Version 1.0.0 and source code are here!

I'm proud to announce that the version 1.0.0 of the application is here. With this release, the source code is published, too.

• The app now features a class-balanced replay mechanism that will prevent forgetting on old classes
• Many unused resources have been eliminated
• The main readme now features a minimal guide to the app
• The app has an icon now :)

We also released the paper "Continual Learning at the Edge: Real-Time Training on Smartphone Devices" which has been accepted at ESANN 2021. The paper explains how the app works in more detail. More info can be found in the main README.

Version v0.2.0-alpha is now available

I'm proud to announce that the v0.2.0 version is now available. It still in its alpha stage, but it should work on most devices. This new version features:

• Enhanced training speed. It should now complete in less than a second on mid-tier devices!
• Release configuration: even more speed!
• Reduced APK size

This is a self-signed release app and it will work on Android arm-v8 devices only. You should be able to download and install it using your browser.

A security prompt may warn you that the app was signed from an unknown source. Don't worry, this is the expected behavior when installing applications from the APK file directly!

Source code available on paper acceptance!

Tested on OnePlus 6 (OxygenOS 10.3.1).

The CL CORe App is now available!

I'm proud to announce that the first alpha version of the CL CORe App is now available.

This is a unsigned debug app and it will work on Android arm-v8 devices only. You should be able to download and install it using your browser.

Source code available soon!