Core ML: Integrate machine learning models into your app(developer.apple.com)
developer.apple.com
Core ML: Integrate machine learning models into your app
https://developer.apple.com/documentation/coreml
40 comments
Apple's new iOS CoreML inference engine supports Keras models!
Developers will be able to design and train model using Keras and then convert the architecture to run on the CoreML engine. I suppose you can run TensorFlow models too if you designed them w/ Keras.
Yeah, that is surprising given that keras is not a language, but a continuously changing python class. How bug complete is the conversion? What about some new features like attention? And what about the future changes in keras?
I think they mentioned in the Core ML lab that the conversion tool is being open sourced - which would help with these issues.
Cool that the file format is officially specified as a Protocol Buffer: https://developer.apple.com/documentation/coreml/converting_... (bottom of page)
Is there a protobuf to capnproto converter?
Federighi says Core ML on iPhone is 6x faster than Google Pixel and Samsung Galaxy S8.
How they actually compare?
How they actually compare?
The probably ran the same tests on Android using Pixel and S8 and then on iPhone (which leveraged CoreML). I would like to see a more detailed analysis myself.
Lots of options to explore, but no reinforcement learning yet.
Also, some converted Core ML Models ready to use here: developer.apple.com/machine-learning
Also, some converted Core ML Models ready to use here: developer.apple.com/machine-learning
Reinforcement Learning relates to the way you train your model. Most of the time, it ends up being a feed-forward neural net (possibly with some convolutional layers), and rarely an RNN, all of which are supported.
The python package, coremltools, to convert the trained model is only for python 2.7??
Is python3 shipped on OS X by default? I'm not sure why they are still shipping 2.x -- anyone know if 3.x will be shipped in high sierra?
A bit disappointed that the model conversion tool only supports an older version of Keras as well (1.2.2). Keras 2.0 is pretty new but I hope they update the conversion tools for it quickly ...
I wonder if the conversion tool will be open source ... seems like they'd want to support the widest net of external models since they don't yet have a way to produce .coreml models directly. Or maybe the intent is to augment Keras/caffe/etc to support saving .coreml directly?
A bit disappointed that the model conversion tool only supports an older version of Keras as well (1.2.2). Keras 2.0 is pretty new but I hope they update the conversion tools for it quickly ...
I wonder if the conversion tool will be open source ... seems like they'd want to support the widest net of external models since they don't yet have a way to produce .coreml models directly. Or maybe the intent is to augment Keras/caffe/etc to support saving .coreml directly?
Both python 2 (/usr/bin/python) and python 3 (/usr/bin/python3) are shipped by default on Ubuntu.
Right now, coremltools is only available for Python 2.7 (https://pypi.python.org/pypi/coremltools), which is annoying as the entire code base I've worked on for months at my current firm is in Python 3.6. Hopefully this is updated soon for Python 3 support.
Right now, coremltools is only available for Python 2.7 (https://pypi.python.org/pypi/coremltools), which is annoying as the entire code base I've worked on for months at my current firm is in Python 3.6. Hopefully this is updated soon for Python 3 support.
macOS comes with Python 2 by default.
Having a directory of trained models to download is an interesting concept. This will certainly accelerate the adoption of ML.
Does it support caffe/TensorFlow/MXnet model inferencing?
Why not caffe2? And why mention libsvm, are we in the 90s?
Because SVM is fast and more than good enough for most problems.
I think he's talking about libsvm, the c++ package
It's there because they want to include SVM.
So, same answer.
So, same answer.
And why mention libsvm, are we in the 90s.
"Core ML is optimized for on-device performance, which minimizes memory footprint and power consumption."
This is major, if they have managed to achieve it reasonably. But before opening a Sekt, I want to see some benchmarks. :)
This is major, if they have managed to achieve it reasonably. But before opening a Sekt, I want to see some benchmarks. :)
"BNNS does not do training, however. Its purpose is to provide very high performance inference on already trained neural networks." :(
https://developer.apple.com/documentation/accelerate/bnns
https://developer.apple.com/documentation/accelerate/bnns
yes, it's an inference engine that runs pre-built architectures. why would you want to train on the device?
Federated Learning. It's something Google does at the Android level, but I don't think they provide an API for app developers to train on device.
Another use case is privacy: ship a pretrained facenet style model and let it learn to map your address book to faces in your photos without your data leaving the device.
Another use case is privacy: ship a pretrained facenet style model and let it learn to map your address book to faces in your photos without your data leaving the device.
Maybe not fully train but fine tune on the user data ( images, voice, surroundings, etc, etc)
oh, you're talking about updating parameters. Maybe sending that new data home and train a new model and then update parameters across all devices in fleet.
> Maybe sending that new data home
I'd assume Apple does not allow this.
I'd assume Apple does not allow this.
Apple would allow this with privacy statements, it's practically the industry standard right now. Of course Facebook is doing this for instance.
However it could be a killer feature to have easy api's for doing personalization all on device.
However it could be a killer feature to have easy api's for doing personalization all on device.
BNNS = “Basic neural network subroutines” if you are not interested in clicking through!
When "Core ML" is no longer a mini language to reason about ML the PL semantics...
Yeah, it crossed my mind as well.
But nowadays every time I see ML on HN, it happens to mean ML = Machine Learning instead of ML = programming language.
But nowadays every time I see ML on HN, it happens to mean ML = Machine Learning instead of ML = programming language.
It's also going to make the name ReasonML even more confusing, considering you can apply the word "Reason" to machine learning...
Someone in the ML programming language community should use the name DeepML to further sow the seeds of confusion.
Machine Learning needs a new acronym. ML is already taken! Get your own!
Which of these 109 entries is objectively the definitive one?
http://www.acronymfinder.com/ML.html
http://www.acronymfinder.com/ML.html
At this point, I would like to nominate Malt Liquor.
using it means need to handle Android with another framework separately
Is there any cross-platform framework recommended?