<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body dir="auto"><div dir="ltr"><b style="font-family: Calibri, sans-serif; font-size: 10pt; -webkit-text-size-adjust: auto;"><span style="font-size: 12pt; font-family: Helvetica;"> Samuel Hopkins, PhD</span></b><meta http-equiv="content-type" content="text/html; charset=utf-8"><div dir="ltr"></div></div><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><b><span style="font-size: 11pt; font-family: Helvetica;">Massachusetts Institute of Technology</span></b><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><span style="font-size: 11pt;"> </span><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"> <o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><b><span style="font-size: 11pt; font-family: Helvetica;"><img width="230" height="230" id="Picture_x0020_2" src="cid:74628B7B-5DE2-487F-BC25-02EB976B80B7" alt="image001.png" _mf_state="1" title="null" style="width: 2.3958in; height: 2.3958in;"> </span></b><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"> <o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><b><span style="font-size: 11pt;"> </span></b><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><b><span style="font-size: 11pt;"><span dir="ltr">Tuesday, </span></span></b><b><span style="font-size: 11pt;"><span dir="ltr">May 16</span><span dir="ltr">, 202</span><span dir="ltr">3</span><span dir="ltr"> at 3:30pm</span></span></b><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><b><span style="font-size: 11pt; background-color: yellow;">Kent Chemical Laboratory, Room 120 (New Room)</span></b><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><span style="font-size: 11pt;"> </span><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><span style="font-size: 11pt;"> </span><o:p></o:p></p><p style="-webkit-text-size-adjust: auto; margin: 0in;"><b><i><span style="font-size: 12pt;">Title:</span></i></b><span style="font-size: 12pt;"> <span style="color: rgb(33, 33, 33);">Robustness Implies Privacy in Statistical Estimation</span></span><o:p></o:p></p><p class="MsoNormal" style="-webkit-text-size-adjust: auto; margin: 0in; font-size: 10pt; font-family: Calibri, sans-serif;"><span style="font-size: 11pt;"> </span><o:p></o:p></p><p style="-webkit-text-size-adjust: auto; margin: 0in;"><b><i><span style="font-size: 12pt;">Abstract:</span></i></b><b><span style="font-size: 12pt;"> </span></b><span style="font-size: 12pt;"> </span> <span style="font-size: 12pt;">We study the relationship between adversarial robustness and differential privacy in high-dimensional algorithmic statistics. We give the first black-box reduction from privacy to robustness which can produce private estimators with optimal tradeoffs among sample complexity, accuracy, and privacy for a wide range of fundamental high-dimensional parameter estimation problems, including mean and covariance estimation. We show that this reduction can be implemented in polynomial time in some important special cases. In particular, using nearly-optimal polynomial-time robust estimators for the mean and covariance of high-dimensional Gaussians which are based on the Sum-of-Squares method, we design the first polynomial-time private estimators for these problems with nearly-optimal samples-accuracy-privacy tradeoffs. Our algorithms are also robust to a constant fraction of adversarially-corrupted samples. Joint work with Gautam Kamath, Mahbod Majid, and Shyam Narayanan, to appear in STOC 2023.</span></p></body></html>