Cyber Elephant Tech Team

Aug 4, 2021 · Artificial Intelligence

How Tag-Based Explicit Recall Boosts Recommendation Performance with Multi-Task Learning

This article explains how a two‑stage recommendation pipeline uses explicit tag‑based recall, inverted indexes, and a multi‑task learning model to improve click‑through and dwell time by dynamically balancing loss weights across tasks.