Things you need to know about Apple's Core Machine Learning

Pros and cons of Apple's Core Machine Learning

Apple has changed the game with its introduction of Core ML and moved the finish line into devices improving battery life and performances drastically. This helps to grow the market for mobile app development in London. Machine learning is proving to be foundational tech that will be in everything and not some light frosting on application code. Apple’s Core ML, although it is very limited right now, points to a mainstream future for machine learning.

Machine learning works on our devices

ML completely depends on training data from large sets. Once you are done figuring out the predictive model you have to feed the machines with large quantities of data to train them to understand it and keep tuning the model. Just because such training sets require such a large amount of data and a lot of computing power, ML has only been possible on the cloud.

With the introduction of Core ML, Apple is moving towards pushing machine learning on their devices and if rumors about the latest device iPhone X are true then there will be a separate chip dedicated only for AI in the new device. Although Apple will need to do the initial machine learning using the cloud, pushing the learning to its devices will mean a significant boost in benefits.

Core ML will minimize memory footprint and significantly reduce power consumption as it is optimized for on-device performance. The on-device processing means greater privacy and security for user data and you IOS app remains functional even if no network connection is available.

Core ML is not all that merry as it sounds

Core ML does not have any provisions for federated learning or model retraining, where model accuracy is improved by the data collected in the field. That will have to implement locally by asking IOS app users to opt for data collection. Once the data is collected it can be used to retrain the model for a future version of the app.

• Here’s how federated learning works,

Your device will download the current model. By learning from data collected on your phone it will summarize the responses as a small update patch. And only this patch is sent to the cloud through encrypted communication. Here all those updates collected from various devices are averaged to improve the shared model. No individual updates are stored in the cloud.

According to the iOS app developer instead of utilizing a large number of servers in the cloud, you can easily use an army of mobile devices out there which would be more efficient. The improved model is immediately available to the device makes the user experience personalized.

Apple seems to be falling behind its peers again.

iCloud, Apple Maps and Siri is either a late or underpowered tech when compared to the cloud and AI heavyweights like Google. AWS (Amazon Web Services) also faced criticism when they released developer-facing machine learning services like Recognition, Polly, and Lex about its basic nature and limitations. According to Swaminathan Sivasubramanian, general manager for AWS, he said, the goal “is to bring machine learning to every AWS developer and not to overwhelm them with the inherent complexity of machine learning”. Apple is creating an easy pathway to getting started with machine learning in a similar manner. It not perfect but it will raise a generation of new developers for machine learning.

App Developers in London are of the view that Apple should probably have kept Core ML Opensource, thus giving developers a chance to customize Core ML to their needs. Most other machine learning toolkits are open source then why not Core ML as well.

Although Core ML is an astounding piece of tech it is still limited and in its infancy, it is still a step closer to machine learning in everything.closer to machine learning in everything.