framework
active
framework:temporal-feature-analysisTemporal Feature Analysis
Proposed objective decomposing activations into predictable and novel components.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (3)
method
- UMAPusesUsed to visualize models in 2D space based on representational distance
- TortuosityusesMeasure of trajectory straightness.
- Used to test event alignment.
Concepts (2)
concept
- Novel componentusesFast-moving residual orthogonal to context.
- Slow-moving context-derived part of activation.
Claims (1)
claim
- Acknowledged limitation.
Frameworks (2)
framework
- A second-order correlational similarity method compared against MAS in the paper.
- Kalman filtersanalogous_toAnalogy to TFA structure and suggested training objective.
Artifacts (2)
artifact
- NeuronpediaaboutSource of auto-interp labels.
- Released code.
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.
- Interpretation of smooth predictive trajectories.
- Method of optimizing input to cause a neuron to fire maximally, used to characterize what a neuron detects; establishes causal link
- §7.2 assertion.
- Interpretable concept representations hypothesized as basic units of computation.
- Method comparing brain activity in conscious vs. unconscious conditions.
- In active inference, distant future predictions are naturally less precise, leading to Bayes-optimal discounting without an explicit discount factor.
- Examining downstream neurons that rely on a given feature to verify its functional role
- Component of the contrastive retrieval pipeline analyzing activation statistics.