Melola Discovery

← Melola Discovery

Jurnal Tanpa peringkat SINTA

Robustness of Bayesian-Optimized Random Forest Against Feature-Space Evasion Attacks for Spam Detection

Recursive Journal of Informatics · Vol. 4 No. 2 (2026) · hlm. 137–148

DOI
10.15294/rji.v4i2.62732
ISSN
2986-6588, 2963-5551
Terbit
2026-09-30
Penerbit
Universitas Negeri Semarang
Bahasa
eng

Abstrak

Abstract. Email spam filters may perform well on unmodified messages yet remain vulnerable to deliberate feature manipulation. Although hyperparameter optimization can improve clean-data performance, such gains do not establish robustness under test-time perturbations. A paired, leakage-controlled evaluation is therefore needed to assess both properties under identical conditions. Purpose: This study compared the clean classification performance and feature-space robustness of a library-default Random Forest (RF-default) and a Bayesian-optimized Random Forest (BO-RF) on Spambase. Methods/Study design/approach: Exact duplicates were removed, and predictor-identical records were grouped to prevent cross-fold leakage. Evaluation used two repetitions of five-fold nested stratified group-aware cross-validation with shared outer folds. Bayesian Optimization maximized inner-validation spam-class F1. Frozen outer-test folds were assessed on clean inputs, class-center-directed perturbations at α = 0.05, 0.10, 0.20, and 0.40, and 30 magnitude-matched random directions. Retained inference assessed within-model clean-versus-directed and directed-versus-random contrasts in F1, Recall, and model-specific attack success rate (ASR). Result/Findings: BO-RF yielded slightly higher means across all four clean metrics; clean F1 was 93.65 ± 0.88% versus 93.53 ± 0.98% for RF-default. Bayesian Optimization selected the effective default configuration in nine of ten outer folds. At directed α = 0.40, F1 fell to 85.20% for RF-default and 85.31% for BO-RF, while Recall fell to 77.94% and 78.03%, respectively; directed ASR was approximately 15.5% for both models. All 48 retained within-model contrasts met Holm-adjusted thresholds, indicating degradation relative to clean inputs and a stronger effect of directed than magnitude-matched random perturbations. No direct between-model inferential family was retained, so robustness comparisons remain descriptive. Novelty/Originality/Value: Under the evaluated Spambase feature-space setting, clean-F1 Bayesian Optimization yielded only marginal descriptive gains and did not demonstrate a clear robustness advantage over the effective RF-default configuration. The findings show that clean-metric optimization should not be treated as a proxy for evasion robustness.

Kata kunci

Jaringan kolaborasi para penulis

40 penulis · 96 tautan kepenulisan bersama · 8 lintas organisasi · menampilkan 40 yang paling banyak terhubung

Arahkan kursor atau ketuk sebuah nama untuk menyorot rekan penulisnya. Klik dua kali untuk membuka halaman penulis.

  • State University of Semarang
  • International Islamic University Malaysia
  • Universitas PGRI Palembang
  • Afiliasi tidak diketahui
  • Organisasi berbeda
  • Organisasi sama
  • Afiliasi tidak diketahui
  • Garis lebih tebal = lebih banyak artikel bersama · lingkaran lebih besar = lebih banyak artikel
Tampilkan sebagai tabel
Rekan penulisOrganisasiArtikel bersama
Endang SugihartiState University of Semarang6
Florentina Yuni AriniState University of Semarang3
Rochmad RochmadInternational Islamic University Malaysia3
Amin SuyitnoState University of Semarang3
Alamsyah AlamsyahState University of Semarang2
Novanka Agnes SekartajiState University of Semarang2
Irfan Fajar MuttaqinState University of Semarang2
Ramadhan Ridho ArrohmanState University of Semarang2
Dwijanto DwijantoState University of Semarang2
Putriaji HendikawatiState University of Semarang2
Budi PrasetiyoState University of Semarang1
Much Aziz MuslimState University of Semarang1
Jumanto JumantoState University of Semarang1
Rofik RofikState University of Semarang1
Arif WidiyatmokoState University of Semarang1
Stephani Diah PamelasariState University of Semarang1
Arka YanitamaState University of Semarang1
Melissa Salma DarmawanState University of Semarang1
Aji SaputraUniversitas PGRI Palembang1
Desy Fitria AstutianingtyasAfiliasi tidak diketahui1
Syafa Nabilah SyahputraState University of Semarang1
Anggyi Trisnawan PutraState University of Semarang1
Athoillah AthoillahState University of Semarang1
Kholiq BudimanAfiliasi tidak diketahui1
Anggyi Trisnawan PutraAfiliasi tidak diketahui1
zaenal abidinAfiliasi tidak diketahui1
Anggoro Sakti AnindhitoState University of Semarang1
Fajar SujitoAfiliasi tidak diketahui1
Anita Wahyu MelindaAfiliasi tidak diketahui1
Isa AkhlisState University of Semarang1
Achmad Wahyu HidayatAfiliasi tidak diketahui1
Dian Rahman SyahputraAfiliasi tidak diketahui1
Safit FirmansyahAfiliasi tidak diketahui1
Anggil AgustaAfiliasi tidak diketahui1
Fachrizal Ahdy AndoyoState University of Semarang1
Margareta Ayu RiantikState University of Semarang1
Dandi Indra WijayaAfiliasi tidak diketahui1
Hanif Nur CahyaniAfiliasi tidak diketahui1

Afiliasi berasal dari OpenAlex; metadata OAI-PMH jurnal tidak memuatnya. Penulis tanpa ID OpenAlex dicocokkan berdasarkan nama, sehingga dua orang bisa tergabung atau satu orang bisa terpisah.

