mobileML para Android

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 node's value, edge's value, and layer's 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.