Second ML-INFN Hackathon: Starting Level

Survey

ML_INFN Second Hackathon Survey

General Info

Please state here again (the survey is anonymous, we cannot reuse the info provided at registration) your initial level

Python proficiency before the course*

Low

Medium

Full Proficiency

Consider this on "the day before the course", so including the self training you might have done following our suggestions

Machine learning tools (Keras, Tensorflow) level before the course*

Low

Medium

Full Proficiency

Consider this on "the day before the course", so including the self training you might have done following our suggestions

Proficiency in Cloud technologies before the course*

Low

Medium

Full Proficiency

Consider this on "the day before the course", so including the self training you might have done following our suggestions

Did you find useful the self-training suggestions?*

No

Yes, but did not study on them

Yes, and I used them

Any comment you want to leave us?

ML Introductory Lessons

• Maurizio Pierini, "What is Machine Learning and why it is relevant to INFN", Monday morning
  
  * Stefano Giagu, "Neural Network architectures", Monday morning

How did you find them?*

Not very interesting...

Too complex for the entry-level hackathon

Too simple (or generic) and lacking of depth or detail

A nice opportunity

Duration: was it appropriate to the purpose of the hackathon?*

Allocated time was not sufficient

Duration was right for the training

Too much time was spent on theoretical training

Any comment you want to leave us?

Seminars dedicated to state-of-the-art applications within INFN research fields

• Tommaso Boccali, "Machine Learning in HEP", Tuesday morning
  
  * Alessandra Retico, "Machine Learning in Medicine and Medical Physics", Tuesday morning

Did you find them useful?*

No - their time should be allocate in other ways

No - their level was too low

No - their level was too complex

Yes - their length / number was just about right

Yes - I would have liked more time spent on them

Is there any other INFN-related specific field you would have liked to see covered?

Hands-on sessions

• Lucio Anderlini, "Playing with numpy, pandas and CERN OpenData", Monday afternoon
  
  * Andrea Rizzi, "Introduction to keras", Monday afternoon
  
  * Lucio Anderlini, "A simple classification in HEP", Tueday morning
  
  * Alessandro Bombini, "Digital restoration of paintures from XRF spectra", Tueday morning
  
  * Andrea Rizzi, "Regression task and Convolutional Neural Network", Tuesday morning

Did you follow the hands-on sessions?*

Yes

No

Partially

Rate the form of the hands on (a teacher filling/discussing a notebook online)*

Good. Easy to follow and interactive enough

Notebooks are ok, but they were too complete: speaker should have written more code

Speakers were too fast and the lack of slide support was frustrating

Bad. Focus was on code rather than ML techniques

I did not follow the hands on

Was the time dedicated to hands-on correct?*

Too short

Too long

It was ok!

I did not follow the hands on

Was the difficulty level appropriate?*

Too complex

Too easy

It was ok!

I did not follow the hands on

Any further comment you want to leave?

The hackathon session

• The hackathon session on Wednesday morning
  
  * The report session on Wednesday afternoon
  
  
  
  While everyone will easily agree that an in-person session would have be smoother, which was your experience on the group work in the hackathon?

Which use case was assigned to you?*

Virgo - Gravitational Waves

CMS - Btag

CMS - Higgs

Image Segmentation - Medicine and Medical Physics

Did you participate to the hackathon?*

Yes

No

For a fraction of the time

About the proposed exercises, did you find their level :*

Too easy

Too simple

It was ok!

I did not follow the hackathon

About the proposed exercises, did you find them interesting?*

Yes

No

I did not follow the hackathon

About the continuous availability of a tutor, did you find it:*

Useful

Useful but not essential

Not useful

Stressful

I did not follow the hackathon

How ofter did your group had to request help from the tutors?*

Rarely

Sometimes

Often

Practically all the time

I did not follow the hackathon

Was your group able to successfully complete the exercise?*

No or in very small part

The first part without the additional ones

All the first part plus some of the additional parts

Yes! Including the additional parts

I did not follow the hackathon

Did you find the final discussion with reports from the 3 groups useful?

No selection

Yes, it was ok!

Yes, but it would deserve more time

Yes, but allocated time was too much

No, I would have preferred more time for the hackathon

No, reports from other groups are not very useful

I did not follow the hackathon

Was the technical setup reasonable?*

Yes

No

Partially (please specify below)

I did not follow the hackathon

Again, in the current covid situation. Was 6 students + 1 tutor a reasonable balance? Was the use of Zoom breakout rooms good enough?

Please add specific comments in the free text form below

Suggestions?

The Cloud / Jupyter setup

Did you find the technical setup using Cloud + Jupyter reasonable for ML oriented analyses?*

Yes, worked for me

Yes, it generally worked for me (please add comment below)

No (please explain below why)

Specific comments

Do you think you will be using Cloud based setups similar to this in the future?*

No

No as I already use another Cloud setup (for example, SWAN@CERN)

Yes if INFN provides me with enough / adequate resources

Yes

Additional comments

Please add any additional comment below

Feel free to add comments

Your answers will be submitted anonymously.