[Lisa_teatalk] [Lisa_labo] ICML Practice talk: Thursday June 2nd by Kelvin Xu

Kelvin Xu iamkelvinxu at gmail.com
Thu Jul 2 16:41:29 EDT 2015


Thank you again to everyone who came to the practice talk. It was very
useful for me and I really appreciated the number of thoughtful comments.

Best,
-- Kelvin

On Thu, Jul 2, 2015 at 1:28 PM, Laurent Dinh <dinh.laurent at gmail.com> wrote:

> This is the reminder for the talk today.
>
> Cheers,
> --
> Laurent
>
>
> Le mardi 30 juin 2015, Yoshua Bengio <yoshua.bengio at gmail.com> a écrit :
>
>> Hi all,
>>
>> For those who are not acquainted with this tradition, we often do these
>> practice talks, which give an opportunity to
>> a presenter to practice in front of a friendly audience, and *more
>> importantly* get critical and constructive feedback
>> in order to improve the presentation. For a 20 min talk we often spend a
>> whole hour discussing ways to improve
>> things. Such talks, especially at major conferences like ICML, are
>> important for the visibility of the lab to the outside world!
>>
>> Your participation in this event is thus greatly appreciated!
>>
>> See you on Thursday, 3:30pm.
>>
>> -- Yoshua
>>
>>
>> On Tue, Jun 30, 2015 at 1:03 PM, Jörg Bornschein <bornj at iro.umontreal.ca>
>> wrote:
>>
>>> Hi everyone,
>>>
>>>
>>> this Thursday Kelvin will present
>>>
>>> Show, Attend and Tell: Neural Image Caption Generation with Visual
>>> Attention
>>>
>>> which was accepted for this years ICML. Looking forward to see you there
>>> and to get your feedback.
>>>
>>> Location: AA-3195
>>> Time: Thursday, July 2nd, 3:30pm
>>>
>>> Abstract
>>>
>>> Inspired by recent work in machine translation and object detection, we
>>> introduce an attention based model that automatically learns to describe
>>> the content of images. We describe how we can train this model in a
>>> deterministic manner using standard backpropagation techniques and
>>> stochastically by maximizing a variational lower bound. We also show
>>> through visualization how the model is able to automatically learn to fix
>>> its gaze on salient objects while generating the corresponding words in the
>>> output sequence. We validate the use of attention with state-of-the-art
>>> performance on three benchmark datasets: Flickr8k, Flickr30k and MS COCO.
>>>
>>>
>>>
>>>    Jorg
>>>
>>>
>>>
>>> _______________________________________________
>>> Lisa_labo mailing list
>>> Lisa_labo at iro.umontreal.ca
>>> https://webmail.iro.umontreal.ca/mailman/listinfo/lisa_labo
>>>
>>>
>>
> _______________________________________________
> Lisa_labo mailing list
> Lisa_labo at iro.umontreal.ca
> https://webmail.iro.umontreal.ca/mailman/listinfo/lisa_labo
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://webmail.iro.umontreal.ca/pipermail/lisa_teatalk/attachments/20150702/5db6f787/attachment.html 


More information about the Lisa_teatalk mailing list