method
active
method:gradient-methodGradient method
Optimization technique that computes weight changes by following the gradient of an error function; contrasted with evolutionary stochastic search.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (1)
method
- Gradient Descentrelated_toUsed for updating hidden state expectations; provides dynamical process theory testable against neuronal data
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.
- The property that qualities vary slowly, subtly, gradually across the extent of each living thing; gradients arise as natural responses to changing circumstances and create field-like character that points toward and establishes centers
- Method for assessing consciousness in nonhuman animals by identifying behavioral/anatomical markers from humans and extrapolating; proposed adaptation for AI.
- Baseline method against which probe-based ranking is compared; more computationally expensive.
- When gradients of different tasks have negative cosine similarity, harming multi-task learning.
- A structure-preserving transformation: using gradual change across space to soften and intensify transitions.
- Gradient that tells a cell its correct position; stress arises from deviation from this gradient.
- During RL training on ATLAS, sparse functional tokens (2.3% of sequences) receive diluted gradient signals from sequence-level advantages propagated across all tokens.
- Optimization procedure for simultaneously updating action selection and perception; uses step size ζ (default 4).