method
active
method:metabalanceMetaBalance
Improving recommendations by adapting gradient magnitudes of auxiliary tasks.
Neighborhood — ranked by edge-count
Artifacts (1)
artifact
- The paper proposing the Dual-Balancing Multi-Task Learning method.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Condition where a pattern memory cannot settle on a unique outcome, producing stochastic switching; common to cognitive and morphogenetic systems.
- The capability of GPT-3 to learn tasks from few-shot prompts during runtime.
- Affiliation of Ziyu Guo and Rain Liu.
- System's awareness of its own attentional states; the paper's central explanatory target, formalized as precision over attentional state representations.
- The problem of ensuring all tasks in MTL perform well, avoiding dominance by some tasks.
- The ability to model one's own cognition; linked to consciousness and decision-making across theories.
- A system component outside the application domain that provides infrastructure (e.g., backplane, interface repository).
- Subspaces whose contributions to a layer's output are canceled by opposing weight values, making them non-causally active under natural inputs