\r\nautomatically constructed from learning data, are based on the

\r\nsteepest descent method (SDM). However, this approach often

\r\nrequires a large convergence time and gets stuck into a shallow

\r\nlocal minimum. One of its solutions is to use fuzzy rule modules

\r\nwith a small number of inputs such as DIRMs (Double-Input Rule

\r\nModules) and SIRMs (Single-Input Rule Modules). In this paper,

\r\nwe consider a (generalized) DIRMs model composed of double

\r\nand single-input rule modules. Further, in order to reduce the

\r\nredundant modules for the (generalized) DIRMs model, pruning and

\r\ngenerative learning algorithms for the model are suggested. In order

\r\nto show the effectiveness of them, numerical simulations for function

\r\napproximation, Box-Jenkins and obstacle avoidance problems are

\r\nperformed.","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 111, 2016"}