<html>
  <head>

    <meta http-equiv="content-type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p><br>
    </p>
    <p>Just In case you managed not to know of this course up till
      now...<br>
    </p>
    <div class="moz-forward-container"><br>
      <div class="moz-forward-container"> <br>
        -------- Forwarded Message --------
        <table class="moz-email-headers-table" cellspacing="0"
          cellpadding="0" border="0">
          <tbody>
            <tr>
              <th nowrap="nowrap" valign="BASELINE" align="RIGHT">Subject:
              </th>
              <td>BU course on adaptive data analysis</td>
            </tr>
            <tr>
              <th nowrap="nowrap" valign="BASELINE" align="RIGHT">Date:
              </th>
              <td>Mon, 4 Sep 2017 23:06:09 -0400</td>
            </tr>
            <tr>
              <th nowrap="nowrap" valign="BASELINE" align="RIGHT">From:
              </th>
              <td>Adam Smith <a class="moz-txt-link-rfc2396E"
                  href="mailto:ads22@bu.edu" moz-do-not-send="true">&lt;ads22@bu.edu&gt;</a></td>
            </tr>
            <tr>
              <th nowrap="nowrap" valign="BASELINE" align="RIGHT">To: </th>
              <td>Dwork, Cynthia <a class="moz-txt-link-rfc2396E"
                  href="mailto:dwork@seas.harvard.edu"
                  moz-do-not-send="true">&lt;dwork@seas.harvard.edu&gt;</a>,
                Salil Vadhan <a class="moz-txt-link-rfc2396E"
                  href="mailto:salil-privacytools@g.harvard.edu"
                  moz-do-not-send="true">&lt;salil-privacytools@g.harvard.edu&gt;</a>,
                Jon Ullman <a class="moz-txt-link-rfc2396E"
                  href="mailto:jullman@gmail.com" moz-do-not-send="true">&lt;jullman@gmail.com&gt;</a>,
                Leonid Reyzin <a class="moz-txt-link-rfc2396E"
                  href="mailto:reyzin@cs.bu.edu" moz-do-not-send="true">&lt;reyzin@cs.bu.edu&gt;</a>,
                Ran Canetti <a class="moz-txt-link-rfc2396E"
                  href="mailto:canetti@bu.edu" moz-do-not-send="true">&lt;canetti@bu.edu&gt;</a>,
                Sharon Goldberg <a class="moz-txt-link-rfc2396E"
                  href="mailto:goldbe@cs.bu.edu" moz-do-not-send="true">&lt;goldbe@cs.bu.edu&gt;</a>,
                Vinod Vaikuntanathan <a class="moz-txt-link-rfc2396E"
                  href="mailto:vinodv@csail.mit.edu"
                  moz-do-not-send="true">&lt;vinodv@csail.mit.edu&gt;</a>,
                Constantinos Daskalakis <a
                  class="moz-txt-link-rfc2396E"
                  href="mailto:costis@csail.mit.edu"
                  moz-do-not-send="true">&lt;costis@csail.mit.edu&gt;</a>,
                Kolaczyk, Eric D <a class="moz-txt-link-rfc2396E"
                  href="mailto:kolaczyk@bu.edu" moz-do-not-send="true">&lt;kolaczyk@bu.edu&gt;</a>,
                Nazer, Bobak <a class="moz-txt-link-rfc2396E"
                  href="mailto:bobak@bu.edu" moz-do-not-send="true">&lt;bobak@bu.edu&gt;</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:pi@bu.edu" moz-do-not-send="true">pi@bu.edu</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:cuhler@mit.edu" moz-do-not-send="true">cuhler@mit.edu</a>,
                Madhu Sudan <a class="moz-txt-link-rfc2396E"
                  href="mailto:madhusudan@g.harvard.edu"
                  moz-do-not-send="true">&lt;madhusudan@g.harvard.edu&gt;</a>,
                Shelat, Abhi <a class="moz-txt-link-rfc2396E"
                  href="mailto:a.shelat@northeastern.edu"
                  moz-do-not-send="true">&lt;a.shelat@northeastern.edu&gt;</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:orecchia@bu.edu" moz-do-not-send="true">orecchia@bu.edu</a>,
                Raskhodnikova, Sofya <a class="moz-txt-link-rfc2396E"
                  href="mailto:sofya@bu.edu" moz-do-not-send="true">&lt;sofya@bu.edu&gt;</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:aene@bu.edu" moz-do-not-send="true">aene@bu.edu</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:b@boazbarak.org" moz-do-not-send="true">b@boazbarak.org</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:rakhlin@wharton.upenn.edu"
                  moz-do-not-send="true">rakhlin@wharton.upenn.edu</a>,
                <a class="moz-txt-link-abbreviated"
                  href="mailto:rigollet@math.mit.edu"
                  moz-do-not-send="true">rigollet@math.mit.edu</a></td>
            </tr>
          </tbody>
        </table>
        <br>
        <br>
        <pre>Dear colleagues,

I will be teaching a graduate course at BU this semester on adaptive
data analysis and associated topics, including differential privacy.
Aaron Roth (at Penn) and I are developing the materials together.

Please circulate to interested students or relevant lists! Auditors
from other universities are welcome.

Best regards,
Adam

===============


CS 591: Adaptive Data Analysis, Algorithmic Stability and Privacy

Times: Mondays and Wednesdays, 4:30-5:45pm
           (Lectures start this Wednesday, September 6.)

Location: MCS Building B25, Boston University
            (111 Cummington Mall, Boston)

Description: Adaptive data analysis is the increasingly common
practice by which insights gathered from data are used to inform
further analysis of the same data sets. This is common practice both
in machine learning, and in scientific research, in which data-sets
are shared and re-used across multiple studies. Standard theory
assumes that the analysis to be performed on a data set is selected
independently of the data set. Unfortunately, when the set of analyses
run is itself a function of the data, much of this theory becomes
invalid. The resulting disconnect has been blamed as one of the causes
of the crisis of reproducibility in empirical science.

This course will look at recently developed approaches to this
problem. We will see approaches stemming from the literature on
"differential privacy", approaches rooted in measuring leaked
information, and approaches coming from more standard statistical
tools.

The course is aimed at graduate students in computer science,
statistics and electrical engineering. The prerequisites are a solid
background in probability, and general "mathematical maturity"
(comfort reading and writing definitions, theorems and proofs). The
course will involve reading and reviewing research papers, pencil and
paper assignments, and some programming problems.

Aaron Roth at U. Penn and I are developing the materials (lecture
notes, homework) together, but each delivering lectures locally.
Materials will be posted here:
  <a class="moz-txt-link-freetext" href="https://adaptivedataanalysis.com/" moz-do-not-send="true">https://adaptivedataanalysis.com/</a>
</pre>
      </div>
    </div>
  </body>
</html>