English versionEN

Build Intelligence at the Edge - Machine Learning with React Native

Have you ever wondered if we can really build machine learning models in React, rather than in the mainstream languages like Python or R? Afterall, React is the most used language by the developers, according to a 2019 survey by Stack Overflow. Well, this sounds like a crazy idea, because React is not designed for high performance computing and neural networks are compute-intensive! But, wait a minute - we have libraries such as Onnx.js, Tensorflow.js to our rescue! In this talk, I’ll be delving deeper into the process of building and deploying machine learning applications using React.

This talk has been presented at React Day Berlin 2023, check out the latest edition of this React Conference.

Watch video on a separate page

FAQ

Machine learning is a subset of artificial intelligence (AI) where algorithms are programmed to learn from data and make decisions or predictions based on that data. It involves training models using large datasets to perform tasks like spam detection or trend forecasting.

Edge computing offers several benefits for machine learning, including cost reduction by minimizing cloud computing expenses, improved performance in environments with unstable internet connections, reduced network latency, and enhanced security and privacy for sensitive data.

Common applications include recommendation systems, patient monitoring systems in healthcare, and predictive maintenance alerts in industrial settings. These applications leverage the real-time processing capabilities of edge devices to deliver timely and relevant results.

Challenges include memory constraints due to the limited capacity of edge devices, data quality issues from lower-quality sensors, and computational limitations that affect the model's ability to process data efficiently.

Optimizing models for edge devices involves techniques like model quantization and pruning, choosing the right framework for the task, and employing concurrency to enhance computational efficiency. Benchmarking models at various stages of the machine learning pipeline is crucial for achieving optimal performance.

Deploying models on edge devices keeps personally identifiable information (PII) local, reducing the risk of data breaches during transmission over networks. This localized processing helps maintain user privacy and enhances overall data security.

Resources for further learning include online courses, technical documentation of React Native and relevant ML libraries, community forums, and expert talks. These resources can provide detailed guidance and practical examples for implementing ML models on edge devices.

Machine learning models can be deployed on edge devices using React Native by installing necessary packages, creating a model with various layers, and training the model with datasets. This process involves using JavaScript frameworks and libraries to handle machine learning operations within a React Native application.

machine learning

Rashmi Nagpal

13 min

12 Dec, 2023

Comments

Video Summary and Transcription

The Talk is about building intelligence at the edge with machine learning and React Native. It covers machine learning concepts, building ML models with React, challenges, best practices, and resources.

Available in Español: Construir Inteligencia en el Borde - Aprendizaje Automático con React Native

1. Introduction to Building Intelligence at the Edge

Short description:

Hi everyone, I'm Rashmi Nagpal. Today my talk is about building intelligence at the edge with machine learning and react native. We'll discuss machine learning concepts, building ML models with React, challenges, best practices, and resources.

Hi everyone, I'm Rashmi Nagpal. I'm a software engineer by profession and a researcher by passion. So today my talk title is build intelligence at the edge with machine learning and using react native.

So let's begin. So the agenda of this talk is firstly we'll discuss what is machine learning and its related concepts and how we can build our machine learning model using the react as one such technical framework.

Then there will be like some of the applications which are possible. Then we'll see what are the challenges which exist while using machine learning on the edge and then what are the best practices so that we can overcome those challenges and bunch of resources that I'll leave for you for the so without any further ado let's begin.