Variational Embedding Multiscale Sample Entropy: A Tool for Complexity Analysis of Multichannel Systems

Entropy-based methods have received considerable attention in the quantification of structural complexity of real-world systems. Among numerous empirical entropy algorithms, conditional entropy-based methods such as sample entropy, which are associated with amplitude distance calculation, are quite intuitive to interpret but require excessive data lengths for meaningful evaluation at large scales. To address this issue, we propose the variational embedding multiscale sample entropy (veMSE) method and conclusively demonstrate its ability to operate robustly, even with several times shorter data than the existing conditional entropy-based methods. The analysis reveals that veMSE also exhibits other desirable properties, such as the robustness to the variation in embedding dimension and noise resilience. For rigor, unlike the existing multivariate methods, the proposed veMSE assigns a different embedding dimension to every data channel, which makes its operation independent of channel permutation. The veMSE is tested on both stimulated and real world signals, and its performance is evaluated against the existing multivariate multiscale sample entropy methods. The proposed veMSE is also shown to exhibit computational advantages over the existing amplitude distance-based entropy methods.

Reconstruction of Phase Space and Eigenvalue Decomposition from a Biological Time Series: A Malayalam Speech Signal Case Study

Our objective is to describe the speech production system from a non-linear physiological system perspective and reconstruct the attractor from the experimental speech data. Mutual information method is utilized to find out the time delay for embedding. The False Nearest Neighbour (FNN) method and Principal Component Analysis (PCA) method are used for optimizing the embedding dimension of time series. The time series obtained from the typical non-linear systems, Lorenz system and Rössler system, is used to standardize the methods and the Malayalam speech vowel time series of both genders of different age groups, sampled at three sampling frequencies (16[Formula: see text]kHz, 32[Formula: see text]kHz, 44.1[Formula: see text]kHz), are taken for analysis. It was observed that time delay varies from sample to sample and, it ought to be better to figure out the time delay with the embedding dimension analysis. The embedding dimension is shown to be independent of gender, age and sampling frequency and can be projected as five. Hence a five-dimensional hyperspace will probably be adequate for reconstructing attractor of speech time series.

A Study On Methods for Determining Phase Space Reconstruction Parameters

Several pairs of algorithms were used to determine the phase space reconstruction parameters to analyze the dynamic characteristics of chaotic time series. The reconstructed phase trajectories were compared with the original phase trajectories of the Lorenz attractor, Rössler attractor, and Chens attractor to obtain the optimum method for determining the phase space reconstruction parameters with high precision and efficiency. The research results show that the false nearest neighbor method and the complex auto-correlation method provided the best results. The saturated embedding dimension method based on the saturated correlation dimension method is proposed to calculate the time delay. Different time delays are obtained by changing the embedding dimension parameters of the complex auto-correlation method. The optimum time delay occurs at the point where the time delay is stable. The validity of the method is verified through combing the application of correlation dimension, showing that the proposed method is suitable for the effective determination of the phase space reconstruction parameters.

Graph Entropy Guided Node Embedding Dimension Selection for Graph Neural Networks

Graph representation learning has achieved great success in many areas, including e-commerce, chemistry, biology, etc. However, the fundamental problem of choosing the appropriate dimension of node embedding for a given graph still remains unsolved. The commonly used strategies for Node Embedding Dimension Selection (NEDS) based on grid search or empirical knowledge suffer from heavy computation and poor model performance. In this paper, we revisit NEDS from the perspective of minimum entropy principle. Subsequently, we propose a novel Minimum Graph Entropy (MinGE) algorithm for NEDS with graph data. To be specific, MinGE considers both feature entropy and structure entropy on graphs, which are carefully designed according to the characteristics of the rich information in them. The feature entropy, which assumes the embeddings of adjacent nodes to be more similar, connects node features and link topology on graphs. The structure entropy takes the normalized degree as basic unit to further measure the higher-order structure of graphs. Based on them, we design MinGE to directly calculate the ideal node embedding dimension for any graph. Finally, comprehensive experiments with popular Graph Neural Networks (GNNs) on benchmark datasets demonstrate the effectiveness and generalizability of our proposed MinGE.

