Synthesization of high-capacity auto-associative memories using complex-valued neural networks

Abstract

In this paper, a novel design procedure is proposed for synthesizing high-capacity auto-associative memories based on complex-valued neural networks with real-imaginary-type activation functions and constant delays. Stability criteria dependent on external inputs of neural networks are derived. The designed networks can retrieve the stored patterns by external inputs rather than initial conditions. The derivation can memorize the desired patterns with lower-dimensional neural networks than real-valued neural networks, and eliminate spurious equilibria of complex-valued neural networks. One numerical example is provided to show the effectiveness and superiority of the presented results.

Bearing Fault Vibration Signal Feature Extraction and Recognition Method Based on EEMD Superresolution Sparse Decomposition

“…is article uses multiparameter, distributed online detection and corresponding signal analysis and processing methods for the large load and limit speed conditions of the rotating machinery system, so as to avoid adverse effects that come from sensor noise, signal disturbance, or instrument performance degradation while obtaining effective information [19]. Feature extraction methods based on numerical optimization calculations and artificial intelligence, such as self-associative memory neural networks, provide a good theoretical support for the fault diagnosis and decisionmaking of rotating machinery [20]; Eklund et al [22] used the Rank Permutation Transformation (RPT method). Reduce the influence of noise in the state measurement, and make the statistical state events more clear through the classification method [20].…”

“…Feature extraction methods based on numerical optimization calculations and artificial intelligence, such as self-associative memory neural networks, provide a good theoretical support for the fault diagnosis and decisionmaking of rotating machinery [20]; Eklund et al [22] used the Rank Permutation Transformation (RPT method). Reduce the influence of noise in the state measurement, and make the statistical state events more clear through the classification method [20]. e exponential smoothing and average filtering methods are widely used in signal processing because of their good noise reduction effects.…”

An Improved Complex-Valued Recurrent Neural Network Model for Time-Varying Complex-Valued Sylvester Equation

“…Now the complex-valued neural networks have shown more advantages than the real-valued neural networks in some fields, such as the high-capacity auto-associative memories [36], the spectral domain [37], the millimeter-wave active imaging [38], and the geometric measures [39]. Inspired by the previous studies for the ZNNs, we explore a novel complex-valued ZNN model for solving the complexvalued time-varying Sylvester equation (CVTVSE) in this paper.…”