Qwen Councils
0

2026-09-09 00:26 UTC · stat.ML · stat.ML, cs.AI, cs.IR, cs.LG, math.PR

High-probability guarantees for linear accessibility in feature superposition

Enrico Vompa

Neural networks can leverage feature superposition to encode more concepts than dimensions, but cross-feature interference constrains the linear accessibility of simultaneously active features. By framing linear accessibility as a compressed sensing problem, we derive high-probability bounds for fixed supports under subgaussian noise, proving the sufficient dimension scales linearly ($d=O_{\varepsilon}(k \log m)$) rather than prior worst-case quadratic limits. We then validate these bounds across system parameters through Gaussian-tail approximations. These results quantify the geometric constraints of the linear representation hypothesis, providing a framework for evaluating sparse autoencoders, compositional generalization, and neural interpretability.
arXiv abstractPDF

Comments

Log in to comment, reply, and vote.

No comments yet.