Cotangent spaces and separating re-embeddings

Given an affine algebra [Formula: see text], where [Formula: see text] is a polynomial ring over a field [Formula: see text] and [Formula: see text] is an ideal in [Formula: see text], we study re-embeddings of the affine scheme [Formula: see text], i.e. presentations [Formula: see text] such that [Formula: see text] is a polynomial ring in fewer indeterminates. To find such re-embeddings, we use polynomials [Formula: see text] in the ideal [Formula: see text] which are coherently separating in the sense that they are of the form [Formula: see text] with an indeterminate [Formula: see text] which divides neither a term in the support of [Formula: see text] nor in the support of [Formula: see text] for [Formula: see text]. The possible numbers of such sets of polynomials are shown to be governed by the Gröbner fan of [Formula: see text]. The dimension of the cotangent space of [Formula: see text] at a [Formula: see text]-linear maximal ideal is a lower bound for the embedding dimension, and if we find coherently separating polynomials corresponding to this bound, we know that we have determined the embedding dimension of [Formula: see text] and found an optimal re-embedding.

Principled approach to the selection of the embedding dimension of networks

Network embedding is a general-purpose machine learning technique that encodes network structure in vector spaces with tunable dimension. Choosing an appropriate embedding dimension – small enough to be efficient and large enough to be effective – is challenging but necessary to generate embeddings applicable to a multitude of tasks. Existing strategies for the selection of the embedding dimension rely on performance maximization in downstream tasks. Here, we propose a principled method such that all structural information of a network is parsimoniously encoded. The method is validated on various embedding algorithms and a large corpus of real-world networks. The embedding dimension selected by our method in real-world networks suggest that efficient encoding in low-dimensional spaces is usually possible.

HRV analysis: undependability of approximate entropy at locating optimum complexity in malnourished children

Introduction: Approximate Entropy is an extensively enforced metric to evaluate chaotic responses and irregularities of RR intervals sourced from an eletrocardiogram. However, to estimate their responses, it has one major problem – the accurate determination of tolerances and embedding dimensions. So, we aimed to overt this potential hazard by calculating numerous alternatives to detect their optimality in malnourished children. Materials and methods: We evaluated 70 subjects split equally: malnourished children and controls. To estimate autonomic modulation, the heart rate was measured lacking any physical, sensory or pharmacologic stimuli. In the time series attained, Approximate Entropy was computed for tolerance (0.1→0.5 in intervals of 0.1) and embedding dimension (1→5 in intervals of 1) and the statistical significances between the groups by their Cohen’s ds and Hedges’s gs were totalled. Results: The uppermost value of statistical significance accomplished for the effect sizes for any of the combinations was −0.2897 (Cohen’s ds) and −0.2865 (Hedges’s gs). This was achieved with embedding dimension = 5 and tolerance = 0.3. Conclusions: Approximate Entropy was able to identify a reduction in chaotic response via malnourished children. The best values of embedding dimension and tolerance of the Approximate Entropy to identify malnourished children were, respectively, embedding dimension = 5 and embedding tolerance = 0.3. Nevertheless, Approximate Entropy is still an unreliable mathematical marker to regulate this.

Cyclotomic numerical semigroup polynomials with at most two irreducible factors

A numerical semigroup S is cyclotomic if its semigroup polynomial $$\mathrm {P}_S$$ P S is a product of cyclotomic polynomials. The number of irreducible factors of $$\mathrm {P}_S$$ P S (with multiplicity) is the polynomial length $$\ell (S)$$ ℓ ( S ) of S. We show that a cyclotomic numerical semigroup is complete intersection if $$\ell (S)\le 2$$ ℓ ( S ) ≤ 2 . This establishes a particular case of a conjecture of Ciolan et al. (SIAM J Discrete Math 30(2):650–668, 2016) claiming that every cyclotomic numerical semigroup is complete intersection. In addition, we investigate the relation between $$\ell (S)$$ ℓ ( S ) and the embedding dimension of S.

A multiple-parameter approach for short-term traffic flow prediction

This study proposes a short-term traffic flow prediction approach based on multiple traffic flow basic parameters, in which the chaos theory and support vector regression are utilized. First, a high-dimensional variable space can be obtained according to the traffic flow fundamental function. Then, a maximum conditional entropy method is proposed to determine the embedding dimension. And multiple time series are reconstructed based on the phase space reconstruction theory using the time delay obtained by mutual information method and the embedding dimension captured by the maximum conditional entropy method. Finally, the reconstructed phase space is used as the input and the support vector regression optimized by the genetic algorithm is utilized to predict the traffic flow. Numerical experiments are performed and the results show that the approach proposed has strong fitting capability and better prediction accuracy.