From the course: Machine Learning in Mobile Applications
Join today to access over 24,000 courses taught by industry experts.
Why use client-side vs. server-side models?
From the course: Machine Learning in Mobile Applications
Why use client-side vs. server-side models?
We know that model creation doesn't usually happen today on mobile devices, but doesn't answer the question of where should the model run. Where the model is created, and the need to access large amounts of data to create the model in the first place, really shouldn't be large deciding factors when considering where the model should be run. When looking at what does matter, there are several advantages of running the model server-side. Server-side machine learning offerings usually have pre-built endpoints to send back information for retraining. This makes creating a feedback mechanism relatively simple for a developer. Since the model's server-side, and all the evaluation of it happens there, any retraining that takes place to update the model can immediately be used by all clients. They also tend to be implemented using REST APIs, so it doesn't really matter if a client SDK exists or not. Sure, the client SDKs can make the work easier, but most client applications are going to be…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.