[Theory] NOW: 5/16 TTIC Colloquium: Elizaveta Rebrova, Princeton University

[Mary Marre]

*When:*        Monday, May 16h at* 11:30 am CT*


*Where:       *Talk will be given *live, in-person* at

                   TTIC, 6045 S. Kenwood Avenue

                   5th Floor, Room 530


*Where:*       Zoom Virtual Talk (*register in advance here
<https://uchicagogroup.zoom.us/webinar/register/WN_QFCPyH-nSv6ovsQ_tgTwag>*)


*Who: *         Elizaveta Rebrova, Princeton University


*Title:  * Modewise Methods for Tensor Compression and Recovery
*Abstract: *Tensor methods are unsurprisingly in demand since lots of
natural data is inherently high-order (e.g., temporal or multi-factor) and
high-dimensional. It is typical to avoid developing new specialized
techniques that work directly on tensors, but rather apply existent
algorithms on the data reshaped into matrices or vectors. However, such
structural simplification of the data can lead to loss of interpretability
as well as to immense comparative memory loss. Modewise tensor maps are
linear operators acting on each tensor dimension (mode) separately rather
than on the vectorizations of tensors. They are efficient: they are faster
and require significantly less memory than the measurements working on
vectorized tensors. They are also more sophisticated and present additional
mathematical challenges due to their inherent Kronecker structure. In this
talk, I will discuss data-oblivious tensor dimension reduction with
these modewise maps and its applications to the tensor low-rank fitting and
recovery. Based on the joint work with Mark Iwen, Deanna Needell, Mike
Perlmutter, and Ali Zare.

*Bio:* Liza Rebrova is an Assistant Professor at Princeton University, ORFE
Department (Operations Research and Financial Engineering). She received
her Ph.D. degree in Mathematics (Non-asymptotic Random Matrix Theory) from
the University of Michigan in 2018. Then, she worked as an Assistant
Adjunct Professor in the Department of Mathematics at UCLA (2018-2020) and
as a Postdoctoral Scholar at the Computational Research Division at
Lawrence Berkeley National Laboratory (in 2020). Her research interests are
in high-dimensional probability, and include developing randomized and
high-dimensional methods for numerical linear algebra, compressed sensing,
and the math of data.

*Host:* *Yury Makarychev* <yury at ttic.edu>
***********************************************************************************

For more information on the colloquium series or to subscribe to the
mailing list, please see http://www.ttic.edu/colloquium.php


Mary C. Marre
Faculty Administrative Support
*Toyota Technological Institute*
*6045 S. Kenwood Avenue*
*Chicago, IL  60637*
*mmarre at ttic.edu <mmarre at ttic.edu>*


On Mon, May 16, 2022 at 10:30 AM Mary Marre <mmarre at ttic.edu> wrote:

