claim
active
claim:curve-neurons-wrapping-at-180-are-reading-from-the-second-harmonic-discovered-by-bsfsCurve neurons wrapping at 180° are reading from the second harmonic discovered by BSFs.
Recontextualizes Olah et al. 2020.
Source paper
extracted_from(2026) · Thomas Fel · Matthew Kowal · Mozes Jacobs · Dron Hazra +21
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Higher-order modes.
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.
- Empirical basis for treating curve detectors as a canonical example of meaningful, understandable features
- Curve manifold recovery.
- Neuroscience analogy.
- Superposition hypothesis: neural networks represent more features than dimensions using almost-orthogonal directions.hypothesis0.727Explanation for why dictionary learning can recover many more features than dimensions.
- Conclusion of null test.
- Superposition hypothesis: networks represent more features than neurons as almost-orthogonal directionshypothesis0.722Central hypothesis.
- Structural finding showing modular organization within the sparse neuron set
- The 28 identified neurons can be partitioned into disjoint clusters each computing a different Fourier period sum