Is it possible to use Encog for complex semi-reccurent network structures?

I'm currently using Neuroph to build a network to evaluate complex logical expressions.

The method is called CIL2P for anyone interested and the method generates semi-recurrent networks (only recursive connections between some neurons).

开发者_运维知识库Neuroph is able to build this structure quite easily with its high level approach to network design. However the performance and further training possibilities I'd really like to move this to the Encog Engine.

Is it possible to recreate this kind of network within Encog?