NEATsuch

NEATsuch is a theoretical framework in artificial intelligence and computational creativity that extends the NeuroEvolution of Augmenting Topologies (NEAT). It emphasizes the integration of explicit constraints or objectives into the evolution process, so that evolving neural architectures not only optimize fitness but also satisfy predefined properties. The name suggests a combination of NEAT principles with constraint-oriented refinement.

Core ideas include a constraint module that encodes "such that" conditions, a fitness function reflecting multiple objectives (performance, efficiency, stylistic attributes), and a speciation mechanism to maintain diversity while steering populations toward compliant designs. Unlike standard NEAT, NEATsuch prioritizes feasibility of generated networks as a primary objective alongside conventional performance.

Applications of NEATsuch appear in procedural content generation, texture synthesis, and generative art, where designers require networks that produce consistent outputs within resource limits or adhere to a given aesthetic. It is explored in research discussions as a way to combine evolvable architectures with controlled behavior, enabling more predictable creative outcomes.

Status: As a conceptual extension with varying implementations, NEATsuch is not a widely standardized framework, and practical work often uses related approaches that merge neuroevolution with constraint programming or multi-objective optimization. The topic serves to examine the balance between exploration and constraint satisfaction in evolving autonomous systems.

Related topics include NEAT, neuroevolution, constraint-based optimization, and generative art.