Instagram fake profile detection using an ensemble learning method.

Area of Science:

• Computer Science
• Social Media Security
• Machine Learning

Background:

• Counterfeit accounts on Instagram contribute to spam, harmful information, and deceptive content, eroding user trust and compromising online security.
• Existing methods for identifying fake profiles are insufficient, necessitating advanced solutions to maintain platform integrity.

Purpose of the Study:

• To develop and evaluate a highly accurate machine learning model for identifying counterfeit Instagram accounts.
• To enhance the effectiveness and trustworthiness of online identity verification systems.

Main Methods:

• Implementation of a hybrid system combining XGBoost, SMOTE for class balancing, and GridSearchCV for hyperparameter tuning.
• Utilizing scale_pos_weight optimization and adaptive discovery for trend analysis in counterfeit accounts.
• Leveraging Random Forest hyperparameter fine-tuning for improved detection accuracy.

Main Results:

• The proposed model achieved an F1 score of 98%, recall of 98%, precision of 98.3%, and accuracy of 98.24%.
• Demonstrated significant reduction in counterfeit accounts, enhancing platform security.
• Established a new standard for trust safeguarding in social media environments.

Conclusions:

• The developed hybrid machine learning system offers a state-of-the-art solution for detecting counterfeit accounts on Instagram.
• This research improves online identity verification systems and strengthens user trust.
• The findings lay the groundwork for future advancements in social media security and combating deceptive online practices.