Cold-Start Active Preference Learning in Socio-Economic Domains

Name: Cold-Start Active Preference Learning in Socio-Economic Domains
Uploaded: 2026-10-03T07:44:55.943740+00:00
Duration: 309 s
Description: The method addresses the cold-start problem in active preference learning by using Principal Component Analysis (PCA) to create initial pseudo-labels from unlabeled data. This is followed by an active learning loop that strategically queries a simulated noisy oracle for true labels, outperforming standard methods without prior information.

The method addresses the cold-start problem in active preference learning by using Principal Component Analysis (PCA) to create initial pseudo-labels from unlabeled data. This is followed by an active learning loop that strategically queries a simulated noisy oracle for true labels, outperforming standard methods without prior information.

Open the full episode