Recommendation System using Core ML or Cloud ML?

Question

Created Aug ’20

Replies 1

Boosts 0

Participants 2

Hey, I am a 17 year old aspiring programmer developing an app which would need a recommender system to provide personalized recommendations to users for new photos, videos, ads, etc. relevant to their interests.

Can this be implemented using CoreML because I am hesitant to use Cloud ML services as I want the users’ data to be private.
Are there any limitations to this? (Because I was wondering why don't all companies use on-device ML to serve ads to protect privacy?)

Boost

Answer 1

mchogan OP

Oct ’20

Hi @aedengeo.

I’m new to CoreML, so can’t answer the first part of your question yet. However, my guess is that if you can get the content on the device then CoreML can be used to process it. Have you tried it? How has your experiment gone?

In the second part of your question you asked, “why don’t all companies use on-device ML to serve ads to protect privacy?” There are many answers to that question. Two I’m aware of are that (1) devices have less computing power than racks of servers in the cloud, so on-device ML requires developers to make more efficient use of computing power; (2) not all companies have the same business model and a commitment to privacy can mean some business models can’t be used.

0