[Theory] NOW: 10/10 Thesis Defense: Ankita Pasad, TTIC
Mary Marre via Theory
theory at mailman.cs.uchicago.edu
Thu Oct 10 11:30:05 CDT 2024
*When*: Thursday, October 10th from* 11**:30am - 12:30am CT*
*Where*: Talk will be given *live, in-person* at
TTIC, 6045 S. Kenwood Avenue
5th Floor, *Room 530*
*Virtually*: via *Zoom*
<https://uchicago.zoom.us/j/97405333371?pwd=KaGiUd7tjzjqk6worbEUJaIz5AioNN.1>
*Who: * Ankita Pasad, TTIC
*Title:* What do Speech Foundation Models Learn? Analysis and Applications
*Abstract: *Speech foundation models (SFMs) are designed to serve as
general-purpose representations for a broad spectrum of speech-processing
tasks. The last four years have seen an influx of increasingly successful
SFMs with impressive performance on various downstream tasks. While the zoo
of SFMs continues to grow, we are far behind in understanding the knowledge
acquired during pre-training. Our limited understanding makes it difficult
to navigate the space of SFMs, as extensive trial and error is cumbersome
and impractical for these computationally demanding models.
In this talk, I will present our lightweight analysis framework,
highlighting our findings from a comparative analysis of a diverse set of
SFMs and how our insights can inform design choices for model adaptation. I
will also discuss our contribution to the Spoken Language Understanding
Evaluation (SLUE) benchmark, the first large-scale collection of
open-source and natural-speech SLU datasets.
Collectively, this thesis tackles previously unanswered questions about
SFMs, providing tools and datasets to further our understanding and enable
the community to make informed design choices for future model development
and adoption.
*Thesis Committee: *Karen Livescu *(Thesis Advisor)*, Kevin Gimpel
(QuillBot), Michael Auli (Meta), Allyson Ettinger (AI2)
Mary C. Marre
Faculty Administrative Support
*Toyota Technological Institute*
*6045 S. Kenwood Avenue, Rm 517*
*Chicago, IL 60637*
*773-834-1757*
*mmarre at ttic.edu <mmarre at ttic.edu>*
On Wed, Oct 9, 2024 at 2:46 PM Mary Marre <mmarre at ttic.edu> wrote:
> *When*: Thursday, October 10th from* 11**:30am - 12:30am CT*
>
> *Where*: Talk will be given *live, in-person* at
> TTIC, 6045 S. Kenwood Avenue
> 5th Floor, *Room 530*
>
> *Virtually*: via *Zoom*
> <https://uchicago.zoom.us/j/97405333371?pwd=KaGiUd7tjzjqk6worbEUJaIz5AioNN.1>
>
>
> *Who: * Ankita Pasad, TTIC
>
>
> *Title:* What do Speech Foundation Models Learn? Analysis and Applications
>
> *Abstract: *Speech foundation models (SFMs) are designed to serve as
> general-purpose representations for a broad spectrum of speech-processing
> tasks. The last four years have seen an influx of increasingly successful
> SFMs with impressive performance on various downstream tasks. While the zoo
> of SFMs continues to grow, we are far behind in understanding the knowledge
> acquired during pre-training. Our limited understanding makes it difficult
> to navigate the space of SFMs, as extensive trial and error is cumbersome
> and impractical for these computationally demanding models.
>
> In this talk, I will present our lightweight analysis framework,
> highlighting our findings from a comparative analysis of a diverse set of
> SFMs and how our insights can inform design choices for model adaptation. I
> will also discuss our contribution to the Spoken Language Understanding
> Evaluation (SLUE) benchmark, the first large-scale collection of
> open-source and natural-speech SLU datasets.
>
> Collectively, this thesis tackles previously unanswered questions about
> SFMs, providing tools and datasets to further our understanding and enable
> the community to make informed design choices for future model development
> and adoption.
