パーセプトロン

Input: Training data $\mathcal{D} = \{(x, y)\}$, Learning Rate $\eta$, Initial weights $w$, Bias $b$

for each epoch do

$\text{errors} = 0$

for each $(x, y) \in \mathcal{D}$ do

$\hat{y} = \text{Activation}(w \cdot x + b)$; // Forward pass

$e = y - \hat{y}$; // Error computation

if $e \neq 0$ then

$w = w + \eta \cdot e \cdot x$; // Update weights

$b = b + \eta \cdot e$; // Update bias

$\text{errors} = \text{errors} + 1$

end for

if $\text{errors} = 0$ then break; // Convergence check

end for

Output: Trained parameters $(w, b)$

Mathematics is the language with which God has written the universe.





















minibpe 希少性に伴う戦略の切り替え 追記型アーキテクチャ WAL WALバッファ EML演算子の万能性