Hybrid YSGOA and neural networks based software failure prediction in cloud systems

Area of Science:

• Cloud Computing
• Software Engineering
• Artificial Intelligence

Background:

• Ensuring software dependability in complex cloud environments is challenging.
• Identifying and rectifying software anomalies proactively is crucial for robust systems.
• Existing methods face obstacles in the dynamic nature of cloud infrastructures.

Purpose of the Study:

• To develop an innovative methodology for predicting software malfunctions in cloud computing.
• To augment the purity metric of software malfunction prediction.
• To enhance the robustness and dependability of cloud-based software systems.

Main Methods:

• Amalgamation of hybrid optimization algorithms with Neural Networks (NN).
• Utilizing Yellow Saddle Goat Fish Algorithm (YSGA) for identifying pivotal features related to software failures.
• Employing Grasshopper Optimization Algorithm (GOA) for refining feature selection.
• Processing selected features with Neural Networks (NN) for accurate classification.

Main Results:

• The hybrid optimization strategy significantly curtails complexity in feature selection.
• The methodology expedites the processing of software malfunction prediction.
• Evaluation using the Failure-Dataset-OpenStack database demonstrated improved performance.
• The approach enhances the accuracy of predicting software anomalies.

Conclusions:

• The proposed hybrid optimization and Neural Network approach offers an effective solution for software malfunction prediction in cloud computing.
• This methodology improves efficiency and reduces complexity in identifying software failures.
• The study highlights the potential of combining advanced optimization algorithms with NNs for robust cloud system management.