>
> *Thesis Committee: *Karen Livescu *(Thesis Advisor)*, Kevin Gimpel
> (QuillBot), Michael Auli (Meta), Allyson Ettinger (AI2)
>
>
>
>
> Mary C. Marre
> Faculty Administrative Support
> *Toyota Technological Institute*
> *6045 S. Kenwood Avenue, Rm 517*
> *Chicago, IL 60637*
> *773-834-1757*
> *mmarre at ttic.edu <mmarre at ttic.edu>*
>
>
> On Mon, Oct 7, 2024 at 10:05 PM Mary Marre <mmarre at ttic.edu> wrote:
>
>> *When*: Thursday, October 10th from* 11**:30am - 12:30am CT*
>>
>> *Where*: Talk will be given *live, in-person* at
>> TTIC, 6045 S. Kenwood Avenue
>> 5th Floor, *Room 530*
>>
>> *Virtually*: via *Zoom*
>> <https://uchicago.zoom.us/j/97405333371?pwd=KaGiUd7tjzjqk6worbEUJaIz5AioNN.1>
>>
>>
>> *Who: * Ankita Pasad, TTIC
>>
>>
>> *Title:* What do Speech Foundation Models Learn? Analysis and
>> Applications
>>
>> *Abstract: *Speech foundation models (SFMs) are designed to serve as
>> general-purpose representations for a broad spectrum of speech-processing
>> tasks. The last four years have seen an influx of increasingly successful
>> SFMs with impressive performance on various downstream tasks. While the zoo
>> of SFMs continues to grow, we are far behind in understanding the knowledge
>> acquired during pre-training. Our limited understanding makes it difficult
>> to navigate the space of SFMs, as extensive trial and error is cumbersome
>> and impractical for these computationally demanding models.
>>
>> In this talk, I will present our lightweight analysis framework,
>> highlighting our findings from a comparative analysis of a diverse set of
>> SFMs and how our insights can inform design choices for model adaptation. I
>> will also discuss our contribution to the Spoken Language Understanding
>> Evaluation (SLUE) benchmark, the first large-scale collection of
>> open-source and natural-speech SLU datasets.
>>
>> Collectively, this thesis tackles previously unanswered questions about
>> SFMs, providing tools and datasets to further our understanding and enable
>> the community to make informed design choices for future model development
>> and adoption.
>>
>> *Thesis Committee: *Karen Livescu *(Thesis Advisor)*, Kevin Gimpel
>> (QuillBot), Michael Auli (Meta), Allyson Ettinger (AI2)
>>
>>
>>
>>
>> Mary C. Marre
>> Faculty Administrative Support
>> *Toyota Technological Institute*
>> *6045 S. Kenwood Avenue, Rm 517*
>> *Chicago, IL 60637*
>> *773-834-1757*
>> *mmarre at ttic.edu <mmarre at ttic.edu>*
>>
>>
>> On Tue, Oct 1, 2024 at 4:37 PM Mary Marre <mmarre at ttic.edu> wrote:
>>
>>> *When*: Thursday, October 10th from* 11**:30am - 12:30am CT*
>>>
>>> *Where*: Talk will be given *live, in-person* at
>>> TTIC, 6045 S. Kenwood Avenue
>>> 5th Floor, *Room 530*
>>>
>>> *Virtually*: via *Zoom*
>>> <https://uchicago.zoom.us/j/97405333371?pwd=KaGiUd7tjzjqk6worbEUJaIz5AioNN.1>
>>>
>>>
>>> *Who: * Ankita Pasad, TTIC
>>>
>>>
>>> *Title:* What do Speech Foundation Models Learn? Analysis and
>>> Applications
>>>
>>> *Abstract: *Speech foundation models (SFMs) are designed to serve as
>>> general-purpose representations for a broad spectrum of speech-processing
>>> tasks. The last four years have seen an influx of increasingly successful
>>> SFMs with impressive performance on various downstream tasks. While the zoo
>>> of SFMs continues to grow, we are far behind in understanding the knowledge
>>> acquired during pre-training. Our limited understanding makes it difficult
>>> to navigate the space of SFMs, as extensive trial and error is cumbersome
>>> and impractical for these computationally demanding models.
>>>
>>> In this talk, I will present our lightweight analysis framework,
>>> highlighting our findings from a comparative analysis of a diverse set of
>>> SFMs and how our insights can inform design choices for model adaptation. I
>>> will also discuss our contribution to the Spoken Language Understanding
>>> Evaluation (SLUE) benchmark, the first large-scale collection of
>>> open-source and natural-speech SLU datasets.
>>>
>>> Collectively, this thesis tackles previously unanswered questions about
>>> SFMs, providing tools and datasets to further our understanding and enable
>>> the community to make informed design choices for future model development
>>> and adoption.
>>>
>>> *Thesis Committee: *Karen Livescu *(Thesis Advisor)*, Kevin Gimpel
>>> (QuillBot), Michael Auli (Meta), Allyson Ettinger (AI2)
>>>
>>>
>>>
>>>
>>> Mary C. Marre
>>> Faculty Administrative Support
>>> *Toyota Technological Institute*
>>> *6045 S. Kenwood Avenue, Rm 517*
>>> *Chicago, IL 60637*
>>> *773-834-1757*
>>> *mmarre at ttic.edu <mmarre at ttic.edu>*
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/theory/attachments/20241010/d323d96d/attachment-0001.html>
More information about the Theory
mailing list