Delay-dependent stability for recurrent neural networks with time-varying delays

Area of Science:

• Control Theory
• Artificial Intelligence
• Dynamical Systems

Background:

• Static neural networks are crucial in various applications.
• Time-varying delays can significantly impact network stability.
• Existing stability conditions may be overly conservative.

Purpose of the Study:

• To develop novel stability criteria for static neural networks with time-varying delays.
• To propose conditions that are less conservative than existing methods.
• To ensure asymptotic stability of the neural network.

Main Methods:

• Derivation of delay-independent stability conditions.
• Derivation of delay-dependent stability conditions for fast time-varying delays.
• Formulation of conditions using linear matrix inequalities (LMIs).

Main Results:

• Proposed delay-independent conditions are less conservative than prior work.
• Proposed delay-dependent conditions offer further conservatism reduction.
• Stability conditions are verifiable using established LMI algorithms.

Conclusions:

• The novel stability conditions effectively guarantee asymptotic stability.
• The proposed methods offer improved performance over existing techniques.
• The LMIs provide a computationally efficient way to check stability.