A Secured Artificial Intelligence (AI) Assisted Personal Data Prediction and Leakage Prevention System Using Deep Learning Logic
Bibliographic record
Abstract
Securing personal data against prediction misuse and leakage threats has emerged as a pressing concern in the era of artificial intelligence. The paper suggests an AI-Assisted Secured Personal Data Prediction and Leakage Prevention System, which combines a hybrid CapsuleNetxGBoost-based system with federated privacy/differentiated privacy models. Data preprocessing includes sanitization, anonymization and synthetic data generation to make sure privacy is preserved. The CapsuleNet extracts hierarchical relationships in sensitive attributes, whereas XGBoost narrows down predictive decision-making. In order to protect against attacks further, adversarial training and immutable logging based on blockchain is added, and homomorphic encryption is also used to process queries securely. The experimental results indicate that the given method is notably superior to traditional deep learning models. In particular, it had a prediction accuracy of 98.6 (a higher score than CNN, 93.5), a precision of 98.3, a recall of 98.8, and F1-score of 98.5. Strongness against adversarial examples and leakage probability were higher than 94 and 2.1 respectively under a rigorous privacy constraint <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\varepsilon=1.2)$</tex>. The Data Leakage Risk Index (DLRI), which was assisted by AI, also allowed identifying insider threats and abnormal access patterns in a dynamic way. Finally, the proposed model is not only more predictive accurate, but also resists leakage and adversarial exploitation. It has had wide applications in sensitive areas like health care, financial services and e-governance. Future studies will be aimed at extending the architecture by adding quantum-resistant-based encryption, reinforcement-based adaptive access control, as well as extending the DLRI architecture to multi-cloud and IoT-based settings. This paper takes the secure AI frameworks a step further to predictive intelligence and privacy protection.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.002 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.000 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".