Red-billed blue magpie optimization for training feedforward neural networks.

Area of Science:

• Artificial Intelligence
• Machine Learning
• Optimization Algorithms

Background:

• Feedforward Neural Networks (FNNs) are crucial for nonlinear fitting, classification, and regression tasks.
• Existing training methods for FNNs can be improved in terms of efficiency and accuracy.
• Nature-inspired optimization algorithms offer potential for enhancing machine learning model training.

Purpose of the Study:

• To introduce the Red-Billed Blue Magpie Optimization (RBMO) algorithm for training FNNs.
• To evaluate the RBMO's effectiveness in measuring output discrepancies and quantifying training efficiency.
• To optimize connection weights and biases in FNNs using the RBMO algorithm.

Main Methods:

• The Red-Billed Blue Magpie Optimization (RBMO) algorithm is developed, mimicking bird foraging and hunting behaviors.
• RBMO is applied to train FNNs, adjusting weights and biases to minimize prediction errors.
• Seventeen diverse datasets are used for empirical validation and comparison with other optimization techniques.

Main Results:

• RBMO demonstrates superior performance in enhancing training efficiency and classification precision compared to 12 other algorithms.
• The algorithm effectively balances global exploration and local exploitation, leading to robust and stable solutions.
• RBMO shows accelerated convergence speed, heightened calculation accuracy, and diminished computational complexity.

Conclusions:

• The RBMO algorithm is a highly effective and robust method for training Feedforward Neural Networks.
• RBMO offers significant advantages in convergence speed, accuracy, and computational efficiency.
• This optimization technique shows great promise for improving machine learning applications requiring precise classification and regression.