mobileML

mobileML lets you harness the power of neural networks at your fingertips!

***MAIN FEATURES***

LEARN about neural networks

- mobileML includes a detailed Help Center with in-depth tutorials about how to use mobileML and the mathematics behind basic neural networks

- Download sample networks from MagMHJ.com to learn how neural networks work inside mobileML

DESIGN your own neural networks

- Construct your own fully-connected neural networks with custom dimensions

- Add input nodes that are updated with sensor data, stock prices, cryptocurrency prices, microphone data, or many other sources

- Use data from HealthKit to create networks to help you manage your health more effectively

- Apply presets to automatically generate common neural networks

- Edit every nodes value, edges value, and layers activation function

TRAIN neural networks

- Live train networks to generate large datasets

- Train networks with HealthKit data to develop networks that make health management easier

- Automatic control of learning rate during network training (optional)

- Retrain networks on datasets saved earlier or imported datasets from elsewhere

- Export datasets as a text file and share easily with friends or import into other networks

RUN neural networks and automate tasks

- Automatically propagate through networks each time input nodes are updated

- Attach actions, like sending a Bluetooth message or opening a website, to different output nodes

- Chain networks together by attaching actions that launch other networks

SHARE networks with AirDrop, messaging & social media

- Networks (including images and actions) can be packaged together and shared over various services, like iMessage and AirDrop

EXPORT networks as Python scripts

- Convert networks to PyTorch (Python 3) scripts to run them on other platforms

- PyTorch scripts can be packaged together with a companion dataset to make training on other platforms effortless

- mobileML can be used as a prototyping tool for enterprise-grade machine learning solutions,

Page navigation:

• < Prev
• Next >