton benchmark data sets.

Download Full-text

Wavelet-denoising multiple echo state networks for multivariate time series prediction

Information Sciences ◽

10.1016/j.ins.2018.07.015 ◽

2018 ◽

Vol 465 ◽

pp. 439-458 ◽

Cited By ~ 10

Author(s):

Meiling Xu ◽

Min Han ◽

Hongfei Lin

Keyword(s):

Time Series ◽

Time Series Prediction ◽

Multivariate Time Series ◽

Wavelet Denoising ◽

Echo State Networks

Download Full-text

Multivariate Time Series Data Prediction Based on ATT-LSTM Network

Applied Sciences ◽

10.3390/app11209373 ◽

2021 ◽

Vol 11 (20) ◽

pp. 9373

Author(s):

Jie Ju ◽

Fang-Ai Liu

Keyword(s):

Time Series ◽

Deep Learning ◽

Time Series Data ◽

Mutual Influence ◽

Multivariate Time Series ◽

The Other ◽

Series Data ◽

Data Sets ◽

Learning Models ◽

Prediction Problems

Deep learning models have been widely used in prediction problems in various scenarios and have shown excellent prediction effects. As a deep learning model, the long short-term memory neural network (LSTM) is potent in predicting time series data. However, with the advancement of technology, data collection has become more accessible, and multivariate time series data have emerged. Multivariate time series data are often characterized by a large amount of data, tight timeline, and many related sequences. Especially in real data sets, the change rules of many sequences will be affected by the changes of other sequences. The interacting factors data, mutation information, and other issues seriously impact the prediction accuracy of deep learning models when predicting this type of data. On the other hand, we can also extract the mutual influence information between different sequences and simultaneously use the extracted information as part of the model input to make the prediction results more accurate. Therefore, we propose an ATT-LSTM model. The network applies the attention mechanism (attention) to the LSTM to filter the mutual influence information in the data when predicting the multivariate time series data, which makes up for the poor ability of the network to process data. Weaknesses have greatly improved the accuracy of the network in predicting multivariate time series data. To evaluate the model’s accuracy, we compare the ATT-LSTM model with the other six models on two real multivariate time series data sets based on two evaluation indicators: Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). The experimental results show that the model has an excellent performance improvement compared with the other six models, proving the model’s effectiveness in predicting multivariate time series data.

Download Full-text

Kaleidomaps: A New Technique for the Visualization of Multivariate Time-Series Data

Information Visualization ◽

10.1057/palgrave.ivs.9500154 ◽

2007 ◽

Vol 6 (2) ◽

pp. 155-167 ◽

Cited By ~ 9

Author(s):

Kim Bale ◽

Paul Chapman ◽

Nick Barraclough ◽

Jon Purdy ◽

Nizamettin Aydin ◽

...

Keyword(s):

Time Series ◽

Time Series Data ◽

Multivariate Time Series ◽

Rapid Identification ◽

Series Data ◽

Data Sets ◽

Periodic Patterns ◽

Large Complex ◽

Visualization Tools ◽

Glass Patterns

In this paper, we describe a new visualization technique that can facilitate our understanding and interpretation of large complex multivariate time-series data sets. ‘Kaleidomaps’ have been carefully developed taking into account research into how we perceive form and structure within Glass patterns. We have enhanced the classic cascade plot using the curvature of a line to alter the detection of possible periodic patterns within multivariate dual periodicity data sets. Similar to Glass patterns, the concentric nature of the Kaleidomap may induce a motion signal within the brain of the observer facilitating the perception of patterns within the data. Kaleidomaps and our associated visualization tools alter the rapid identification of periodic patterns not only within their own variants but also across many different sets of variants. By linking this technique with traditional line graphs and signal processing techniques, we are able to provide the user with a set of visualization tools that permit the combination of multivariate time-series data sets in their raw form and also with the results of mathematical analysis. In this paper, we provide two case study examples of how Kaleidomaps can be used to improve our understanding of large complex multivariate time dependent data.

Download Full-text

Outlier Detection for Time Series with Recurrent Autoencoder Ensembles

Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2019/378 ◽

2019 ◽

Cited By ~ 21

Author(s):

Tung Kieu ◽

Bin Yang ◽

Chenjuan Guo ◽

Christian S. Jensen

Keyword(s):

Time Series ◽

Outlier Detection ◽

Time Series Data ◽

State Of The Art ◽

Multivariate Time Series ◽

Series Data ◽

Data Sets ◽

Network Connection ◽

Detection Quality ◽

Insight Into

We propose two solutions to outlier detection in time series based on recurrent autoencoder ensembles. The solutions exploit autoencoders built using sparsely-connected recurrent neural networks (S-RNNs). Such networks make it possible to generate multiple autoencoders with different neural network connection structures. The two solutions are ensemble frameworks, specifically an independent framework and a shared framework, both of which combine multiple S-RNN based autoencoders to enable outlier detection. This ensemble-based approach aims to reduce the effects of some autoencoders being overfitted to outliers, this way improving overall detection quality. Experiments with two large real-world time series data sets, including univariate and multivariate time series, offer insight into the design properties of the proposed frameworks and demonstrate that the resulting solutions are capable of outperforming both baselines and the state-of-the-art methods.

Download Full-text

Statistical approaches for reconstructing neuro-cognitive dynamics from high-dimensional neural recordings

e-Neuroforum ◽

10.1007/s13295-010-0011-0 ◽

2010 ◽

Vol 16 (4) ◽

Cited By ~ 2

Author(s):

D. Durstewitz ◽

E. Balaguer-Ballester

Keyword(s):

Time Series ◽

Neural Coding ◽

Imaging Techniques ◽

Multivariate Time Series ◽

High Dimensional ◽

Data Sets ◽

Single Trial ◽

Neural Recordings ◽

Neural Ensemble ◽

Spatio Temporal

AbstractRecent advances in multiple single-unit recording and optical imaging techniques now routinely enable observation of the activity from tens to hundreds of neurons simultaneously. The result is high-dimensional multivariate time series which offer an unprecedented range of possibilities for gaining insight into the detailed spatio-temporal neural dynamics underlying cognition. For instance, they may pave the way for reliable single-trial analyses, for investigating the role of higher-order correlations in neural coding, the mechanisms of neural ensemble formation, or more generally of transitions among attractor states accompanying cognitive processes. At the same time, exploiting the information in these multivariate time series may require more sophisticated statistical methods beyond the commonly employed repertoire. Here we review, using specific experimental examples, some of these methods for visualizing structure in high-dimensional data sets, for statistical inference about the apparent structure, for single-trial analysis of neural time series, and for reconstructing some of the dynamical properties of neural systems that can only be inferred from simultaneous recordings.

Download Full-text