All Segments
Michael Kearns: Differential Privacy
5 segments available
Full episode with Michael Kearns (Nov 2019): https://www.youtube.com/watch?v=AzdxbzHtjgs New clips channel (Lex Clips): https://www.youtube.com/lexclips Once it reaches 20,000 subscribers, I'll start posting the clips there instead. (more links below) For now, new full episodes are released once or twice a week and 1-2 new clips or a new non-podcast video is released on all other days. Clip from full episode: https://www.youtube.com/watch?v=AzdxbzHtjgs If you enjoy these clips, subscribe to the new clips channel (Lex Clips): https://www.youtube.com/lexclips Once it reaches 20,000 subscribers, I'll start posting the clips there instead. For now, new full episodes are released once or twice a week and 1-2 new clips or a new non-podcast video is released on all other days. (more links below) Podcast full episodes playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 Podcasts clips playlist: https://www.youtube.com/playlist?list=PLrAXtmErZgOeciFP3CBCIEElOJeitOr41 Podcast website: https://lexfridman.com/ai Podcast on Apple Podcasts (iTunes): https://apple.co/2lwqZIr Podcast on Spotify: https://spoti.fi/2nEwCF8 Podcast RSS: https://lexfridman.com/category/ai/feed/ Michael Kearns is a professor at University of Pennsylvania and a co-author of the new book Ethical Algorithm that is the focus of much of our conversation, including algorithmic fairness, bias, privacy, and ethics in general. But, that is just one of many fields that Michael is a world-class researcher in, some of which we touch on quickly including learning theory or theoretical foundations of machine learning, game theory, algorithmic trading, quantitative finance, computational social science, and more. Subscribe to this YouTube channel or connect on: - Twitter: https://twitter.com/lexfridman - LinkedIn: https://www.linkedin.com/in/lexfridman - Facebook: https://www.facebook.com/lexfridman - Instagram: https://www.instagram.com/lexfridman - Medium: https://medium.com/@lexfridman - Support on Patreon: https://www.patreon.com/lexfridman
Segments Timeline
Understanding Differential Privacy
Michael Kearns introduces the concept of differential privacy, explaining its significance as a stronger alternative to traditional anonymization methods. He discusses how differential privacy allows researchers to analyze sensitive data, like medical records, while ensuring that the inclusion of an individual's data does not significantly alter the outcome of the analysis.
"so is there hope for any kind of privacy in a world where a few likes can can identify you so there is differential privacy right what is differential differential privacy basically is a kind of alter..."
The Counterfactual Approach
Kearns elaborates on the counterfactual framework used in differential privacy. He describes how researchers compare two scenarios: one where an individual's data is included in a dataset and one where it is not. This comparison helps to assess the potential harms that could arise from data analysis, emphasizing that differential privacy aims to ensure that the risks remain consistent regardless of data inclusion.
"counterfactual question we basically compare two alternatives one is when I do this I build this model on the database of medical records including your medical record and the other one is where I do ..."
Real-World Implications of Data Inclusion
In this segment, Kearns provides a concrete example of how data inclusion can lead to real-world consequences, such as increased insurance premiums due to the identification of health risks. He discusses the historical context of smoking and lung cancer research, illustrating how differential privacy can mitigate the risks associated with data analysis while acknowledging that some harm may still occur.
"even if your data wasn't included and to give a very concrete example right you know you know like we discussed at some length the the study that you know the in the 50s that was done that created the..."
Mechanisms of Differential Privacy
Kearns explains the mechanisms behind differential privacy, focusing on the addition of noise to computations. He describes how probabilistic algorithms can produce varying outputs by incorporating noise, which helps to protect individual data points while still allowing for meaningful analysis. This segment highlights the technical foundation of differential privacy and its application in statistical computations.
"very little harm is done great but how what is the mechanism of differential privacy so that's the kind of beautiful statement of it well what's the mechanism by which privacy's preserve yeah so it's ..."
The Future of Privacy in Data Science
In this concluding segment, Kearns discusses the maturity of the differential privacy field and its implications for machine learning and statistical analysis. He emphasizes that many algorithms can be adapted to ensure privacy without sacrificing their utility. This segment reassures that the data science community can continue to leverage data while providing robust privacy guarantees to individuals.
"average so noise noise is the Savior how many algorithms can be aided by miam by adding noise yeah so I'm a relatively recent member of the differential privacy community my co-author Aaron Roth is yo..."