[4eyes] [sen-lab] Re: Machine learning discussion section

Fri Apr 12 09:36:56 PDT 2019

Hi everyone,

This week I might be late to this meeting, I will be more regular next week.

Thanks
Ekta

On Fri, Apr 12, 2019 at 8:23 AM Pradeep Sen <psen at ucsb.edu> wrote:

> Thanks, Peter for coordinating . Everyone, please see updated location.
>
> Best,
>
> -Pradeep
>
> ---
> Pradeep Sen
> Associate Professor
> UCSB MIRAGE Lab
> Dept. of Electrical & Computer Engineering
> University of California, Santa Barbara
> Santa Barbara, CA 93106-9560
>
> On Apr 12, 2019, at 7:57 AM, Peter Zhe Fu <peterzhefu at gmail.com> wrote:
>
> Hi all,
>
>       Just a reminder of our machine learning discussion group meeting
> today at 10 am at HFH 1152 (the smaller conference room near the elevator).
>       PS: the meeting agenda can be found in trailing emails :-D
>
> Thanks and best regards,
> Zhe Fu (Peter)
>
>
> On Wed, Apr 10, 2019 at 3:23 PM Pradeep Sen <psen at ece.ucsb.edu> wrote:
>
>> Okay, that sounds good. Please let all in 4Eyes lab know about the change.
>> We can do it there from now on.
>>
>> Best,
>>
>> -Pradeep
>>
>> ---
>> Pradeep Sen
>> Associate Professor
>> UCSB MIRAGE Lab
>> Dept. of Electrical & Computer Engineering
>> University of California, Santa Barbara
>> Santa Barbara, CA 93106-9560
>>
>>
>> On Wed, Apr 10, 2019 at 3:18 PM Peter Zhe Fu <peterzhefu at gmail.com>
>> wrote:
>>
>>> Hi Pradeep,
>>>
>>>     I have checked with Samantha from CS office and have booked HFH
>>> 1152. This is the small conference room attached to HFH 1132, but still
>>> larger than our HFH 5110 lab conference room :-D
>>>
>>> Thanks and best regards,
>>> Zhe Fu (Peter)
>>>
>>>
>>> On Wed, Apr 10, 2019 at 1:26 PM Jędrzej Kozerawski <
>>> jedrzej.kozerawski at gmail.com> wrote:
>>>
>>>> That sounds great to me. I'm also available at 10 am on Friday.
>>>>
>>>> Best,
>>>> Jedrzej
>>>>
>>>>
>>>>
>>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=icon> Virus-free.
>>>> www.avast.com
>>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=link>
>>>> <#m_-2055277601389949151_m_-5728017780921240589_m_4798832457742260780_m_-7285137942775776007_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>>>>
>>>> On Wed, Apr 10, 2019 at 1:15 PM Pradeep Sen <psen at ece.ucsb.edu> wrote:
>>>>
>>>>> Actually, let me revise that slightly.  It might be helpful for
>>>>> everyone else to hear
>>>>> a *SHORT* blurb about what your project is about at the beginning. What
>>>>> are you trying to accomplish?  What is your general approach?  What
>>>>> kind
>>>>> of results have you gotten till now?
>>>>>
>>>>> Let's keep these ULTRA short, say 2 minutes or so. I'll cut you off
>>>>> and get started
>>>>> on the main discussion because the point of the meeting is not to
>>>>> present your
>>>>> project but rather to have more discussion on it.
>>>>>
>>>>> Best,
>>>>>
>>>>> -Pradeep
>>>>>
>>>>>
>>>>> ---
>>>>> Pradeep Sen
>>>>> Associate Professor
>>>>> UCSB MIRAGE Lab
>>>>> Dept. of Electrical & Computer Engineering
>>>>> University of California, Santa Barbara
>>>>> Santa Barbara, CA 93106-9560
>>>>>
>>>>>
>>>>> On Wed, Apr 10, 2019 at 1:04 PM Pradeep Sen <psen at ucsb.edu> wrote:
>>>>>
>>>>>> Okay, then. We can talk about Peter’s project.  Guys, I’d like to do
>>>>>> it like we have our one-on-one meetings so it’s NOT a presentation and you
>>>>>> don’t have to “drive” it. Rather it is a back and forth discussion of
>>>>>> ideas, and I want to focus on the nitty gritty details that could be
>>>>>> tremendously useful for other projects.
>>>>>>
>>>>>> Peter, for your project for example I really want to talk about this
>>>>>> Perlin Noise idea we discussed, which I think is absolutely the right way
>>>>>> to think about multiscale detail. This looks like it could be very useful
>>>>>> for Ekta’s image restoration project and obviously Abhishek’s project as
>>>>>> well (although there are many other similarities with that project as
>>>>>> well).  We can also talk about some of the architectural things of your
>>>>>> project that could relate to others and then on getting the dataset which
>>>>>> again is of interest to many of you.
>>>>>>
>>>>>> Best,
>>>>>>
>>>>>> -Pradeep
>>>>>>
>>>>>> ---
>>>>>> Pradeep Sen
>>>>>> Associate Professor
