[bioundgrd] FW: Jelani Nelson - Sketching Big Data - Wed., October 25th at 7pm, Science Research Public Lecture - Harvard Science Center, Hall C, One Oxford St.

Joshua Stone stonej at mit.edu
Thu Oct 12 12:39:17 EDT 2017


From: Science Research Public Lecture Series <science_lectures=fas.harvard.edu at mail28.atl111.rsgsv.net> on behalf of Science Research Public Lecture Series <science_lectures at fas.harvard.edu>
Reply-To: Science Research Public Lecture Series <science_lectures at fas.harvard.edu>
Date: Thursday, October 12, 2017 at 12:34 PM
Subject: Jelani Nelson - Sketching Big Data - Wed., October 25th at 7pm, Science Research Public Lecture - Harvard Science Center, Hall C, One Oxford St.





Science Research Public Lecture

Wednesday, October 25 @ 7:00pm


Harvard University, Science Center Hall C
One Oxford Street, Cambridge, MA




[https://gallery.mailchimp.com/2a42f9e6c5264bb277175db04/images/acea3471-c199-47d0-a7cc-9086c05c5259.jpg]






Sketching Big Data

Jelani Nelson
Associate Professor of Computer Science
John A. Paulson School of Engineering and Applied Science
Harvard University




A "sketch" is a data structure supporting some pre-specified set of queries and updates to a database while consuming space substantially (often exponentially) less than the information theoretic minimum required to store everything seen, and thus can also be seen as some form of functional compression. The advantages of sketching include less memory consumption, faster algorithms, and reduced bandwidth requirements in distributed computing environments.

This talk will touch on some of the magic made possible by sketching techniques, such as:

* (Approximately) counting up to an integer N in exponentially less memory than what's required to actually write the digits of N down.

* (Approximately) computing the number of distinct words ever appearing in any of Shakespeare's works, via a method that reads through them all once while only remembering 3 lines' worth of text in memory at any given time.

* Detecting trending keywords queried to a search engine, such as 'bigly', or 'Irma', while never remembering more than a negligible fraction of the query stream seen thus far.


































This email was sent to undergradbio at mit.edu<mailto:undergradbio at mit.edu>
why did I get this?<https://harvard.us9.list-manage.com/about?u=2a42f9e6c5264bb277175db04&id=5025e76076&e=2b613e0cdc&c=b60406381b>    unsubscribe from this list<https://harvard.us9.list-manage.com/unsubscribe?u=2a42f9e6c5264bb277175db04&id=5025e76076&e=2b613e0cdc&c=b60406381b>    update subscription preferences<https://harvard.us9.list-manage.com/profile?u=2a42f9e6c5264bb277175db04&id=5025e76076&e=2b613e0cdc>
Harvard University/Department of Physics · 17 Oxford Street · Cambridge, MA 02138 · USA

[mail Marketing Powered by MailChimp]<http://www.mailchimp.com/monkey-rewards/?utm_source=freemium_newsletter&utm_medium=email&utm_campaign=monkey_rewards&aid=2a42f9e6c5264bb277175db04&afl=1>



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/bioundgrd/attachments/20171012/9cd4f6e6/attachment-0001.html


More information about the bioundgrd mailing list