Sumber metadata: OAI-PMH (metadata jurnal) https://journal.unnes.ac.id/journals/rji/oai, pertama di-harvest 6 Nov 2026, terakhir terlihat 8 Nov 2026.
Metadata asli (oai_dc)
<oai_dc:dc xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/  http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
	<dc:title xml:lang="en">Robustness of Bayesian-Optimized Random Forest Against Feature-Space Evasion Attacks for Spam Detection</dc:title>
	<dc:creator xml:lang="en">Aminudin, Moh.</dc:creator>
	<dc:creator xml:lang="en">Arifudin, Riza</dc:creator>
	<dc:subject xml:lang="en">Email spam detection</dc:subject>
	<dc:subject xml:lang="en">Random Forest</dc:subject>
	<dc:subject xml:lang="en">adversarial robustnes</dc:subject>
	<dc:subject xml:lang="en">feature-space evasion attacks</dc:subject>
	<dc:description xml:lang="en">Abstract. Email spam filters may perform well on unmodified messages yet remain vulnerable to deliberate feature manipulation. Although hyperparameter optimization can improve clean-data performance, such gains do not establish robustness under test-time perturbations. A paired, leakage-controlled evaluation is therefore needed to assess both properties under identical conditions.
Purpose: This study compared the clean classification performance and feature-space robustness of a library-default Random Forest (RF-default) and a Bayesian-optimized Random Forest (BO-RF) on Spambase.
Methods/Study design/approach: Exact duplicates were removed, and predictor-identical records were grouped to prevent cross-fold leakage. Evaluation used two repetitions of five-fold nested stratified group-aware cross-validation with shared outer folds. Bayesian Optimization maximized inner-validation spam-class F1. Frozen outer-test folds were assessed on clean inputs, class-center-directed perturbations at α = 0.05, 0.10, 0.20, and 0.40, and 30 magnitude-matched random directions. Retained inference assessed within-model clean-versus-directed and directed-versus-random contrasts in F1, Recall, and model-specific attack success rate (ASR).
Result/Findings: BO-RF yielded slightly higher means across all four clean metrics; clean F1 was 93.65 ± 0.88% versus 93.53 ± 0.98% for RF-default. Bayesian Optimization selected the effective default configuration in nine of ten outer folds. At directed α = 0.40, F1 fell to 85.20% for RF-default and 85.31% for BO-RF, while Recall fell to 77.94% and 78.03%, respectively; directed ASR was approximately 15.5% for both models. All 48 retained within-model contrasts met Holm-adjusted thresholds, indicating degradation relative to clean inputs and a stronger effect of directed than magnitude-matched random perturbations. No direct between-model inferential family was retained, so robustness comparisons remain descriptive.
Novelty/Originality/Value: Under the evaluated Spambase feature-space setting, clean-F1 Bayesian Optimization yielded only marginal descriptive gains and did not demonstrate a clear robustness advantage over the effective RF-default configuration. The findings show that clean-metric optimization should not be treated as a proxy for evasion robustness.</dc:description>
	<dc:publisher xml:lang="en">Universitas Negeri Semarang</dc:publisher>
	<dc:date>2026-09-30</dc:date>
	<dc:type>info:eu-repo/semantics/article</dc:type>
	<dc:type>info:eu-repo/semantics/publishedVersion</dc:type>
	<dc:type xml:lang="en">Peer-reviewed Article</dc:type>
	<dc:format>application/pdf</dc:format>
	<dc:identifier>https://journal.unnes.ac.id/journals/rji/article/view/62732</dc:identifier>
	<dc:identifier>10.15294/rji.v4i2.62732</dc:identifier>
	<dc:source xml:lang="en">Recursive Journal of Informatics; Vol. 4 No. 2 (2026): September 2026; 137-148</dc:source>
	<dc:source>2986-6588</dc:source>
	<dc:source>2963-5551</dc:source>
	<dc:source>10.15294/j2r8b957</dc:source>
	<dc:language>eng</dc:language>
	<dc:relation>https://journal.unnes.ac.id/journals/rji/article/view/62732/11625</dc:relation>
	<dc:rights xml:lang="en">Copyright (c) 2026 Recursive Journal of Informatics</dc:rights>
</oai_dc:dc>