>>>>>> UCSB MIRAGE Lab
>>>>>> Dept. of Electrical & Computer Engineering
>>>>>> University of California, Santa Barbara
>>>>>> Santa Barbara, CA 93106-9560
>>>>>>
>>>>>> On Apr 10, 2019, at 12:31 PM, Peter Zhe Fu <peterzhefu at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>> Hi Ekta and all,
>>>>>>       No worries Ekta, TA workload can be really a lot, especially at
>>>>>> the start of the quarter. If Ekta is busy, I am happy to volunteer to lead
>>>>>> or drive our meeting this Friday.
>>>>>>     The topic would be using data driven machine learning methods to
>>>>>> improve sparse point cloud reconstruction (from structure from motion), on
>>>>>> campus scale buildings :-D
>>>>>>
>>>>>> Regards,
>>>>>> Peter
>>>>>>
>>>>>> Ekta Prashnani <ektaprashnani at gmail.com>于2019年4月10日 周三下午12:15写道：
>>>>>>
>>>>>>> Hi everyone,
>>>>>>>
>>>>>>> It can be tough for me to regularly attend this meeting, given my
>>>>>>> meeting commitments for capstone TAship + other lab/research meetings
>>>>>>> (which can often amount to >= 5 meetings a week - as has been the case this
>>>>>>> week) and upcoming deadlines.
>>>>>>> This week has been busy for me with TAship responsibilities due to
>>>>>>> the start of spring quarter. Could someone else please volunteer to present
>>>>>>> their work?
>>>>>>> I will be sure to present my work in the near future.
>>>>>>>
>>>>>>> I request this because whenever we are discussing my research, I
>>>>>>> would like to drive that discussion and that would need some time to
>>>>>>> prepare. Sorry for the inconvenience!
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Ekta
>>>>>>>
>>>>>>> On Wed, Apr 10, 2019 at 11:33 AM Pradeep Sen <psen at ece.ucsb.edu>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Okay, let's do it then this Friday at 10am.  Ekta, can we discuss
>>>>>>>> your work
>>>>>>>> on GANs and the related ideas we have been discussing?
>>>>>>>>
>>>>>>>> I can drive the discussion, so you don't have to prepare anything
>>>>>>>> specific for
>>>>>>>> this discussion. But just be prepared to show some of your results,
>>>>>>>> discuss
>>>>>>>> your application, etc.
>>>>>>>>
>>>>>>>> Peter et al., I am going to email the full 4Eyes list to see if
>>>>>>>> there are folks
>>>>>>>> interesting in attending as well (don't want to leave anyone out).
>>>>>>>> The
>>>>>>>> conference room would be great, but I am concerned about capacity.
>>>>>>>> So let's do a headcount of who can attend.
>>>>>>>>
>>>>>>>> Best,
>>>>>>>>
>>>>>>>> -Pradeep
>>>>>>>>
>>>>>>>> ---
>>>>>>>> Pradeep Sen
>>>>>>>> Associate Professor
>>>>>>>> UCSB MIRAGE Lab
>>>>>>>> Dept. of Electrical & Computer Engineering
>>>>>>>> University of California, Santa Barbara
>>>>>>>> Santa Barbara, CA 93106-9560
>>>>>>>>
>>>>>>>>
>>>>>>>> On Wed, Apr 10, 2019 at 12:32 AM Abhishek Badki <
>>>>>>>> abhishek_badki at umail.ucsb.edu> wrote:
>>>>>>>>
>>>>>>>>> For meeting venue, 4eyes conference room is an option.
>>>>>>>>> Peter/Pushkar/Jacob: Would this be possible?
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Abhishek
>>>>>>>>>
>>>>>>>>> On Tue, Apr 9, 2019 at 11:59 PM CY Xu <hello.xu at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> This would become a great venue to ask research questions and get
>>>>>>>>>> instant feedback and discussion, especially beneficial for ML rookies like
>>>>>>>>>> myself. Thank you. 10am every Friday works for me.
>>>>>>>>>>
>>>>>>>>>> CY Xu
>>>>>>>>>> cyxu.tv
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Tue, Apr 9, 2019 at 4:07 PM Pradeep Sen <psen at ece.ucsb.edu>
>>>>>>>>>> wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi everyone,
>>>>>>>>>>>
>>>>>>>>>>> I have mentioned to a few of you that we should start a regular
>>>>>>>>>>> discussion meeting for
>>>>>>>>>>> machine learning topics and their applications (different from
>>>>>>>>>>> the
>>>>>>>>>>> reading group).
>>>>>>>>>>>
>>>>>>>>>>> Specifically, I feel that there is a lot of overlap in various
>>>>>>>>>>> projects that we are involved with,
>>>>>>>>>>> and find myself discussing the same ideas over and over again
>>>>>>>>>>> with
>>>>>>>>>>> different people, where
>>>>>>>>>>> it might be useful to bring everyone together to brainstorm
>>>>>>>>>>> ideas at
