Selective Unlearning in Face Recognition: Forgetting Faces without Compromising Accuracy

Facial recognition technology is increasingly being used in current applications ranging from smartphone unlocking to security and surveillance systems. This extensive adoption, however, has generated severe concerns about individual privacy and data policies. In response to these concerns, this study introduces a novel methodology that gives people more control over their data in facial recognition systems. This study’s main contribution is the proposal of a strategy for selectively unlearning certain faces from trained facial recognition algorithms. An iterative model adaption approach is used to achieve this selective unlearning process. It becomes feasible to empower individuals to control the presence of their facial data in these systems by iteratively fine-tuning the model. This not only improves individual privacy but also conforms with ethical technology development standards. Furthermore, the methodology presented in this study is an important step towards improving privacy in the age of ubiquitous facial recognition. It tackles increasing concerns about surveillance, tracking, and unauthorized use of personal data by adapting and personalizing facial recognition algorithms to accommodate individual tastes. This study provides an innovative answer to the ethical and privacy concerns raised by facial recognition technologies. It empowers users to exercise greater control over their privacy and data in the face of increasingly prevalent facial recognition technologies by allowing them to selectively unlearn their data from training algorithms.