Detecting Spam Email with Machine Learning Optimized with Bio-Inspired Metaheuristic Algorithms

Metadane

• Autorzy: S. Gibson, B. Issac, L. Zhang, S.M. Jacob
• Rok: 2020
• Źródło: IEEE Access, Volume 8, Pages 187914–187932
• DOI/Link: 10.1109/ACCESS.2020.3030751
• Status: to-read
• Pochodzenie: Wyekstrahowane z al-subaiey-web-ai-phishing-2024 ([27] - GA-SGD 99.21% accuracy, highest in Table 3)
• Tagi: to-read reference genetic-algorithm particle-swarm-optimization spam-detection metaheuristic hyperparameter-optimization

Notatki

Publikacja dodana automatycznie z bibliografii.

Kontekst cytowania w Al-Subaiey 2024:

• Referenced jako [27] w paper (Table 1, Table 3 comparisons)
• Method: Genetic Algorithm + SGD (GA-SGD) - bio-inspired optimization for hyperparameter tuning
• Dataset: Ling, Enron, PUA, SpamAssassin (separately tested) - 20,170 spam, 16,545 ham
• Results: Accuracy 99.21%, Precision 98.68%, Recall 99.54% ← HIGHEST recall in Table 3
• Comparison: Slightly higher accuracy than Al-Subaiey (99.21% vs 99.10%), but Al-Subaiey provides web deployment + XAI

Technical Innovation:

• Bio-inspired algorithms (GA, PSO) as “coaches” fine-tuning ML models
• Achieved impressive accuracy (even 100% in some cases per Al-Subaiey citation)
• Combines ML with optimization techniques

Significance:

• Demonstrates hyperparameter optimization importance
• State-of-art performance (99.21% accuracy)
• Al-Subaiey comparable with simpler approach (no metaheuristic optimization needed)
• Higher computational cost likely (GA/PSO iterations)

Future Direction (implied):

• Could Al-Subaiey’s SVM+TF-IDF benefit from GA optimization? Trade-off: performance vs complexity

Dodaj PDF aby wygenerować pełne podsumowanie używając /summarize-paper gibson-bio-inspired-spam-detection-2020