<div dir="ltr">A gentle reminder that the talk will be tomorrow (Tuesday) noon in <b>NSH 1507.</b></div><div class="gmail_extra"><br><div class="gmail_quote">On Sat, Apr 14, 2018 at 3:17 AM, Adams Wei Yu <span dir="ltr"><<a href="mailto:weiyu@cs.cmu.edu" target="_blank">weiyu@cs.cmu.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)">Dear faculty and students,</div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)"><br></div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)"><span style="font-weight:400">We look forward to seeing you next Tuesday, April 17, at noon in </span><b>NSH 1507</b><b style="font-weight:400"> </b>for AI Seminar sponsored by Apple. To learn more about the seminar series, please visit the AI Seminar <a href="http://www.cs.cmu.edu/~aiseminar/" style="font-weight:400;color:rgb(17,85,204)" target="_blank">webpage</a>.</div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)"><br></div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)"><span style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-transform:none;white-space:normal;word-spacing:0px">On Tuesday,<span> </span></span><span style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-transform:none;white-space:normal;word-spacing:0px"> <a href="http://xycking.wixsite.com/yichongxu" target="_blank">Yichong Xu</a></span><span style="font-size:12.8px"><span> </span>will give the following talk: </span></div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)"><br></div><div style="text-align:start;text-indent:0px;text-decoration-style:initial;text-decoration-color:initial;background-color:rgb(255,255,255)"><div><span style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-transform:none;white-space:normal;word-spacing:0px">Title:  </span><span style="font-size:12.8px">Interactive learning using Comparison Queries</span></div><div><span style="font-size:12.8px"><br></span></div><div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-transform:none;white-space:normal;word-spacing:0px">Abstract: </div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:12.8px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-transform:none;white-space:normal;word-spacing:0px"><br></div><div><span style="text-align:start;text-indent:0px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline;font-size:12.8px"><div style="text-align:start;text-indent:0px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"><div><div style="text-align:start;text-indent:0px;background-color:rgb(255,255,255);text-decoration-style:initial;text-decoration-color:initial"><div><div>In supervised learning, we leverage a labeled dataset to design methods for function estimation. In many practical situations, we are able to obtain alternative feedback, possibly at a low cost. A broad goal is to understand the usefulness of, and to design algorithms to exploit, this alternative feedback. We consider a interactive learning setting where we obtain additional ordinal (or comparison) information for potentially unlabeled samples. In this talk we show the usefulness of such ordinal feedback for two tasks: Binary classification and nonparametric regression. For binary classification, we show that comparison queries can help in improving the label and total query complexity by reducing the learning problem to that of learning a threshold function. We present an algorithm that achieves near-optimal label and total query complexity. For nonparametric regression, we show that it is possible to accurately estimate an underlying function with a very small labeled set, effectively escaping the curse of dimensionality. We develop an algorithm called Ranking-Regression(R^2) and analyze its accuracy as a function of size of the labeled and unlabeled datasets and various noise parameters. We also derive lower bounds to show that R^2 is optimal in a variety of settings. Experiments show that our algorithms outperforms label-only algorithms when comparison information is available.</div><div><br></div><div>Based on joint works with Sivaraman Balakrishnan, Artur Dubrawski, Kyle Miller, Hariank Muthakana, Aarti Singh and Hongyang Zhang.</div></div><div><br></div><div><br></div></div></div></div></span></div></div></div></div>
</blockquote></div><br></div>