A Joint-Entropy Approach To Time-series Classification

Safarihamid, Kimia; Pourafzal, Alireza; Fereidunian, Alireza

doi:10.1109/icspis54653.2021.9729371

Cited by 3 publications

(5 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We realize that the number of trainable parameters in deep learning models (number of filters, kernels, and depth of the network) are chosen based on a trial-and-error process. However, as suggested by Pourafzal and Fereidunian [17] and Safarihamid et al [18], regardless of observation length of chaotic time series, they can be classified using a few features of the complex system [17,18]. This gives us the intuition that there is a feature space in which the given chaotic time series can be sparsified.…”

Section: Motivationmentioning

confidence: 97%

“…We extract the predictability with Hurst exponent [17] with 20 lags, self-organization with Disequilibrium [17] with 𝐾 = 10 and emergence using distribution entropy with 10 frequency bins [18]. In addition, complexity is the product of emergence and selforganization.…”

Section: Complex Systems Approachmentioning

confidence: 99%

“…Moreover, approaches by Pourafzal and Fereidunian [17] and Safarihamid et al [18] took advantages of the link between complex systems and chaotic systems. According to Poincaré, the unpredictable behaviour of non-linear dynamical systems can be interpreted as an extreme point of complexity instead of disorder [19].…”

Section: Introductionmentioning

confidence: 99%

“…Additionally, Safarihamid et al [18] reviewed the definitions of characteristics and they substituted with measured-theoretic entropies (based on Kolmogorov complexity). Specifically, it is shown that emergence is indeed the violation of regularities in a time-series, whereas self-organization can be defined as the expectation of limited outcomes in a system.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Chaotic Time Series Recognition: A Deep Learning Model Inspired by Complex Systems Characteristics

Pourafzal

Fereidunian

Safarihamid

2023

IJE

View full text Add to dashboard Cite

A deep learning method is developed for chaotic time series classification. We investigate the chaotic state of a dynamical system, based on the output of the system. One of the main obstacles in time series classification is mapping a high-dimensional vector into a scalar value. To reduce the dimensions, it is common to use an average pooling layer block after feature extraction block. This blind process results in models with high computational complexity and potent to overfitting. One alternative is to extract the features manually, then apply shallow learning models to classify the time series. In fact, since complexity lies between the chaos and order, it is a sound idea to refer to complex systems characteristics to explore the chaotic region entrance. Therefore, chaotic state of a dynamical system can be recognized solely based on these characteristics. Inspired by this concept, we conclude that there is a feature space in which the output vector can be sparsified. Thus, we propose a deep learning method which the feature space dimensions successively are reduced in the feature extraction process. Specifically, we employ a fully convolutional network and add on two maximum pooling layers to the relevant feature extraction block. To validate the proposed model, the Lorenz system is employed which exhibits chaotic/nonchaotic states. We generate a labeled dataset containing 10000 samples each with 20000 features of the output of Lorenz system. The proposed model achieves 99.45 percent accuracy over 2000 unseen samples, higher than all the other competitor methods.

show abstract

Section: Motivationmentioning

confidence: 97%

Section: Complex Systems Approachmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Chaotic Time Series Recognition: A Deep Learning Model Inspired by Complex Systems Characteristics

Pourafzal

Fereidunian

Safarihamid

2023

IJE

View full text Add to dashboard Cite

show abstract

“…With a discussion on the pros and cons of sample entropy and permutation entropy, Rostaghi [ 3 ] explored the application of discrete entropy in a time series. Safari [ 4 ] introduced different entropy indicators in association with emergence and self-organization for time series classification. Based on these associations, the authors proposed an alternative joint entropy in which each feature was represented using a specific entropy indicator.…”

Section: Introductionmentioning

confidence: 99%

Dynamic System Stability Modeling Approach with Sparrow-Inspired Meta-Heuristic Optimization Algorithm

Xia,

Zhang,

Wang

2023

Biomimetics

View full text Add to dashboard Cite

Aiming at the accurate prediction of the inception of instability in a compressor, a dynamic system stability model is proposed based on a sparrow-inspired meta-heuristic optimization algorithm in this article. To achieve this goal, a spatial mode is employed for flow field feature extraction and modeling object acquisition. The nonlinear characteristic presented in the system is addressed using fuzzy entropy as the identification strategy to provide a basis for instability determination. Using Sparrow Search Algorithm (SSA) optimization, a Radial Basis Function Neural Network (RBFNN) is achieved for the performance prediction of system status. A Logistic SSA solution is first established to seek the optimal parameters of the RBFNN to enhance prediction accuracy and stability. On the basis of the RBFNN-LSSA hybrid model, the stall inception is detected about 35.8 revolutions in advance using fuzzy entropy identification. To further improve the multi-step network model, a Tent SSA is introduced to promote the accuracy and robustness of the model. A wider range of potential solutions within the TSSA are explored by incorporating the Tent mapping function. The TSSA-based optimization method proves a suitable adaptation for complex nonlinear dynamic modeling. And this method demonstrates superior performance, achieving 42 revolutions of advance warning with multi-step prediction. This RBFNN-TSSA model represents a novel and promising approach to the application of system modeling. These findings contribute to enhancing the abnormal warning capability of dynamic systems in compressors.

show abstract

A hybrid method combining Lévy process and neural network for predicting remaining useful life of rotating machinery

Lv,

Liu,

et al. 2024

Advanced Engineering Informatics

View full text Add to dashboard Cite

A Joint-Entropy Approach To Time-series Classification

Cited by 3 publications

References 29 publications

Chaotic Time Series Recognition: A Deep Learning Model Inspired by Complex Systems Characteristics

Chaotic Time Series Recognition: A Deep Learning Model Inspired by Complex Systems Characteristics

Dynamic System Stability Modeling Approach with Sparrow-Inspired Meta-Heuristic Optimization Algorithm

A hybrid method combining Lévy process and neural network for predicting remaining useful life of rotating machinery

Contact Info

Product

Resources

About