Papers
20
Total Citations
214
H-Index
9
About
Shingo Mabu is a computational intelligence researcher whose work spans evolutionary algorithms, graph-based optimization, and intelligent robotics systems. His most significant contributions lie in the development and refinement of Genetic Network Programming (GNP), a graph-structured evolutionary framework that enhances traditional genetic algorithms through richer representational capacity. By integrating reinforcement learning into GNP, Mabu extended the paradigm's applicability to dynamic, real-world tasks such as mobile robot navigation and adaptive behavior generation in changing environments — work that has accumulated over 22 citations in its foundational form. Among his most impactful contributions is a graph-based Estimation of Distribution Algorithm (EDA) that leverages reinforcement learning to build probabilistic models, earning 37 citations and representing a notable advance in avoiding building-block disruption common in conventional evolutionary approaches. He further innovated by incorporating infeasible individuals into probabilistic model building, challenging the prevailing assumption that only high-quality solutions inform useful distributions. Beyond optimization theory, Mabu has made applied contributions in computer vision and robotics, including a deep learning-based citrus-harvesting visual system (30 citations) and gesture recognition using self-organizing maps. Together, his body of work — spanning over a decade — reflects a sustained effort to bridge theoretical evolutionary computation with practical intelligent systems applications.
Research Focus
Key Achievements
Top Papers
- 1
- 2A Visual System of Citrus Picking Robot Using Convolutional Neural Networks30 citations · 2018
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10