<div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Tue, Jun 27, 2023 at 6:16 PM Artur Dubrawski <<a href="mailto:awd@cs.cmu.edu">awd@cs.cmu.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">This will be fun!<div><br></div><div>Please mark your calendars and join to cheer for Sebastian while enjoying a presentation of exceptional quality.</div><div><br></div><div>Cheers</div><div>Artur<br><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">---------- Forwarded message ---------<br>From: <strong class="gmail_sendername" dir="auto">Diane Stidle</strong> <span dir="auto"><<a href="mailto:stidle@andrew.cmu.edu" target="_blank">stidle@andrew.cmu.edu</a>></span><br>Date: Tue, Jun 27, 2023 at 4:24 PM<br>Subject: Thesis Defense - July 12, 2023 - Sebastian Caldas - Collaborative learning by leveraging siloed data<br>To: <a href="mailto:ml-seminar@cs.cmu.edu" target="_blank">ml-seminar@cs.cmu.edu</a> <<a href="mailto:ML-SEMINAR@cs.cmu.edu" target="_blank">ML-SEMINAR@cs.cmu.edu</a>>,  <<a href="mailto:cler@pitt.edu" target="_blank">cler@pitt.edu</a>>,  <<a href="mailto:martin.jaggi@epfl.ch" target="_blank">martin.jaggi@epfl.ch</a>><br></div><br><br>
  

    
  
  <div>
    <p><i><b>Thesis Defense</b></i></p>
    <p>Date: July 12, 2023<br>
      Time: 4:00pm (EDT)<br>
      Place: GHC 8102 & Remote<br>
      PhD Candidate: Sebastian Caldas</p>
    <p><b>Title: </b><b>Collaborative learning by leveraging siloed
        data</b></p>
    <div>Abstract:<br>
      <div>Regulations can often limit stakeholders’ modeling
        capabilities by preventing data sharing. For example, in order
        to protect patient privacy, clinical centers may be unable to
        share their data and thus lack representative records to learn
        about a rare condition. To address this challenge, previous work
        in machine learning has shown that these stakeholders benefit
        from training models in a collaborative fashion, improving their
        predictive performance. However, as we start training these
        collaborative models in real-world settings, and in order to be
        truly useful, they need to provide utility along dimensions
        beyond predictive performance. In this thesis, we propose
        methods and algorithms to improve collaborative models that
        leverage siloed data along three dimensions. In the first part,
        we propose methods to reduce the communication footprint of
        models learned by mobile devices cooperating over edge networks,
        allowing for higher capacity models to be trained. Then, in the
        second part, we introduce an algorithm that provides
        explanations about predictions of models trained across clinical
        centers, thus improving their clinical utility. Finally, in the
        third part, we address the need to encode expert supervision
        into collaborative models trained using on-device data,
        increasing the class of problems we can tackle in these
        scenarios.</div>
      <div><b><br>
        </b></div>
      <div><b>Thesis Committee:</b> <br>
        Artur Dubrawski (Chair)<br>
        Virginia Smith<br>
        Gilles Clermont (University of Pittsburgh) <a href="mailto:cler@pitt.edu" target="_blank">cler@pitt.edu</a><br>
        Martin Jaggi (EPFL) <a href="mailto:martin.jaggi@epfl.ch" target="_blank">martin.jaggi@epfl.ch</a></div>
      <div><br>
      </div>
      <b>Link to the draft document: </b><a href="https://drive.google.com/file/d/1xTPUERARdnQnMSNCnu2VXrWk1YFKQOqp/view?usp=sharing" target="_blank">https://drive.google.com/file/d/1xTPUERARdnQnMSNCnu2VXrWk1YFKQOqp/view?usp=sharing</a> 
      <div><br>
      </div>
      <div><b>Zoom meeting link:</b> <a href="https://www.google.com/url?q=https://cmu.zoom.us/j/92902855666?pwd%3Da3l4V3ppQlp1K252K2doSjNTRlhPdz09&sa=D&source=calendar&ust=1688311717182551&usg=AOvVaw0fNQndryXEeSLoOO-ZhNfp" style="font-family:Roboto,Arial,sans-serif;font-size:14px;letter-spacing:0.2px" target="_blank">https://cmu.zoom.us/j/92902855666?pwd=a3l4V3ppQlp1K252K2doSjNTRlhPdz09</a></div>
    </div>
    <pre cols="72">-- 
Diane Stidle
PhD Program Manager
Machine Learning Department
Carnegie Mellon University
<a href="mailto:stidle@andrew.cmu.edu" target="_blank">stidle@andrew.cmu.edu</a></pre>
  </div>

</div></div></div>
</blockquote></div></div>