> *When:*        Monday, May 16h at* 11:30 am CT*
>
>
> *Where:       *Talk will be given *live, in-person* at
>
>                    TTIC, 6045 S. Kenwood Avenue
>
>                    5th Floor, Room 530
>
>
> *Where:*       Zoom Virtual Talk (*register in advance here
> <https://uchicagogroup.zoom.us/webinar/register/WN_QFCPyH-nSv6ovsQ_tgTwag>*
> )
>
>
> *Who: *         Elizaveta Rebrova, Princeton University
>
>
> *Title:  * Modewise Methods for Tensor Compression and Recovery
> *Abstract: *Tensor methods are unsurprisingly in demand since lots of
> natural data is inherently high-order (e.g., temporal or multi-factor) and
> high-dimensional. It is typical to avoid developing new specialized
> techniques that work directly on tensors, but rather apply existent
> algorithms on the data reshaped into matrices or vectors. However, such
> structural simplification of the data can lead to loss of interpretability
> as well as to immense comparative memory loss. Modewise tensor maps are
> linear operators acting on each tensor dimension (mode) separately rather
> than on the vectorizations of tensors. They are efficient: they are faster
> and require significantly less memory than the measurements working on
> vectorized tensors. They are also more sophisticated and present additional
> mathematical challenges due to their inherent Kronecker structure. In this
> talk, I will discuss data-oblivious tensor dimension reduction with
> these modewise maps and its applications to the tensor low-rank fitting and
> recovery. Based on the joint work with Mark Iwen, Deanna Needell, Mike
> Perlmutter, and Ali Zare.
>
> *Bio:* Liza Rebrova is an Assistant Professor at Princeton University,
> ORFE Department (Operations Research and Financial Engineering). She
> received her Ph.D. degree in Mathematics (Non-asymptotic Random Matrix
> Theory) from the University of Michigan in 2018. Then, she worked as an
> Assistant Adjunct Professor in the Department of Mathematics at UCLA
> (2018-2020) and as a Postdoctoral Scholar at the Computational Research
> Division at Lawrence Berkeley National Laboratory (in 2020). Her research
> interests are in high-dimensional probability, and include developing
> randomized and high-dimensional methods for numerical linear algebra,
> compressed sensing, and the math of data.
>
> *Host:* *Yury Makarychev* <yury at ttic.edu>
>
> ***********************************************************************************
>
> For more information on the colloquium series or to subscribe to the
> mailing list, please see http://www.ttic.edu/colloquium.php
>
>
>
>
> Mary C. Marre
> Faculty Administrative Support
> *Toyota Technological Institute*
> *6045 S. Kenwood Avenue*
> *Chicago, IL  60637*
> *mmarre at ttic.edu <mmarre at ttic.edu>*
>
>
> On Sun, May 15, 2022 at 3:27 PM Mary Marre <mmarre at ttic.edu> wrote:
>
>> *When:*        Monday, May 16h at* 11:30 am CT*
>>
>>
>> *Where:       *Talk will be given *live, in-person* at
>>
>>                    TTIC, 6045 S. Kenwood Avenue
>>
>>                    5th Floor, Room 530
>>
>>
>> *Where:*       Zoom Virtual Talk (*register in advance here
>> <https://uchicagogroup.zoom.us/webinar/register/WN_QFCPyH-nSv6ovsQ_tgTwag>*
>> )
>>
>>
>> *Who: *         Elizaveta Rebrova, Princeton University
>>
>>
>> *Title:  * Modewise Methods for Tensor Compression and Recovery
>> *Abstract: *Tensor methods are unsurprisingly in demand since lots of
>> natural data is inherently high-order (e.g., temporal or multi-factor) and
>> high-dimensional. It is typical to avoid developing new specialized
>> techniques that work directly on tensors, but rather apply existent
>> algorithms on the data reshaped into matrices or vectors. However, such
>> structural simplification of the data can lead to loss of interpretability
>> as well as to immense comparative memory loss. Modewise tensor maps are
>> linear operators acting on each tensor dimension (mode) separately rather
>> than on the vectorizations of tensors. They are efficient: they are faster
>> and require significantly less memory than the measurements working on
>> vectorized tensors. They are also more sophisticated and present additional
>> mathematical challenges due to their inherent Kronecker structure. In this
>> talk, I will discuss data-oblivious tensor dimension reduction with
>> these modewise maps and its applications to the tensor low-rank fitting and
