Download

Download Poster (780 KB)

Faculty Advisors

Dr. Eric Schearer

Description

Currently, there are few comprehensive guides with detailed development procedures by which brain.js is used for creating ANNs (Artificial Neural Networks) that can recognize patterns in large amounts of data (big data) in sequential and multi-dimensional form.This work is a guide detailing the development of AI (Artificial Intelligence) using brain.js, a library of features which allow for the high-level development of ANNs, all purely written in the JavaScript programing language. It details an Ensemble Styled ML (Machine Learning) procedure with brain.js. The methods in this guide include: the creation of a set of sequential eight-digit binary patterns, its conversion into a JSON format dataset, the partitioning of the dataset, the set-up of a server side node.js JavaScript environment with the ES6 modules importing system - including the brain.js library in the environment - using brain.js to implement LSTM Time Step ANN models in JavaScript, individually training the models on each dataset partition, serializing the models, creating an averaging and voting algorithm which groups and analyzes the collective output of all serialized models upon input and lastly analyzing the model’s performance upon input of trained and untrained data. The goal of this guide is to provide useful basic technical information that is accessible to the inexperienced and yet provide a scalable procedure which the experienced can leverage for more complex ML applications.

Publication Date

2024

College

Washkewicz College of Engineering

Department

Center for Human Machine Systems

Student Publication

This item is part of the McNair Scholars Program.

A Guide to Ensemble Styled Machine Learning with JavaScript and Brain.js

Share

COinS