Universal approximation theorem: Difference between revisions

Content deleted Content added
Formal statement: ehm, i guess that was a typo? doesnt make sense otherwise
Formal statement: cleanup (easier to read)
Line 13:
<blockquote>
 
Let φ(·) be a nonconstant, bounded, and [[monotonic function|monotonically]]-increasing [[continuous function|continuous]] function. Let ''I''<sub>''m''<sub>0</sub></sub> denote the ''m''<sub>0</sub>-dimensional unit hypercube [0,1]<sup>''m''<sub>0</sub></sup>. The space of continuous functions on ''I''<sub>''m''<sub>0</sub></sub> is denoted by ''C''(''I''<sub>''m''<sub>0</sub></sub>). Then, given any function ''f'' ∈ ''C''(''I''<sub>''m''<sub>0</sub></sub>) and є &gt; 0, there exist an integer ''m''<sub>1</sub> and sets of real constants ''α''<sub>''i''</sub>, ''b''<sub>''i''</sub> and∈ ℝ, ''w''<sub>''iji''</sub>, where ''i'' = 1, ..., ℝ<sup>''m''<sub>10</sub></sup>, andwhere ''ji'' = 1, ..., ''m''<sub>01</sub> such that we may define:
 
: <math>
F( x_1 , \dots, x_{m_0} ) =
\sum_{i=1}^{m_1} \alpha_i \varphi \left( \sum_{j=1}w_i^{m_0}T w_{i,j} x_jx + b_i\right)
</math>