>> recovery. Based on the joint work with Mark Iwen, Deanna Needell, Mike
>> Perlmutter, and Ali Zare.
>>
>> *Bio:* Liza Rebrova is an Assistant Professor at Princeton University,
>> ORFE Department (Operations Research and Financial Engineering). She
>> received her Ph.D. degree in Mathematics (Non-asymptotic Random Matrix
>> Theory) from the University of Michigan in 2018. Then, she worked as an
>> Assistant Adjunct Professor in the Department of Mathematics at UCLA
>> (2018-2020) and as a Postdoctoral Scholar at the Computational Research
>> Division at Lawrence Berkeley National Laboratory (in 2020). Her research
>> interests are in high-dimensional probability, and include developing
>> randomized and high-dimensional methods for numerical linear algebra,
>> compressed sensing, and the math of data.
>>
>> *Host:* *Yury Makarychev* <yury at ttic.edu>
>>
>> ***********************************************************************************
>>
>> For more information on the colloquium series or to subscribe to the
>> mailing list, please see http://www.ttic.edu/colloquium.php
>>
>>
>>
>>
>>
>>
>> Mary C. Marre
>> Faculty Administrative Support
>> *Toyota Technological Institute*
>> *6045 S. Kenwood Avenue*
>> *Chicago, IL  60637*
>> *mmarre at ttic.edu <mmarre at ttic.edu>*
>>
>>
>> On Mon, May 9, 2022 at 4:20 PM Mary Marre <mmarre at ttic.edu> wrote:
>>
>>> *When:*        Monday, May 16h at* 11:30 am CT*
>>>
>>>
>>> *Where:       *Talk will be given *live, in-person* at
>>>
>>>                    TTIC, 6045 S. Kenwood Avenue
>>>
>>>                    5th Floor, Room 530
>>>
>>>
>>> *Where:*       Zoom Virtual Talk (*register in advance here
>>> <https://uchicagogroup.zoom.us/webinar/register/WN_QFCPyH-nSv6ovsQ_tgTwag>*
>>> )
>>>
>>>
>>> *Who: *         Elizaveta Rebrova, Princeton University
>>>
>>>
>>> *Title:  * Modewise Methods for Tensor Compression and Recovery
>>> *Abstract: *Tensor methods are unsurprisingly in demand since lots of
>>> natural data is inherently high-order (e.g., temporal or multi-factor) and
>>> high-dimensional. It is typical to avoid developing new specialized
>>> techniques that work directly on tensors, but rather apply existent
>>> algorithms on the data reshaped into matrices or vectors. However, such
>>> structural simplification of the data can lead to loss of interpretability
>>> as well as to immense comparative memory loss. Modewise tensor maps are
>>> linear operators acting on each tensor dimension (mode) separately rather
>>> than on the vectorizations of tensors. They are efficient: they are faster
>>> and require significantly less memory than the measurements working on
>>> vectorized tensors. They are also more sophisticated and present additional
>>> mathematical challenges due to their inherent Kronecker structure. In this
>>> talk, I will discuss data-oblivious tensor dimension reduction with
>>> these modewise maps and its applications to the tensor low-rank fitting and
>>> recovery. Based on the joint work with Mark Iwen, Deanna Needell, Mike
>>> Perlmutter, and Ali Zare.
>>>
>>> *Bio:* Liza Rebrova is an Assistant Professor at Princeton University,
>>> ORFE Department (Operations Research and Financial Engineering). She
>>> received her Ph.D. degree in Mathematics (Non-asymptotic Random Matrix
>>> Theory) from the University of Michigan in 2018. Then, she worked as an
>>> Assistant Adjunct Professor in the Department of Mathematics at UCLA
>>> (2018-2020) and as a Postdoctoral Scholar at the Computational Research
>>> Division at Lawrence Berkeley National Laboratory (in 2020). Her research
>>> interests are in high-dimensional probability, and include developing
>>> randomized and high-dimensional methods for numerical linear algebra,
>>> compressed sensing, and the math of data.
>>>
>>> *Host:* *Yury Makarychev* <yury at ttic.edu>
>>>
>>> ***********************************************************************************
>>>
>>> For more information on the colloquium series or to subscribe to the
>>> mailing list, please see http://www.ttic.edu/colloquium.php
>>>
>>>
>>>
>>> Mary C. Marre
>>> Faculty Administrative Support
>>> *Toyota Technological Institute*
>>> *6045 S. Kenwood Avenue*
>>> *Chicago, IL  60637*
>>> *mmarre at ttic.edu <mmarre at ttic.edu>*
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/theory/attachments/20220516/32d37ca4/attachment-0001.html>