NEURAL NETWORK LIBRARY IN C++

Feedforward neural network from scratch with backpropagation engine, configurable layers, gradient descent optimization with numerical stability.

Overview

A high-performance C++ library for building and training feedforward neural networks from the ground up. Features include a modular layer architecture, an efficient backpropagation engine, multiple activation functions (ReLU, Sigmoid, Tanh), and gradient descent optimizers with momentum and weight decay.