Brief

Researchers have developed a machine unlearning technique to remove the influence of specific subjects from trained models without requiring a full retraining process. This method, applied to engagement recognition datasets like DAiSEE and EngageNet, aims to sanitize models by identifying and excluding problematic data subsets. The unlearned models achieved significant performance recovery, nearing that of models retrained from scratch, at a fraction of the computational cost. AI

Researchers have developed PriorNet, a novel framework designed to improve engagement estimation from face videos. This system addresses challenges like incomplete facial data and subjective annotations by incorporating task-specific priors at multiple stages of the process. PriorNet utilizes techniques such as zero-frame placeholders for missed detections, parameter-efficient adaptation of a pre-trained backbone, and a specialized training objective to enhance accuracy. AI