>>>>>>>>>>> the same time.  So I
>>>>>>>>>>> think we should have a (small) discussion group to talk about
>>>>>>>>>>> things like this.
>>>>>>>>>>>
>>>>>>>>>>> The group can discuss a different topic per week, mostly focused
>>>>>>>>>>> around individual
>>>>>>>>>>> student projects, but also looking at some new ideas that we
>>>>>>>>>>> want to
>>>>>>>>>>> pursue.  Picture
>>>>>>>>>>> the kind of discussion that we have in individual projects, but I
>>>>>>>>>>> would like to have that
>>>>>>>>>>> with a slightly bigger group so we can brainstorm ideas together
>>>>>>>>>>> and better see
>>>>>>>>>>> connections between the different projects. Hopefully, this will
>>>>>>>>>>> generate even better
>>>>>>>>>>> ideas and motivate various people to work together on new
>>>>>>>>>>> projects.
>>>>>>>>>>>
>>>>>>>>>>> For example, here are some of the projects I would like to see us
>>>>>>>>>>> discuss at some
>>>>>>>>>>> point:
>>>>>>>>>>>
>>>>>>>>>>> - Ekta's project on new GAN architectures for synthesis and
>>>>>>>>>>> image restoration
>>>>>>>>>>> - CY's project on using GAN architectures for video style
>>>>>>>>>>> transfer
>>>>>>>>>>> - Steve's project(s) on using machine learning for rendering. He
>>>>>>>>>>> also
>>>>>>>>>>> had another project
>>>>>>>>>>>   using GANs for super-resolution, which is actually quite
>>>>>>>>>>> closely
>>>>>>>>>>> tied to some of the
>>>>>>>>>>>   directions Ekta is exploring
>>>>>>>>>>> - Adam and Chris's early stage projects on using machine
>>>>>>>>>>> learning for
>>>>>>>>>>> denoising and other rendering applications
>>>>>>>>>>> - Abhishek's project on using machine learning for 3D geometry
>>>>>>>>>>> reconstruction with meshes
>>>>>>>>>>> - Peter's project on using machine learning for 3D geometry
>>>>>>>>>>> reconstruction with point cloud
>>>>>>>>>>> - Jedrzej's project on using random networks for interesting
>>>>>>>>>>> things
>>>>>>>>>>>
>>>>>>>>>>> These are just some of the possible projects we could discuss.
>>>>>>>>>>> As you
>>>>>>>>>>> can see there is
>>>>>>>>>>> quite a bit of overlap between them and they could benefit for
>>>>>>>>>>> some
>>>>>>>>>>> in-depth discussion as
>>>>>>>>>>> a group. Note that even though I am just calling out a few
>>>>>>>>>>> projects,
>>>>>>>>>>> there are others working
>>>>>>>>>>> on related projects as well... everyone is invited!
>>>>>>>>>>>
>>>>>>>>>>> I can help drive this thing if someone can help get a room and
>>>>>>>>>>> find a
>>>>>>>>>>> time we can
>>>>>>>>>>> all meet.  I would say something like Friday mornings, say at
>>>>>>>>>>> 10am.  Would that
>>>>>>>>>>> work for everyone?  Please let me know.  We can start this week.
>>>>>>>>>>>
>>>>>>>>>>> Best,
>>>>>>>>>>>
>>>>>>>>>>> -Pradeep
>>>>>>>>>>>
>>>>>>>>>>> ---
>>>>>>>>>>> Pradeep Sen
>>>>>>>>>>> Associate Professor
>>>>>>>>>>> UCSB MIRAGE Lab
>>>>>>>>>>> Dept. of Electrical & Computer Engineering
>>>>>>>>>>> University of California, Santa Barbara
>>>>>>>>>>> Santa Barbara, CA 93106-9560
>>>>>>>>>>>
>>>>>>>>>> --
>>>>>>>> You received this message because you are subscribed to the Google
>>>>>>>> Groups "ECE Sen Lab" group.
>>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>>> send an email to sen-lab+unsubscribe at ece.ucsb.edu.
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>>
>>>>>>> Ekta Prashnani
>>>>>>> Ph.D. Candidate
>>>>>>> Department of Electrical and Computer Engineering
>>>>>>> University of California, Santa Barbara
>>>>>>> Personal Homepage: https://prashnani.github.io/
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>> Thanks and best regards,
>>>>>> Zhe Fu (Peter)
>>>>>>
>>>>>>
>>>>
>>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=icon> Virus-free.
>>>> www.avast.com
>>>> <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail&utm_term=link>
>>>> <#m_-2055277601389949151_m_-5728017780921240589_m_4798832457742260780_m_-7285137942775776007_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>
>>>>
>>> --

Ekta Prashnani
Ph.D. Candidate
Department of Electrical and Computer Engineering
University of California, Santa Barbara
Personal Homepage: https://prashnani.github.io/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cs.ucsb.edu/pipermail/ilab-users/attachments/20190412/8c8571f8/attachment-0001.html>