Friction in data sharing is a large challenge for large scale machine learning. Recently techniques such as Federated Learning, Differential Privacy and Split Learning aim to address siloed and unstructured data, privacy and regulation of data sharing and incentive models for data transparent ecosystems. Split learning is a new technique developed at the MIT Media Lab’s Camera Culture group that allows for participating entities to train machine learning models without sharing any raw data.
Implemented works:
- Simulation of split learning on same colab/PC as vanilla splitNN code.
- Written a threaded version of code, running server and client threads to train the splitted model.
- Developed a code using mpi4py to run the code on different hardware connected to each other on a network in a Distributed fashion.
- Written scripts using zeroMQ to run as client and server to fulfill the training. Ran the code successfully on GPC as server and laptop as a client. Check different folders for different types of execution.