Leveraging Deep Learning Techniques for the Stability Principles of Current Artificial Neural Networks Are Emerging Into Their Activation Functions

@article{1705256,
      author = {Dr. CH Narasimha Chary, Mocharla Ramesh Babu, More Sadanandam, S Krishna Reddy},
      title = {Leveraging Deep Learning Techniques for the Stability Principles of Current Artificial Neural Networks Are Emerging Into Their Activation Functions},
      journal = {Iconic Research And Engineering Journals},
      year = {2023},
      volume = {7},
      number = {5},
      pages = {244-247},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1705256.pdf},
      abstract = {Continuous-time recurrent neural network stability issues have been thoroughly researched. This paper aims to present a thorough analysis of the literature on the stability of continuous-time recurrent neural networks, encompassing models such as Cohen-Grasberg[1] and Hopfield neural networks. The stability results of recurrent neural networks with various classes of time delays are thoroughly examined, as time delays are an inherent part of real-world applications. The findings of dealing with the constant/variable delay in recurrent neural networks for the case of delay-dependent stability are compiled. Different forms, including algebraic inequality forms, -matrix forms, and linear forms, are produced by the relationship between stability. It is addressed and compared with Lyapunov diagonal stability forms and matrix inequality[2] forms. Additionally covered are certain adequate and essential stability requirements for recurrent neural networks in the absence of time delays. Finally, some thoughts are shared on the stability analysis of recurrent neural networks going forward.},
      keywords = {Balance points, continuous neural networks, monotonic conduct and fixed point hypothesis.},
      month = {November},
  }