Hi! 👋 Welcome to The Big Y!
Privacy is a recurring theme for good reasons, and it is a continually important and increasingly difficult problem to solve. So much progress within machine learning applications depends on the accumulation and use of a lot of personal data, yet we still struggle to ensure the safety of our personal information. It’s hard to trust our data to large corporations that have financial incentives to exploit it, and this is amplified when it comes to personal health data. A couple of years ago, the NHS had a controversial data sharing agreement with DeepMind, so the question is how do we innovate within an area that is ripe with privacy concerns?
I’ve sung the praises of synthetic data previously, but I want to touch on another solution called federated machine learning: a privacy-preserving decentralized collaborative ML technique. Running the ML algorithms locally on edge devices avoiding transferring data off of the device can be beneficial especially when it comes to sensitive data. By eliminating privacy friction points, we are able to continue to innovate to help drive health benefits (and many other types of privacy-forward solutions). This type of ML is clearly gaining traction, and I believe it will play an upfront role as we continue to address the importance of adhering to privacy concerns in ML development.
Some Covid-19 contact tracing apps have had some issues since they have relied on centralized data collection rather than using the suggested privacy forward solutions from Apple or Google.
Thanks for reading! Share this with a friend if you think they'd like it too. Have a great week! 😁