CK-based Android applications to crowdsource AI/SW/HW co-design experiments

You can download and install two CK-based applications for your Android device to participate in collaborative benchmarking, optimization, autotuning and co-design of efficient model/SW/HW stacks for emerging workloads. You can view results from all crowdsourced experiments (quantum, ReQuEST, etc) here (see previous version of a public repository). We use this open-source technology in open and reproducible ACM ReQuEST tournaments to co-design Pareto-efficient deep learning stacks (speed, accuracy, costs) and in a joint project with Raspberry Pi foundation to crowdsource compiler autotuning and machine learning (see interactive article).

To our surprise, our open science applications were removed from Google Play in August 2018 after 3 years of successful participation by thousands of users because it apparently violates recent Google's terms of use: it's not allowed to download extra code or data from external sources. However, the whole point of our applications and our novel crowd-tuning concept is to download slightly differently modified code and models to different users to collaboratively find the most efficient solutions across diverse platforms and environments provided by volunteers similar to SETI@home in terms of speed, accuracy, costs and other metrics or automatically detect software and hardware bugs (see paper 1 and paper 2).

While we are discussing with Google how to move our applications back to Google Play Store, you can download and install them here. Just make sure that your Android device is configured to allow installation from sources other than the Play Store before downloading and installing our CK-based Android apps. You can check it as follows:

You can now download, install and run two CK-based Android applications to

Please feel free to get in touch with the CK community to provide your feedback or discuss results!