Multi-task learning gradient balancing

Methods addressing loss-scale and gradient-magnitude imbalances in multi-task learning, with DB-MTL achieving state-of-the-art results on dense prediction benchmarks like NYUv2.

3 members. Each node is clickable.

Loading graph…

Drawn from 1 source

The papers/notes whose extracted claims & findings make up this cluster.

Dual-Balancing for Multi-Task Learning3 members

Bridges (2)

Other communities that share members with this one — cross-cutting threads or papers that sit at the seam between two themes.

Dual balancing multi-task learning3 shared
Dual-balancing multi-task learning3 shared

Claims (2)

DB-MTL is a simple yet effective method that addresses both loss-scale and gradient-magnitude imbalances.Core claim of the paper.
Task balancing requires simultaneous consideration of both loss scales and gradient magnitudesCore interpretive position of DB-MTL: complementarity of loss and gradient perspectives

Findings (1)

DB-MTL achieves ∆p = +1.15±0.16 on NYUv2, outperforming all baselines including state-of-the-artPrimary empirical validation on scene understanding task