An Elixir framework for evolutive neural networks with a modular DSL and OTP-based architecture.
EXNN is an Elixir framework for evolutive neural networks, enabling the creation of adaptive neural systems through genetic algorithms and real-time fitness evaluation. It provides a modular DSL for configuring sensors, actuators, and network topologies, built on Elixir's OTP for concurrency and fault tolerance.
Elixir developers and researchers interested in neuroevolution, genetic algorithms, or building adaptive, learning-based systems with neural networks.
Developers choose EXNN for its Elixir-native implementation, OTP-based architecture, and modular DSL, offering a flexible and concurrent approach to evolutive neural networks compared to traditional frameworks.
An Elixir Evolutive Neural Network framework à la G.Sher
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Uses Elixir macros for easy definition of sensors, actuators, and fitness functions, as shown in the configuration example with clear, code-based setup.
Built on Elixir/OTP with GenServer processes, enabling fault-tolerant and concurrent neural network components, leveraging Erlang's battle-tested runtime.
Fitness modules assess performance in real-time to guide evolution, allowing adaptive learning systems without batch processing, as described in the core concepts.
Initial patterns define network structure with support for layered neurons, facilitating evolutive changes in topology through genetic algorithms.
The framework is in a very early stage and can currently only train a trivial XOR function, severely limiting its practical use for complex tasks.
Documentation is being slowly built, with only basic ex_doc available, which hinders onboarding and troubleshooting for new users.
Requires deep knowledge of Elixir and OTP patterns, making it inaccessible for developers from other backgrounds and increasing the learning curve.
Lacks advanced neural network capabilities and instrumentation, with future plans indicating missing features like exploratory search or production states.