Privacy-Friendly Setup?

Hi everyone,

I think that Colab could monitor my workflow. Indeed, the answer of Ali Zaidi about the fingerprinting in Paperspace didn’t reassure me. I understand that if it was just about tweaking some stuffs, it’s “okay” to use Colab, but I want to learn, experiment and create as much as I can freely!

Obviously, I need a setup to learn. So, which of these are privacy-friendly?

Thanks!

Can someone help me please?

Buy the hardware to run analysis locally.

2 Likes

And what about the setups that I proposed ? Are there any that are privacy-friendly ?

Hi @jjohnson :wave: , maybe you can visit the landing page for the websites mentioned after installing https://tosdr.org/
Also, having startpage.com as an extension in your browser then looking at the privacy score That way you’d have two sources of information. Then you can come to a decision about which service is privacy friendly as opposed to me pointing out which is the best.

Some practices that can help to some extent is removing dataset headers you are working on and adjusting personally identifiable information through K-anonymization and I-diversity. Credit: chapter 5 Data Privacy: A runbook for engineers by Nishant Bhajaria.

1 Like

I have been actively learning this area. I have come across differential privacy and how they do it in Apple where you can add noise to your data but continue to use your normal workflows. Lastly, Principal component analysis could be used as a preprocessing step could also help as well. @jjohnson