Jacob Deasy (@jacobdeasy) 's Twitter Profile
Jacob Deasy

@jacobdeasy

AI PhD @Cambridge_Uni | Former AI Resident @Theteamatx | Former Research Intern @ Huawei AI Research | Comp Bio @Cambridge_Uni | Maths @imperialcollege

ID: 297011825

linkhttps://jacobdeasy.github.io/ calendar_today11-05-2011 19:39:57

44 Tweet

164 Followers

284 Following

Aleco Kastanos (@alecokas) 's Twitter Profile Photo

Here's another post from my recent excursion into Julia. It's a step-by-step tutorial on convolutional variational autoencoders in Flux: alecokas.github.io/julia/flux/vae… #JuliaLang #FluxML #JuliaCon

_hylandSL - not here (@_hylandsl) 's Twitter Profile Photo

MIMIC-IV is out! Alistair Johnson has a great two-part tutorial at chilconference explaining the data (and what's changed from MIMIC-III), and walking through a full study using MIMIC-IV chilconference.org/tutorial_d.html chilconference.org/tutorial_f.html

Imperial College London (@imperialcollege) 's Twitter Profile Photo

"Maths is truth. Once you discover something in maths, it applies to all eternity" A huge congratulations to Professor Sir Martin Hairer from our Imperial_Maths who is the single winner of the $3m 2021 Breakthrough Prize in Mathematics (Breakthrough): ow.ly/fYY550Bn2SF

Jacob Deasy (@jacobdeasy) 's Twitter Profile Photo

*After 3 years* my Imperial Masters project on "The r-Hunter–Saxton equation, smooth and singular solutions and their approximation" has finally been published in non-linearity! iopscience.iop.org/article/10.108…

AriErcole (@ariercole) 's Twitter Profile Photo

Time-sensitive prediction for the #ICU using thousands of features with 𝙣𝙀 π™£π™šπ™šπ™™ 𝙩𝙀 π™¨π™šπ™‘π™šπ™˜π™©, π™˜π™ͺπ™§π™–π™©π™š 𝙀𝙧 π™˜π™‘π™šπ™–π™£ π™©π™π™š 𝙙𝙖𝙩𝙖 π™›π™žπ™§π™¨π™© by Jacob Deasy Pietro Lio' and AriErcole using #MachineLearning nature.com/articles/s4159… doi.org/10.1038/s41598…

Time-sensitive prediction for the #ICU using thousands of features with 𝙣𝙀 π™£π™šπ™šπ™™ 𝙩𝙀 π™¨π™šπ™‘π™šπ™˜π™©, π™˜π™ͺπ™§π™–π™©π™š 𝙀𝙧 π™˜π™‘π™šπ™–π™£ π™©π™π™š 𝙙𝙖𝙩𝙖 π™›π™žπ™§π™¨π™© by <a href="/JacobDeasy/">Jacob Deasy</a> <a href="/pl219_Cambridge/">Pietro Lio'</a> and <a href="/AriErcole/">AriErcole</a> using #MachineLearning

nature.com/articles/s4159…
doi.org/10.1038/s41598…
Tim Cook (@tim_cook) 's Twitter Profile Photo

We believe users should have the choice over the data that is being collected about them and how it’s used. Facebook can continue to track users across apps and websites as before, App Tracking Transparency in iOS 14 will just require that they ask for your permission first.

We believe users should have the choice over the data that is being collected about them and how it’s used. Facebook can continue to track users across apps and websites as before, App Tracking Transparency in iOS 14 will just require that they ask for your permission first.
Papers with Code (@paperswithcode) 's Twitter Profile Photo

βͺ Papers with Code: Year in Review. We’re ending the year by taking a look back at the top trending papers, libraries and benchmarks for 2020. Read on below! medium.com/paperswithcode…

βͺ Papers with Code: Year in Review. We’re ending the year by taking a look back at the top trending papers, libraries and benchmarks for 2020. Read on below!

medium.com/paperswithcode…
Andrei Margeloiu (@andreimargeloiu) 's Twitter Profile Photo

πŸŽ‰New paper reviewπŸ“Ή Even LSTMs suffer from forgetting the input. Neural Turing Machines use external memory, similar to a CPU-Registers-RAM architecture πŸ’». Find out how the memory is addressed using end-to-end differentiable operations πŸ‘‡ youtu.be/XBgwq4IePLw

πŸŽ‰New paper reviewπŸ“Ή Even LSTMs suffer from forgetting the input.

Neural Turing Machines use external memory, similar to a CPU-Registers-RAM architecture πŸ’». Find out how the memory is addressed using end-to-end differentiable operations πŸ‘‡

youtu.be/XBgwq4IePLw
Aleco Kastanos (@alecokas) 's Twitter Profile Photo

Our paper on Graph Convolutional Networks in semi-supervised Swahili news classification has been accepted at #AfricaNLP #EACL2021! Have a look at the arxiv: arxiv.org/abs/2103.09325 and code: github.com/alecokas/swahi… Dataset thanks to Davis David β„’ #NLProc

Our paper on Graph Convolutional Networks in semi-supervised Swahili news classification has been accepted at #AfricaNLP #EACL2021! Have a look at the arxiv: arxiv.org/abs/2103.09325  and code: github.com/alecokas/swahi…

Dataset thanks to <a href="/Davis_McDavid/">Davis David β„’</a> 
#NLProc
Aleco Kastanos (@alecokas) 's Twitter Profile Photo

Find our presentation on GCNs for Swahili news classification in the #AfricaNLP2021 presentations playlist: youtube.com/watch?v=UtvOIh…

Cristian Bodnar (@crisbodnar) 's Twitter Profile Photo

Happy that our MPSN paper has been accepted to #ICML2021. Final version with improvements & code coming soon. Work w/ Fabrizio Frasca*, Yu Guang Wang*, Nina Otter, G. MontΓΊfar*, Pietro Lio', Michael Bronstein arXiv: arxiv.org/abs/2103.03212

Happy that our MPSN paper has been accepted to #ICML2021. Final version with improvements &amp; code coming soon. Work w/ <a href="/ffabffrasca/">Fabrizio Frasca</a>*, <a href="/wangyg85/">Yu Guang Wang</a>*, <a href="/kneppkatt/">Nina Otter</a>, G. MontΓΊfar*, <a href="/pl219_Cambridge/">Pietro Lio'</a>, <a href="/mmbronstein/">Michael Bronstein</a>  

arXiv: arxiv.org/abs/2103.03212
Google DeepMind (@googledeepmind) 's Twitter Profile Photo

Today with EMBL-EBI, we're launching the #AlphaFold Protein Structure Database, which offers the most complete and accurate picture of the human proteome, doubling humanity’s accumulated knowledge of high-accuracy human protein structures - for free: dpmd.ai/alphafolddb 1/

Sepp Hochreiter (@hochreitersepp) 's Twitter Profile Photo

ArXiv arxiv.org/abs/2112.09788: Denoising score matching / Langevin dynamics with generalized normal distribution (eg Laplacian, heavy-tailed) for noising and denoising. Iterative noise scaling algorithm to handle multiple levels of noise in annealed Langevin dynamics.