Gael Varoquaux   @GaelVaroquaux

Research & code: Research director @inria ►Data, Health, & Computer science ►Python coder, (co)founder of @scikit_learn & joblib ►Art on @artgael ►Physics PhD





  Tweets by Gael Varoquaux  

Gael Varoquaux    @GaelVaroquaux     10/18/2021
Quantifying mental health? In a new @GigaScience paper, we develop the notion of population-derived proxy measures, built with machine learning to extract indicators of aging, intelligence, or personality combining brain imaging and sociodemographics 馃憞1/ https://t.co/mqkUbYTbXR
 Reply      Retweet   13      Like     20    

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     9/29/2021
This problem is broader than our team. Only 17% of the attendees of the free MOOC on machine learning with scikit-learn were women. Machine learning and data science are great career paths, where women perfectly belong.
 Reply      Retweet   3      Like     20    

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/18/2021
We found that brain imaging is much less important than socialdemographics, to model psychological traits, an important result for population-imaging research. This result has personally pushed me to focus more on non-imaging data. 5/
 
 Reply      Retweet   2      Like     11    

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/5/2021
We conjecture that it is best to jointly optimize an imputation function (often close to conditional imputation) and a regression on top. A neural architecture integrating NeuMiss (as supervised differentiable imputation) gives such chained procedure https://t.co/9RaYmohUZE 5/6
 Reply      Retweet   1      Like     2    

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/12/2021
Rather, in some settings, reweighting samples (importance weighting) during extraction of the predictive model can create a model suited for the shifted distribution. 4/7
 
 Reply      Retweet   1      Like     1    

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     9/30/2021
I want to acknowledge the challenges to women's career, the legitimate fear of unfairness. I've herd so many stories of everyday sexism, stories whispered, unseen by priviliged I also acknowledge that the current lack of women in the scikit-learn does not send the right signals.
 Reply      Retweet   1      Like     1    

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/13/2021
New release of dirty_cat 鉁: machine learning on dirty categories The big deal: the SuperVectorizer: easily ingest a (possibly dirty) pandas dataframe in a machine-learning pipeline 馃 https://t.co/mljxVwtnzG Painless data science directly on the dataframe #pydata
 
 Reply      Retweet   2      Like        

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/12/2021
The problem of dataset shift is that the data on which we apply a predictive model (eg biomarker) differs from the data from which the predictive model was learned. The problem of external validity in #epitwitter 2/7
 
 Reply      Retweet   1      Like        

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/5/2021
Experiments show that: - Optimal conditional imputation & fully-observed predictor (chaining oracles)馃憥 - Oracle conditional imputation +MLP馃憤 - Joint optimization (NeuMiss + MLP) gives best learners (outperforming XGboost, MICE...) 馃憤馃憤 in MAR and MNAR missingness mechanism 6/6
 
 Reply      Retweet   1      Like        

More by:   Gael Varoquaux
zoom in  
 



Gael Varoquaux    @GaelVaroquaux     10/18/2021
These measures build upon validated measures (such as age, or a neuroticism assessment) but population modeling puts them in a broader context: brain age captures how a person's brain relates to their age, sociodemographics are central to interpreting neuroticism or IQ scores 2/
 
 Reply      Retweet        Like     3    

More by:   Gael Varoquaux
zoom in  
 









  More from:  


Sebastian Raschka
Author of the 'Python Machine Learning' book. Tweet about Python, deep learning research, open source. Asst Prof of Statistics @UWMadison. Opinions are my own.

Yann LeCun
Professor at NYU. Chief AI Scientist at Facebook. Researcher in AI, Machine Learning, etc. ACM Turing Award Laureate.

Stanford HAI
Advancing AI research, education, policy, and practice to improve the human condition.

Facebook AI
Facebook AI focuses on bringing the world together by advancing AI, powering meaningful and safe experiences, and conducting open research.

Gabriel Peyré
@CNRS researcher at @ENS_ULM. One tweet a day on computational mathematics.

Petar Veličković
Staff Research Scientist @DeepMind | Affiliated Lecturer @Cambridge_Uni. PhD @TrinCollCam. Ex-@Mila_Quebec @BellLabs. GDL, GNNs, Algorithms. 🇷🇸🇲🇪🇧🇦