I lead a team of computer scientists in a French web company (SmartAdServer). We use machine learning and other mathematical tools in everyday life. Most problems we study do not have a solution out of the box, so we use unusual techniques very often, sometimes needing to adapt the classical ideas without a perfect theoretical ground but only empirical validation.
I first studied abstract mathematics (logics, set theory and model theory) and learnt many methods in applied mathematics for a few years. I still think more like a general mathematician than like a machine learning expert (that I'm not). I'm also a developer mainly interested in parallelism, low level optimization and performance: C++, C#, Java.
I try to focus truth and rigour mixed with practical intuition.
You can contact me for professional collaboration at "bsanchez at smartadserver dot com"
Member for 3 years, 7 months
1 profile view
Last seen Mar 26 '18 at 16:01
Top network posts
- 46 What is a manifold?
- 28 Is ridge regression useless in high dimensions ($n \ll p$)? How can OLS fail to overfit?
- 20 Avoid overfitting in regression: alternatives to regularization
- 18 Relationships between correlation and causation
- 17 Is unbiased maximum likelihood estimator always the best unbiased estimator?
- 14 Understanding negative ridge regression
- 14 How does batch size affect convergence of SGD and why?
- View more network posts →