Kohonen: Δ W = ∑ k = 1 R N ( X , W K ) α ( X T − W k ) {\displaystyle \Delta W=\sum _{k=1}^{R}N(X,W_{K})\alpha (X^{T}-W_{k})}
Competitive: Δ W = ∑ k = 1 R α ( X T − W k ) {\displaystyle \Delta W=\sum _{k=1}^{R}\alpha (X^{T}-W_{k})}
Gas: Δ W = ∑ k = 1 R α e − k ( X T − W k ) {\displaystyle \Delta W=\sum _{k=1}^{R}\alpha e^{-k}(X^{T}-W_{k})}
Kernel adatron:
Multipliers: δ α i = η ( 1 − γ i ) {\displaystyle \delta \alpha ^{i}=\eta (1-\gamma ^{i})\,}
Biases: b = 1 2 ( m i n ( z i ) + m a x ( z i ) ) {\displaystyle b={\frac {1}{2}}(min(z_{i})+max(z_{i}))}
daughter wavelet: Ψ j ( z ) = Ψ ( x − m j d j ) {\displaystyle \Psi _{j}(z)=\Psi \left({\frac {x-m_{j}}{d_{j}}}\right)}
Function layer: Y = f ( X + b ) {\displaystyle Y=f(X+b)\,}
Sanger: Δ W i j = α y i ( x j − ∑ k = 1 i − 1 y k w k j ) {\displaystyle \Delta W_{ij}=\alpha y_{i}(x_{j}-\sum _{k=1}^{i-1}y_{k}w_{kj})}
A little: μ A ( x ) 1.3 {\displaystyle \mu _{A}(x)^{1.3}\,}
Slightly: μ A ( x ) 1.7 {\displaystyle \mu _{A}(x)^{1.7}\,}
Very: μ A ( x ) 2 {\displaystyle \mu _{A}(x)^{2}\,}
Extremely: μ A ( x ) 3 {\displaystyle \mu _{A}(x)^{3}\,}
Very Very: μ A ( x ) 4 {\displaystyle \mu _{A}(x)^{4}\,}
More or less: μ A ( x ) {\displaystyle {\sqrt {\mu _{A}(x)}}\,}
Somewhat: μ A ( x ) {\displaystyle {\sqrt {\mu _{A}(x)}}\,}
Indeed: { 2 μ A ( x ) 2 , if 0 ≤ μ A ( x ) ≤ 0.5 1 − 2 ( 1 − μ A ( x ) ) 2 , if 0.5 < μ A ( x ) ≤ 1 {\displaystyle {\begin{cases}2\mu _{A}(x)^{2},&{\mbox{if 0 }}\leq \mu _{A}(x)\leq {\mbox{ 0.5}}\\1-2(1-\mu _{A}(x))^{2},&{\mbox{if 0.5}}<\mu _{A}(x)\leq {\mbox{ 1}}\end{cases}}\,}