Pol del Aguila Pla

Image processing specialist at Qamcom Research and Technology Stockholm. Opinions are my own.

[Pol del Aguila Pla @polsocjo]

@mattforney No, you don't need to carry id on you in many European countries, whoever you are.

[Pol del Aguila Pla @polsocjo]

@nixcraft They should implement `bet` that just does the usual thing. Then one can choose whether to bet or vet according to that day's mood 😅🤣

[Pol del Aguila Pla @polsocjo]

@MarkSchmidtUBC @Dirque_L So picky with those pesky details 🤣😅

[Pol del Aguila Pla @polsocjo]

@grok @MarstenJack @Chrisxy24 @FBIDirectorKash @mattgaetz He could just be asking for him to be impeached, no?

[Pol del Aguila Pla @polsocjo]

@wmhuo168 "The Chinese record processor was Swedish." -Ny teknik.

https://t.co/Gde5uHqqvX

[Pol del Aguila Pla @polsocjo]

@fchollet Wrong. It depends on what you're trying to do. If you want to predict what you cannot see, getting something very correlated to it will help you reconstruct it. Who cares about causality.

[Pol del Aguila Pla @polsocjo]

@fchollet This stopped being science long ago

[Pol del Aguila Pla @polsocjo]

@docmilanfar Yes, definitely yes

[Pol del Aguila Pla @polsocjo]

Super interesting stuff. Congrats @MehrsaPourya

[IEEE Trans. on Computational Imaging @IEEE_TCI]

Inverse Problems: Discrete or Continuous? @MehrsaPourya et al. from @big_epfl explain that box splines and multi-res bridge the gap:

"A Box-Spline Framework for Inverse Problems with Continuous-Domain Sparsity Constraints”
On early access:
https://t.co/DHIBwsHCKu

[IEEE Trans. on Computational Imaging @IEEE_TCI]

Inverse Problems: Discrete or Continuous? @MehrsaPourya et al. from @big_epfl explain that box splines and multi-res bridge the gap:

"A Box-Spline Framework for Inverse Problems with Continuous-Domain Sparsity Constraints”
On early access:
https://t.co/DHIBwsHCKu

[Pol del Aguila Pla @polsocjo]

@airasia I have been trying to get a response to no avail. Can you reach out to me please?

[Pol del Aguila Pla @polsocjo]

Adding noise before some nonlinearities helps preserve information. We found this also in statistical models for electronics :-)

Our sawtooth model and how noise makes or breaks the game: https://t.co/oWulS7GVoU

[Dirk L. @Dirque_L]

Just learned: Adding randomness before quantization helps! One-bit quantization of X is sign(X). If |X|≤c then it's better to store sign(X+Y) where Y is uniformly distributed on [-c,c]!

Why? We still have

E[X] = E[c*sign(X+Y)]

1/n
https://t.co/W6VmZaVtwi

[Pol del Aguila Pla @polsocjo]

@jburnmurdoch No sleep for me tonight...

[Pol del Aguila Pla @polsocjo]

@profdvp Why not?

[Pol del Aguila Pla @polsocjo]

@outrun86 @dehshibi @docmilanfar I'll agree that no one has the experience or expertise to review an enormous volume of papers well. And yep, I just try to laugh at some editorial/reviewer decisions too.

[Pol del Aguila Pla @polsocjo]

@docmilanfar A joke?

[Pol del Aguila Pla @polsocjo]

@ccanonne_ I share your pain

[Pol del Aguila Pla @polsocjo]

@gaurav_ven @dgonzales1990 Oh. I'd like to read more about this 😱

[Pol del Aguila Pla @polsocjo]

@ukmlv @SLCL Strap your seatbelt on. Enjoy