Manolis K

Data Scientist, AI/ML for Diagnostic Cardiology, Physics PhD

[Manolis K @ManolisKarg]

@jwestonhughes @StanfordMed We may be able to demonstrate that bias, if you would like to collaborate. It’s an insight sorely missing from the literature.

[Manolis K @ManolisKarg]

@jwestonhughes @StanfordMed The huge drop in performance, when the DNN is applied on a general population with realistic prevalence, is likely a strong indication of bias. The PPV of such a model would likely be very low if deployed for screening.

[Manolis K @ManolisKarg]

More self-reference is all you need

https://t.co/IrSGxEbHH0

[Manolis K @ManolisKarg]

@nirsd Very cool, which makes “remembering” a much deeper and more creative process.

[Manolis K @ManolisKarg]

@nirsd Where each character goes through periods believing different interpretations of quantum mechanics, in each of their many worlds incarnations.

[Manolis K @ManolisKarg]

Old podcast with great meta learning concepts, on generative models for explainability. Very similar concepts to my undergraduate thesis work on AMIAS (https://t.co/vL3AAND7JH). https://t.co/C6Ag84Iy9r

[Manolis K @ManolisKarg]

A model is only as good as the data it is trained on. Synthetic datasets hold great promise to address difficult problems where labeled samples are hard to obtain.
https://t.co/RK3E2rBOSh

[Manolis K @ManolisKarg]

@GoogleAI Great tool for people who feel they need to explain their joke, otherwise you would have laughed.

[Google AI @GoogleAI]

Introducing the 540 billion parameter Pathways Language Model. Trained on two Cloud #TPU v4 pods, it achieves state-of-the-art performance on benchmarks and shows exciting capabilities like mathematical reasoning, code writing, and even explaining jokes. https://t.co/NFHFAHgUkB

[Manolis K @ManolisKarg]

Wonderful news! Congrats Νικολία, Fermilab is in very capable hands!
https://t.co/DZZAgC57S8 #labdirector #LBNF/DUNE #people #PIP-II

[Manolis K @ManolisKarg]

Nice overview. Completely missing is discussion of needs and challenges for ML in smaller experimental areas outside the 3 examined. https://t.co/w6aInMMEjQ

[Manolis K @ManolisKarg]

It helps that Michael is a brilliant leader in his field. As a graduate student I loved his papers. https://t.co/wdfoSN5jxj

[Manolis K @ManolisKarg]

Reinforcement learning discovering new modes. Reminiscent of that famous AlphaGo move everyone thought was bad. https://t.co/lr4IcIZOpb

[Manolis K @ManolisKarg]

Beautiful and innovative techniques - if apparently computationally heavy. Your experience with kNN ideas is evident. Excellent scholarly discussion too. I will be sure to check out the earlier work.

[Tommaso Dorigo @dorigo]

Deep learning 66M parameters with a SGD-optimized nearest neighbor gets you to the same performance of NNs and BDTs! The power of ML is in differentiable calculus, as expected. https://t.co/BdvOOGB3tw

[Manolis K @ManolisKarg]

@Bucks @Thanasis_ante43 @Giannis_An34 Οκ έρχομαι!

[Fermilab @Fermilab]

We're less than 24 hours away from learning the first results from the Muon g-2 experiment at Fermilab! #gminus2
https://t.co/92KZ5oeb1t

[Manolis K @ManolisKarg]

Are dogs allowed in South Pole, and do we have to apply jointly?

[IceCube Neutrino Observatory @uw_icecube]

It's that time of the year...

📢 We are now accepting applications for next year's South Pole winterovers!

Apply by January 31, 2021 ➡ https://t.co/MM8ui6iQa3

[Manolis K @ManolisKarg]

Very happy with my match, looking ahead to an interesting connection and conversation! This program is aiming to replace conference interactions that are no longer available.

[Manolis K @ManolisKarg]

Fight the incredible cruelty of the #StudentBan

[Chris Marsicano @ChrisMarsicano]

Well, it finally happened. This #ICE International #StudentBan got me back into the “organizing effective lobbying action via twitter” game.

Want to make sure our lovely international students can keep their visas? Here’s how to do it:

THREAD.

https://t.co/k8xmv5dNyE

1/14

[Scientific American @sciam]

Climate change made Europe’s mega-heatwave five times more likely https://t.co/5ppFxAHPEk