Evaluating Machine Unlearning: Applications, Approaches, and Accuracy
ABSTRACT Machine learning (ML) enables computers to learn from experience by identifying patterns and trends. Despite ML's advancements in extracting valuable data, there are instances necessitating the removal or deletion of certain data, as ML models can inadvertently memorize training data....
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-01-01
|
| Series: | Engineering Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/eng2.13081 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832576620350668800 |
|---|---|
| author | Zulfiqar Ali Asif Muhammad Rubina Adnan Tamim Alkhalifah Sheraz Aslam |
| author_facet | Zulfiqar Ali Asif Muhammad Rubina Adnan Tamim Alkhalifah Sheraz Aslam |
| author_sort | Zulfiqar Ali |
| collection | DOAJ |
| description | ABSTRACT Machine learning (ML) enables computers to learn from experience by identifying patterns and trends. Despite ML's advancements in extracting valuable data, there are instances necessitating the removal or deletion of certain data, as ML models can inadvertently memorize training data. In many cases, ML models may memorize sensitive or personal data, raising concerns about data privacy and security. Machine unlearning (MU) techniques offer a solution to these concerns by selectively removing sensitive data from trained models without significantly compromising their performance. Similarly, we can analyze and evaluate whether MU can successfully achieve the “right to be forgotten.” In this paper, we investigate various MU approaches regarding their accuracy and potential applications. Experiments have shown that the data‐driven approach emerged as the most efficient method in terms of both time and accuracy, achieving a high level of precision with a minimal number of training epochs. When fine‐tuning, the full test error rises somewhat to 14.57% from the baseline model's 14.28%. One approach shows a high forget error of 99.90% with a full test error of 20.68%, while retraining yields a 100% forget error and a test error of 21.37%. While error‐minimizing noise preserves performance, the SCRUB technique results in a 21.08% test error and an 81.05% forget error, in contrast to the degradation brought on by error‐maximizing noise. On the other hand, the agnostic approach displayed sluggishness and generated less accurate results compared to the data‐driven approach. Furthermore, the choice of approach may depend on the unique requirements of the task and the available training resources. |
| format | Article |
| id | doaj-art-69d7ab512656467da70f780ba9613281 |
| institution | Kabale University |
| issn | 2577-8196 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Wiley |
| record_format | Article |
| series | Engineering Reports |
| spelling | doaj-art-69d7ab512656467da70f780ba96132812025-01-31T00:22:49ZengWileyEngineering Reports2577-81962025-01-0171n/an/a10.1002/eng2.13081Evaluating Machine Unlearning: Applications, Approaches, and AccuracyZulfiqar Ali0Asif Muhammad1Rubina Adnan2Tamim Alkhalifah3Sheraz Aslam4Department of Computer Science COMSATS University Islamabad Islamabad PakistanFAST School of Computing National University of Computer & Emerging Sciences Islamabad PakistanDepartment of Computer Science COMSATS University Islamabad Islamabad PakistanDepartment of Computer Engineering, College of Computer Qassim University Buraydah Saudi ArabiaDepartment of Computer Science CTL eurocollege Limassol 3077 CyprusABSTRACT Machine learning (ML) enables computers to learn from experience by identifying patterns and trends. Despite ML's advancements in extracting valuable data, there are instances necessitating the removal or deletion of certain data, as ML models can inadvertently memorize training data. In many cases, ML models may memorize sensitive or personal data, raising concerns about data privacy and security. Machine unlearning (MU) techniques offer a solution to these concerns by selectively removing sensitive data from trained models without significantly compromising their performance. Similarly, we can analyze and evaluate whether MU can successfully achieve the “right to be forgotten.” In this paper, we investigate various MU approaches regarding their accuracy and potential applications. Experiments have shown that the data‐driven approach emerged as the most efficient method in terms of both time and accuracy, achieving a high level of precision with a minimal number of training epochs. When fine‐tuning, the full test error rises somewhat to 14.57% from the baseline model's 14.28%. One approach shows a high forget error of 99.90% with a full test error of 20.68%, while retraining yields a 100% forget error and a test error of 21.37%. While error‐minimizing noise preserves performance, the SCRUB technique results in a 21.08% test error and an 81.05% forget error, in contrast to the degradation brought on by error‐maximizing noise. On the other hand, the agnostic approach displayed sluggishness and generated less accurate results compared to the data‐driven approach. Furthermore, the choice of approach may depend on the unique requirements of the task and the available training resources.https://doi.org/10.1002/eng2.13081agnostic approachdata‐driven approachGeneral Data Protection Regulationmachine learningmachine unlearning |
| spellingShingle | Zulfiqar Ali Asif Muhammad Rubina Adnan Tamim Alkhalifah Sheraz Aslam Evaluating Machine Unlearning: Applications, Approaches, and Accuracy Engineering Reports agnostic approach data‐driven approach General Data Protection Regulation machine learning machine unlearning |
| title | Evaluating Machine Unlearning: Applications, Approaches, and Accuracy |
| title_full | Evaluating Machine Unlearning: Applications, Approaches, and Accuracy |
| title_fullStr | Evaluating Machine Unlearning: Applications, Approaches, and Accuracy |
| title_full_unstemmed | Evaluating Machine Unlearning: Applications, Approaches, and Accuracy |
| title_short | Evaluating Machine Unlearning: Applications, Approaches, and Accuracy |
| title_sort | evaluating machine unlearning applications approaches and accuracy |
| topic | agnostic approach data‐driven approach General Data Protection Regulation machine learning machine unlearning |
| url | https://doi.org/10.1002/eng2.13081 |
| work_keys_str_mv | AT zulfiqarali evaluatingmachineunlearningapplicationsapproachesandaccuracy AT asifmuhammad evaluatingmachineunlearningapplicationsapproachesandaccuracy AT rubinaadnan evaluatingmachineunlearningapplicationsapproachesandaccuracy AT tamimalkhalifah evaluatingmachineunlearningapplicationsapproachesandaccuracy AT sherazaslam evaluatingmachineunlearningapplicationsapproachesandaccuracy |