Emergent Models

Alternative machine learning from cellular automata

Emergent Models

Emergent Models is our proposal for an alternative machine learning framework.

Instead of relying on large, carefully engineered neural network architectures, we investigate whether task-solving behavior can emerge from simple, iterative, physics-inspired rules operating on minimal substrates such as cellular automata, trained through evolutionary processes.

More broadly, this line of research explores unconventional computing methods for machine learning.

Cellular Automata

Computation emerges from simple local update rules applied over time.

Evolutionary Training

Model behavior is shaped through evolutionary processes rather than standard gradient-based optimization.

Alternative Foundations

The goal is to explore a different foundation for machine learning, not just a variation of existing neural architectures.

Research

Supported through collaborations and early research funding.