framework
active
framework:singular-learning-theorySingular learning theory
Relates parameters to loss via landscape geometry.
Neighborhood — ranked by edge-count
Papers (3)
paper
Claims (1)
claim
- Current decomposition methods lack solid theoretical foundations and the nature of features is unclearanalogous_toCalls for conceptual clarity.
Frameworks (1)
framework
- Stochastic Parameter Decomposition (SPD)associated_withLinear parameter decomposition with stochastic ablation.
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.
- Process of inferring causes of sensory information; unified with value learning as integral aspects of free energy minimization.
- Inference of parameters encoding contingencies of the world (e.g., likelihood matrix A) at slower timescale than perception.
- Framework for solving inverse problems in which physical systems autonomously adapt their parameters in response to stimuli through local learning rules, without requiring computational design or explicit cost functions
- Learning model parameters through curious, uncertainty-reducing behavior; reducing ignorance about contingencies
- Learning that builds a low-dimensional model of input data without error signals or rewards; Hebbian learning is an example.
- Sentience criterion; capacity occurs even in gene regulatory networks and non-neural morphogenetic agents.
- Learning paradigm that jointly learns multiple related tasks using a single model
- Acquisition of new concepts by Bayesian model expansion and reduction.