Crate rfann

Pure Rust implementation of the Fast Artificial Neural Network (FANN) library

This crate provides a modern, safe, and efficient implementation of neural networks inspired by the original FANN library, with support for generic floating-point types. Includes full cascade correlation support for dynamic network topology optimization.

Re-exports

pub use activation::ActivationFunction;

pub use connection::Connection;

pub use layer::Layer;

pub use network::Network;

pub use network::NetworkBuilder;

pub use network::NetworkError;

pub use neuron::Neuron;

pub use training::ParallelTrainingOptions;

pub use training::TrainingAlgorithm as TrainingAlgorithmTrait;

pub use training::TrainingData;

pub use training::TrainingError;

pub use training::TrainingState;

pub use cascade::CascadeConfig;

pub use cascade::CascadeError;

pub use cascade::CascadeNetwork;

pub use cascade::CascadeTrainer;

pub use errors::ErrorCategory;

pub use errors::RuvFannError;

pub use errors::ValidationError;

Modules

activation

cascade

Cascade Correlation Training for Dynamic Network Topology Optimization

connection

errors

Comprehensive error handling system for rfann

integration

Integration utilities for combining all agent implementations

io

I/O and serialization module for rUv-FANN

layer

memory_manager

Memory management for efficient neural network operations

mock_types

Mock types for testing I/O functionality

network

neuron

simd

SIMD optimizations for neural network computations

training

Training algorithms for neural networks

webgpu

WebGPU compute backend for rfann neural networks

Macros

cascade_error

network_error

Helper macros for error creation with context

training_error

Enums

TrainingAlgorithm

Enumeration of available training algorithms