Hello, Folks! Thank you all for joining us in kicking off MSAIL 2017 with a great talkand discussion of Stein's Paradox. For some reason, the presenter wishesto remain anonymous, but let's nevertheless clap again for him or her! Her or his wish got me thinking: how can we respect his or her privacydespite referencing and congratulating her or him? I can't say too much,lest y'all learn his or her identity! On a completely different note: have you ever wondered why this ArtificialIntelligence club does so much Machine Learning? Surely, equally importantto learning is ignorance. This week, let's discuss Machine Ignorance:techniques to obstruct an algorithm's learning. Specifically, we wonder: How can we prevent over-fitting on our test set despite repeated testing?We can't test too much, lest our algorithms learn the test's answers! We'll meet:     3433 EECS, Wednesday, 2016-01-25, 19:00-20:00.Afterward, we'll have the room reserved an additional half hour forinformal discussion and socializing. We look forward to seeing youthere! For a sneak-peak of the ideas we'll discuss, check out the following!Generalization in Adaptive . . . and Holdout Reuse (Dwork et al. 2015)
Differential Privacy: a Short Tutorial (Yu-Xiang Wang 2012)
  Cheers,
Mas Ioka