"Safe and Pleasant ML": today 4:30-6pm

[Artur Dubrawski]

We will try to finish covering the material in Session 2.

We will be meeting in NSH 3305.

The slides are posted on:
https://drive.google.com/drive/folders/1wd2DHkDBIgWzBjwiDFr0FWRVVESfhhEg?usp=sharing

Cheers,
Artur Dubrawski

On Fri, Jul 22, 2022 at 11:44 AM Artur Dubrawski <awd at cs.cmu.edu> wrote:

> We will continue into Session 2 today, contents of which include:
> * The elusive concept of “perfect data” and getting our real-world data
> closer to perfection
> * Basic feature selection algorithms
> * Feature reduction with Principal Components Analysis (PCA)
>   - Additional uses of PCA
>   - Related techniques
> * Feature reduction with Autoencoders (AE)
> * Folding data with Multi-Dimensional Scaling (MDS)
> * Unfolding data with “Kernel trick”
> * A quick look at preparing textual data for ML
>
> We will probably not be able to cover everything in one seating, so expect
> the session to be continued in the next meetings.
>
> Cheers
> Artur
>
> On Fri, Jul 15, 2022 at 2:48 PM Artur Dubrawski <awd at cs.cmu.edu> wrote:
>
>> We are continuing today, 4:30pm, NSH 3305 and zoom.
>>
>> Artur
>>
>> On Thu, Jul 7, 2022 at 6:32 PM Artur Dubrawski <awd at cs.cmu.edu> wrote:
>>
>>> We will continue our introduction to safe and pleasant machine
>>> learning tomorrow as scheduled.
>>> This time, we will talk about bootstrapping, finish the coverage of ROC
>>> methodology, and briefly talk about McNemar's tests and classifier
>>> calibration.
>>>
>>> Slides from this and previous meetings are available at
>>> https://drive.google.com/drive/folders/1wd2DHkDBIgWzBjwiDFr0FWRVVESfhhEg
>>>
>>> (I'd like to give Dr. Gisolfi credit for coming up with the apt title of
>>> the above directory)
>>> (So please let him know if you have any trouble accessing it :P )
>>>
>>> Cheers,
>>> Artur
>>>
>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/autonlab-users/attachments/20220805/7247ad1d/attachment.html>