framework
active
framework:robust-features-hypothesis

Robust features hypothesis

Models exploit non-robust but predictive patterns.

Neighborhood — ranked by edge-count

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

  • Robustnessconcept0.803
    Ability to maintain function despite perturbations.
  • Model Robustnessconcept0.778
    Area of AI research that uses interventions to test and improve model resilience to perturbations
  • The functional solidity and working character of natural systems, arising from the fifteen properties.
  • Finds data clusters where response concept is chosen/rejected leaning.
  • Reward Hypothesisconcept0.747
    The claim in RL that any goal can be expressed as maximizing the expected cumulative sum of a scalar reward signal.
  • Ability to entertain competing hypotheses within one inference engine; proposed hallmark of mindful inference
  • Synthesizing inputs maximizing component activation.
  • Featuresconcept0.736
    Interpretable concept representations hypothesized as basic units of computation.