[Lisa_seminaires] [TeaTalk] Danny Tarlow, June 9, 13:45, AA5340

Dzmitry Bahdanau dimabgv at gmail.com
Ven 9 Juin 15:15:50 EDT 2017


If you want to book a 1-on-1 slot with Danny this afternoon, please put
your name on the spreadsheet (2 slots left):

https://docs.google.com/spreadsheets/d/1kIXzQVdlv2_yPspJo6U9P_25TJgSvxs3LnuixKLLds4/edit?usp=sharing

Dima

On Fri, 9 Jun 2017 at 14:53 Dzmitry Bahdanau <dimabgv at gmail.com> wrote:

> Hi all,
>
> Thanks for coming everyone, that was the record number of people! If you
> want to talk to Danny, he will be around at 6th floor for some time.
>
> Dima
>
>
> On Fri, 9 Jun 2017 at 08:46 Dzmitry Bahdanau <dimabgv at gmail.com> wrote:
>
>> Hi all,
>>
>> Just a kind reminder about the talk :)
>>
>> Dima
>>
>> On Tue, 6 Jun 2017 at 15:22 Dzmitry Bahdanau <dimabgv at gmail.com> wrote:
>>
>>> Hi all,
>>>
>>> You have just received an email about the tea-talk on June 13, but I
>>> have good news for you: we will have one more before that!
>>>
>>> Our next (really next) speaker is *Danny Tarlow*, who is a Research
>>> Scientist at Google Brain Montreal. He will present on *June 9, 13:45,
>>> at AA5340* (our usual tea-talk slot). Please find detailed information
>>> below.
>>>
>>> For those who are confused, the coming tea-talks will be given on June
>>> 9, 13 and 15 (yet to be announced).
>>>
>>> *Title:* Learning to Code: Machine Learning for Program Induction
>>>
>>> *Abstract:* I'll present two of our recent works on using machine
>>> learning ideas to induce computer programs from input-output examples. The
>>> first system is TerpreT, which casts program synthesis as a continuous
>>> optimization problem on which we perform gradient descent. It enables
>>> comparison of gradient-based program synthesis techniques to discrete
>>> search techniques that are popular in the programming languages community.
>>> Based on our learnings from TerpreT, we develop the second system,
>>> DeepCoder, which induces programs from input-output examples using a neural
>>> network to guide discrete search techniques. DeepCoder achieves an order of
>>> magnitude speedup over optimized search techniques, and it can solve
>>> problems of difficulty comparable to the very simplest problems on
>>> programming competition websites.
>>>
>>> *Bio:* Danny Tarlow is a Research Scientist at Google Brain Montreal.
>>> His main research interests are in the application of machine learning to
>>> problems involving structured data, with a specific interest in the
>>> intersection of machine learning and programming languages. He is a
>>> co-editor of the recent MIT Press book on Perturbations, Optimization, and
>>> Statistics (2017). His work has won awards at UAI (Best Student Paper,
>>> Runner Up), the ICML Workshop on Constructive Machine Learning (Best
>>> Paper), the NIPS Workshop on Neural Abstract Machines and Program Induction
>>> (Best Paper), and NIPS (Best Paper). He holds a Ph.D. from the Machine
>>> Learning group at the University of Toronto (2013) and previously was a
>>> Researcher at Microsoft Research Cambridge UK (2013 - 2017) with a Research
>>> Fellowship at Darwin College, University of Cambridge (2013 - 2016).
>>>
>>> Dima
>>>
>>
-------------- section suivante --------------
Une pièce jointe HTML a été nettoyée...
URL: http://webmail.iro.umontreal.ca/pipermail/lisa_seminaires/attachments/20170609/f54af6ac/attachment-0001.html 


Plus d'informations sur la liste de diffusion Lisa_seminaires