Pinned Tweet
Are you interested in causality and image counterfactuals? What makes a counterfactual a counterfactual? Check out our ICLR 2023 paper: Measuring axiomatic soundness of counterfactual image models. (1/6)
openreview.net/forum?id=lZOUQ…
openreview.net/forum?id=lZOUQ…
1
8
58
10,143
Our paper "Measuring axiomatic soundness of counterfactual image models" is now also available on arXiv.
with: @pwnic @dccastr0 @GlockerBen and Fabio De Sousa Ribeiro
arxiv.org/abs/2303.01274
with: @pwnic @dccastr0 @GlockerBen and Fabio De Sousa Ribeiro
arxiv.org/abs/2303.01274
0
4
13
1,226
Please check out our paper for a deeper explanation of these concepts and more experiments and results!
P.S. I have finished my PhD and am currently looking for work opportunities. (6/6)
openreview.net/forum?id=lZOUQ…
P.S. I have finished my PhD and am currently looking for work opportunities. (6/6)
openreview.net/forum?id=lZOUQ…
1
2
3
664
Miguel Monteiro retweeted
This year, the #MLCN2022 best paper is awarded to Margherita Rosnati @MRosnati for “Automatic Lesion Analysis for Increased Efficiency in Outcome Prediction of Traumatic Brain Injury”.
mlcnws.com/2022/10/05/the…
mlcnws.com/2022/10/05/the…
1
3
10
Miguel Monteiro retweeted
Our work on TBI lesion segmentation won Best Paper Award at @MLCNworkshop!
Thank you to all my co-authors for the hard work!
Thank you to all my co-authors for the hard work!
2
2
14
Miguel Monteiro retweeted
New output from @CenterTBI in @EBioMedicine - Early biomarker concentration after TBI is primarily related to severity of injury and intracranial disease burden, rather than pathoanatomical type of injury. #TBI #Biomarkers authors.elsevier.com/sd/article/S23…
1
4
14
Miguel Monteiro retweeted
Our new study demonstrates noisy labels impacts training, eval & selection of ML models. Reannotating large datasets requires lots of expert time (eg in healthcare), so we also show data-driven strategies enable fixing more labels using fewer resources: arxiv.org/abs/2109.00574
2
7
42
Miguel Monteiro retweeted
We are delighted to have @mabmonteiro as today's speaker. The talk will start at 14:30 CET and will be about their #NeurIPS2020 paper on uncertainty modeling in medical image segmentation. Zoom link and more information: bmic-student-conference.github.io
1
3
6
Miguel Monteiro retweeted
Our #CVPR2021 paper about "Unsupervised Human Pose Estimation through Transforming Shape Templates" took us several iterations to get to this point together with @vlontzos , @Kings_PFIG and @BernhardKainz1 .
arxiv: arxiv.org/abs/2105.04154
website: infantmotion.github.io
👇
arxiv: arxiv.org/abs/2105.04154
website: infantmotion.github.io
👇
2
6
36
Miguel Monteiro retweeted
We'll be presenting our work on building deep structural causal models tomorrow, 09.12., at the poster session 5pm GMT / 12 pm EST @NeurIPSConf.
Join us for discussions live or just watch the video and find us on chat:
neurips.cc/virtual/2020/p…
@dccastr0 @GlockerBen
#NeurIPS2020
Join us for discussions live or just watch the video and find us on chat:
neurips.cc/virtual/2020/p…
@dccastr0 @GlockerBen
#NeurIPS2020
Happy to announce that DSCMs got accepted to #NeurIPS2020! Finally unlocked that milestone of getting a paper into @NeurIPSConf or any other conference for that matter! :D
1
4
32
I will be presenting our work "Stochastic Segmentation Networks" at @NeurIPSConf today at 5pm GMT / 12pm EST. Join the virtual poster session or watch the 3min video summary with link bellow: neurips.cc/virtual/2020/p…
Interested in modelling spatially correlated uncertainty? Our new lightweight, flexible, architecture-independent method can generate multiple plausible label-maps with a single forward pass. Ideal for image segmentation and other image-to-image tasks!
arxiv.org/abs/2006.06015
1
3
29
Miguel Monteiro retweeted
A 2.5 hour tutorial on gradients and integration from the lens of Machine Learning and Middle Earth. Say no more. Thanks @mpd37 & Cheng
This may be my day of #NeurIPS2020
slideslive.com/38935796/there…
This may be my day of #NeurIPS2020
slideslive.com/38935796/there…
2
20
86
Miguel Monteiro retweeted
We needed a bit to write up what we did for medicalood.dkfz.de/web/ and summarized our humble approach: here arxiv.org/abs/2011.04197 . Source code is available: github.com/jemtan/FPI
Looking forward to see results with other use case scenarios as well!
Looking forward to see results with other use case scenarios as well!
We won most of the tasks in the @MICCAI2020 MOOD challenge zenodo.org/record/3961376… with a good margin. Team FTI was led by the absolutely incredible Jeremy Tan @BioMedIAICL @ICComputing.
More details on the method will follow (as soon as we can write this up in a proper paper😄)
1
1
14
Miguel Monteiro retweeted
now hear me out. we tried statistics and it let us down. time to use deep learning to predict elections.
76
95
1,268
Miguel Monteiro retweeted
Excited to share our recent work on predicting cortical metrics across a range of brain development stages and pathologies from MRI using #MachineLearning #DeepLearning #AI. Presented at @MLCNworkshop @MICCAI2020 #MLCN2020 #MICCAI2020 arxiv.org/abs/2010.01669
1
5
14