NeuralN

NeuralN is a neural network library built with C++ for Node.js, designed to overcome JavaScript's memory limitations and accelerate training times. It enables developers to work with datasets larger than Node.js's default memory constraints and leverages multi-threading for faster model training.

Key Features

• Large Dataset Support — Handles datasets exceeding Node.js's 1GB memory limit by utilizing C++ memory management.
• Multi-Threaded Training — Parallelizes training across multiple threads to significantly reduce learning phase duration.
• Flexible Network Configuration — Allows instantiation with custom layers, momentum, learning rates, and bias values.
• State Persistence — Provides methods to serialize and deserialize network states via string or JSON representations.
• Training Control — Supports both single-threaded (blocking) and multi-threaded (non-blocking) training with configurable parameters.

Philosophy

NeuralN prioritizes performance and scalability, leveraging C++ to bypass Node.js's memory limitations and using multi-threading to make large-scale neural network training practical.