✳flâneur — a map of the web's best reading

Hello, KAN! — Kolmogorov Arnold Network documentation

kindxiaoming.github.io · 653 words · saved by 1 readers

Kolmogorov-Arnold representation theorem states that if 𝑓 is a multivariate continuous function on a bounded domain, then it can be written as a finite composition of continuous functions of a single variable and the binary operation of addition. More specifically, for a smooth 𝑓 : [ 0 , 1 ] 𝑛 → 𝑅 , where 𝜙 𝑞 , 𝑝 : [ 0 , 1 ] → 𝑅 and Φ 𝑞 : 𝑅 → 𝑅 . In a sense, they showed that the only true multivariate function is addition, since every other function can be written using univariate functions and sum. However, this 2-Layer width- ( 2 𝑛 + 1 ) Kolmogorov-Arnold representation may not be smooth due to its limited expressive power. We augment its expressive power by generalizing it to arbitrary depths and widths. The Kolmogorov-Arnold representation can be written in matrix form where We notice that both 𝛷 i n and 𝛷 o u t are special cases of the following function matrix 𝛷 (with 𝑛 i n inputs, and 𝑛 o u t outputs), we call a Kolmogorov-Arnold layer: 𝛷 i n

Hello, KAN! - Kolmogorov Arnold Network documentation Hello, KAN! View page source Hello, KAN!  Kolmogorov-Arnold representation theorem  Kolmogorov-Arnold representation theorem states that if \(f\) is a multivariate continuous function on a bounded domain, then it can be written as a finite composition of continuous functions of a single variable and the binary operation of addition. More specifically, for a smooth \(f : [0,1]^n \to \mathbb{R}\) , \[f(x) = f(x_1,...,x_n)=\sum_{q=1}^{2n+1}\Phi_q(\sum_{p=1}^n \phi_{q,p}(x_p))\] where \(\phi_{q,p}:[0,1]\to\mathbb{R}\) and \(\Phi_q:\mathbb{R}\

Explore this link on the map →

related reading