Shape memory polymers (SMPs) and its fabrication process has recently attracted much attention because of their potential application as soft active materials. In this work, a dimension accuracy approach using artificial neural networks (ANN) is presented for overcoming the dimensional challenge in shape memory polymer/graphene oxide (SMP/GO) composites using projection-type Stereolithography (SL) 3D printing. Experimental trials were conducted to achieve proper SMP photo-resin mixing (monomer, cross-linker, photo-initiator) and suitable GO dispersion in SMP/GO composite. An artificial neural network (ANN) was designed based on back propagation theory for modelling dimensional error on specimens. ANN training and testing phases used Stereolithography (SL) historical data and results were compared using two different ANN architecture.  Using the ANN approach, this paper reports a maximum Pearson correlation of 77.7 % during testing. It could be used as a reference for fabrication, and dimensional error modelling of SMP/GO composites fabricated via SL 3D printing technique.