Random Loss Weighting (RLW)

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• RL algorithmsconcept0.721
  The different reinforcement learning algorithms used across conditions, to ensure the alignment result is not algorithm-specific.
• Uncertainty Weighting (UW)method0.711
  Loss balancing using homoscedastic uncertainty.
• loss-scale balancingconcept0.707
  Addressing disparity in loss magnitudes across tasks at the loss level
• Reinforcement learning (RL)concept0.699
  Machine learning paradigm where agents learn to maximize cumulative reward through interaction.
• Random Latent Ablation Controlmethod0.690
  Control experiment ablating random latents matched for activation frequency and magnitude to test OTD specificity
• Equal Weighting (EW)method0.688
  Baseline that minimizes sum of task losses with equal weights.
• Positive Pull Loss (L_dist)method0.688
  Distance-based loss comparing injected representation to class centroids in the active subspace
• Equal Weightingframework0.685
  Baseline MTL approach minimizing sum of task losses with equal weights; suffers from task balancing