One-pass-throw-away learning for cybersecurity in streaming non-stationary environments by dynamic stratum network

Area of Science:

• Computer Science
• Cybersecurity
• Machine Learning

Background:

• Cybersecurity issues are prevalent in business applications, especially in dynamic, non-stationary environments.
• Conventional methods struggle with adaptive solutions due to repeated training processes.
• Existing approaches may deteriorate performance in dynamic settings.

Purpose of the Study:

• To propose an efficient learning method for cybersecurity in dynamic environments.
• To address the limitations of conventional methods in handling streaming non-stationary data.
• To enhance classification accuracy, reduce computational time, and minimize space complexity.

Main Methods:

• Developed a one-pass-throw-away learning algorithm utilizing a dynamical network structure.
• Introduced recursive functions for faster computation and minimal space complexity.
• Implemented information gain-based feature selection to reduce training time.

Main Results:

• The proposed algorithm demonstrated superior performance compared to existing methods.
• Achieved higher classification accuracy in incremental-like and online ensemble learning.
• Showcased significant improvements in space complexity and computational time.

Conclusions:

• The novel one-pass-throw-away learning approach effectively addresses cybersecurity challenges in dynamic environments.
• The method offers a computationally efficient and space-conscious solution for streaming data.
• This research provides a robust alternative to conventional methods for real-time cybersecurity applications.