ConFederate
Zero knowledge transfer federalized machine learning system for accelerating healthcare.
Created on 17th September 2022
•
ConFederate
Zero knowledge transfer federalized machine learning system for accelerating healthcare.
The problem ConFederate solves
Explores the potential value of untapped, localized datasets present with healthcare providers to make healthcare cheaper,faster and more accurate while keeping in mind the primary stakeholders and their interests. Also implements IPFS for internal nodes in individual hospital chains to ensure authenticity and data integrity.
Challenges we ran into
We ran into problems integrating our components with docker and deploying multiple docker images onto nodes. This took over 3 hours to resolve since numpy build wheel kept failing due to a dependency clash and it turned out to be a faulty docker image from the python project that was causing this setback. Also challenging was the training phase for our model to detect Pneumonia based on X-Ray images since the dataset was over 2 Gigabytes in size and our machines were underpowered for it.