Conception of 4-component architecture of information systems on example of Artificial Neural Networks

Nowadays Information Systems (IS) become more and more distributed, complex, and heterogeneous. Such nature of IS make them or their components a Black Box. Although classical software operates according understandable logic, modern complex software often shows non-determinism in its operation. Artificial Intelligence (AI) based on Artificial Neural Networks (ANN) is an example of such systems. This paper considers IS architecture consisting of 4 components, one of which represents non-determinism as an "Machine Intuition". The architecture is derived from 3-tier computer architecture and based on psychological findings. This approach allowed building a simple and user/developer friendly model. Practical value of the architecture is concluded in ability to better understand, design, and develop the IS containing units with non-deterministic behavior, deal with AI overfitting, underfitting, and threat problems. Architecture and principles represented in this paper can be applied not only to AI/ANN but different IS types.

---

1. Track 3: Advances in Information Systems and Technology

---

3. Communication papers