Deep Ensemble Learning for Fake Digital Image Detection: A Convolutional Neural Network-Based Approach
Bibliographic record
Abstract
The advent of deep learning (DL) technologies has paved the way for a plethora of applications, the creation of hyperrealistic images via generative adversarial networks (GANs) being a compelling example. However, these synthetic images, nearly indistinguishable from genuine ones to the human eye, can be exploited for nefarious activities such as cybercrime, extortion, politically motivated campaigns, propaganda, among others. This paper proposes a deep ensemble learning approach to detect such counterfeit images, aiming to mitigate the issues arising from deepfake multimedia. The impetus for engaging ensemble models stems from their capacity to reduce the generalization error of predictions, provided the foundational models exhibit diversity and independence. Consequently, the prediction error diminishes when an ensemble approach is deployed. In this study, the CASIA v2 benchmark datasets, comprising 12,323 color images (5,123 originals and 7,200 counterfeits), were utilized. This investigation employed an ensemble of 13 pre-trained CNN models. The ensemble technique amalgamates these models to form a comprehensive perceptual model, wherein each model contributes to the final outcome. The classification predictions from each model are considered a 'vote', with the majority verdict serving as the final prediction. The proposed methodology was also juxtaposed with prevailing techniques. Assessment of our approach's efficacy revealed a 100% accuracy rate, 97.75% precision, 87.46% recall, and 99.9% AUC, underscoring the improvement offered by the proposed system.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".