Dynamic Fuzzy Machine Learning

2) Gradual rule

In this case, the value of the degree of coincidence is only useful for the comparison conditions and the conclusion. As a result, it is no longer the same as the previous treatment of the coincidence rate of high α-explanatory type.

$\frac{\begin{array}{l}c{f}_{grad}\left(A\left(\overleftarrow{t},\overrightarrow{t}\right)\to C\left(\overleftarrow{t},\overrightarrow{t}\right)\right)=\\ |\{\left(\overleftarrow{x},\overrightarrow{x}\right)\in {\mathcal{H}}^{II}\left|I_{ILP}\right|=\sigma \left[\left(\overleftarrow{t},\overrightarrow{t}\right)/\left(\overleftarrow{x},\overrightarrow{x}\right)\right]\left(A\wedge C\right),\mu \left(\sigma \left[\left(\overleftarrow{t},\overrightarrow{t}/\overleftarrow{x},\overrightarrow{x}\right)\right]C\right)\\ \ge \mu \left(\sigma \left[\left(\overleftarrow{t},\overrightarrow{t}\right)/\left(\overleftarrow{x},\overrightarrow{x}\right)\right]A\right)|\}\end{array}}{\left|\left\{\left(\overleftarrow{x},\overrightarrow{x}\right)\in {\mathcal{H}}^n\left|I_{ILP}\right|=\sigma \left[\left(\overleftarrow{t},\overrightarrow{t}\right)/\left(\overleftarrow{x},\overrightarrow{x}\right)\right]\left(A\right)\right\}\right|}$

When the part values of the rules are a conjunction of the basic text, ...