Cattle Corral Optimization Algorithm (CCO)

A Novel, Meta Heuristic, Nature Inspired, Feature Selection Algorithm

Overview

Machine learning optimization often involves complex challenges like large search spaces and hyperparameter sensitivity, especially in domains like fake news detection. Nature-inspired metaheuristics (e.g., PSO, WOA) have emerged as powerful tools to address these issues.

This repository presents the Cattle Corral Optimization (CCO) algorithm, a novel metaheuristic building upon this field. CCO is inspired by the ecosystem around grazing cattle, translating animal behaviors into optimization strategies.

How CCO Works

CCO uses a two-pronged search mechanism:

1. Egret-Inspired Global Search: Mimics the strategic movement of egrets towards resource-rich areas (high-potential solutions).
2. Fly-Inspired Local Search: Simulates the random, localized exploration of flies around cattle (refining known good solutions).

These behaviors are integrated within a population-based model designed for machine learning pipelines.

Core Functionality & Features

• Hybrid Optimization: Simultaneously optimizes model hyperparameters and selects informative feature subsets.
• Informed Initialization: Uses pre-ranked features (weighted correlation & mutual information) to guide the initial search, improving convergence speed and solution quality.
• Efficiency: Employs an early stopping mechanism based on fitness stagnation to avoid unnecessary computation.

Repository Purpose

This repository provides the source code for the CCO algorithm, enabling its use and further development.

Key Contributions:

• Novel CCO Algorithm blending global (egret) and local (fly) search.
• Hybrid feature selection and hyperparameter tuning framework.
• Empirical validation demonstrating effectiveness.

📊 Feature Selection Results

Metric	None	CCO	HSDO	WOA	HHA	GSF	PSO
AUC	0.637	0.979	0.971	0.952	0.950	0.958	0.971
F1	0.946	0.935	0.933	0.892	0.891	0.895	0.923
Accuracy	0.963	0.933	0.933	0.893	0.892	0.904	0.929
Precision	0.085	0.999	0.999	0.998	0.997	0.999	0.999
FPR	0.001	0.001	0.001	0.002	0.002	0.002	0.001
FNR	0.998	0.130	0.133	0.213	0.213	0.181	0.143
MCC	0.008	0.879	0.874	0.804	0.803	0.822	0.867
n-feat	21	9	9	15	12	10	10

Legend:

• CCO: Cattle Corral Optimization
• HSDO: Hybrid Squirrel Dragonfly Optimization
• WOA: Whale Optimization Algorithm
• HHA: Horse Heard Optimization
• GSF: Grey Sail Fish Optimization
• PSO: Particle Swarm Optimization

🔗 Run on Google Colab