J. Miguel Valverde

PhD student at UEF. MRI. Segmentation. Deep Learning.

[J. Miguel Valverde @jmlipman]

I was recently interviewed by https://t.co/IZrZATBK21 (Open Science)

https://t.co/X17dP71klC

[AvoinTiede @AvoinTiede]

The policy on open research data and methods has been updated with a policy component on open research methods and infrastructures 🎉 Huge thanks to the working group and the whole research community! #OpenScience

https://t.co/r7jktgDuzw

[J. Miguel Valverde @jmlipman]

@Rivsie Ciertamente! Lo han debido de poner nuevo también.

[J. Miguel Valverde @jmlipman]

Acabo de entrar a Twitter después de mucho tiempo y lo primero que veo son tweets de gente que no sigo y tiene la marquita azul. Cómo aguantáis? Lo mejor de haberme ido a Mastodon es que, al no tener como objetivo engancharme (como twitter), cada vez uso menos las redes sociales.

[J. Miguel Valverde @jmlipman]

Y cada pocos segundos me salen como 35 tweets nuevos para ver, siendo que la mayoria son de otras cuentas que no sigo y de hace muchas horas, wtf twitter. Qué eras y en qué te has convertido.

[J. Miguel Valverde @jmlipman]

You can find more details, analyses and experiments in the paper, which is open access 🔓 https://t.co/uanVNoweMt

[J. Miguel Valverde @jmlipman]

🚨 Paper alert 🚨

Our paper "RegionWise loss for biomedical image segmentation" (w/ @jussi_tohka) was recently published in Pattern Recognition https://t.co/uanVNoweMt

We introduce a new stable way to optimize CNNs to account for pixel importance. What does this mean?

👇🧵

[J. Miguel Valverde @jmlipman]

Moreover, we show that several loss functions that were apparently different can be reformulated as RegionWise loss, unveiling their similarities and providing a new understanding of those loss functions.