finding

active

finding:curve-detecting-neurons-found-in-every-non-trivial-vision-model-carefully-examined

Curve detecting neurons found in every non-trivial vision model carefully examined

Empirical basis for treating curve detectors as a canonical example of meaningful, understandable features

Source paper

extracted_from

Zoom In: An Introduction to Circuits

(2020) · Chris Olah · Nick Cammarata · Ludwig Schubert · Gabriel Goh +2

Neighborhood — ranked by edge-count

Claims (1)

claim

Features are the fundamental unit of neural networks; they correspond to directions and can be rigorously studied and understood.
supports
First of three speculative claims forming the foundation of the circuits research agenda

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.

Rosetta Neurons — individual neurons activated by the same patterns across a range of diverse vision models form a common dictionary independently discovered by all modelsfinding0.792
Cited evidence that convergence extends to the neuron level, not just representational geometry
Curve detectors found across AlexNet, InceptionV1, VGG19, ResNetV2-50 and models trained on Places365finding0.780
Anecdotal evidence for the universality of low-level visual features across different architectures and datasets
Can Neurons From Modalities Such As Vision Bequestion0.780
Neurons can correspond to interpretable functional roles but interpretations in terms of individual neurons are unlikely to be the most parsimoniousclaim0.757
Claim from footnote 3, acknowledging neuron-level interpretability while arguing subcomponents are better.
Neuron cluster identification via partitioningmethod0.752
Method used to identify and partition the 28 MLP neurons into disjoint clusters by Fourier period
Superposition hypothesis: neural networks represent more features than dimensions using almost-orthogonal directions.hypothesis0.747
Explanation for why dictionary learning can recover many more features than dimensions.
Olah et al. (2020) found that automatically trained computer vision models, regardless of architecture and training procedure, all arrive at similar functional structures organizing similar features into similar compositional hierarchies, closely resembling the primate visual cortex.finding0.746
Empirical finding supporting the Universality Hypothesis; extended by the paper to consciousness
Non-neural morphogenetic agents satisfy most sentience criteria via electrically active cells rather than neuronsfinding0.741
Empirical basis for expanding sentience frameworks; shows Crump criteria adaptable beyond traditional neurocentric definitions.