Streaming Video of the meetup
Maltem Consulting Group will host and sponsor the networking event after the présentations. Thank you to them. Our meetup group now counts more than 3000 members.
We will have three presentations and two pitches. The slides will be added here before the meetup:
- Presentation Maltem Consulting Group
- Igor Carron, introduction
- Franck Bardol, Intro to ML
- Amine El Helou, The Mathworks (short)
- Chloe-Agathe Azencott, Researcher at the Centre for Computational Biology (CBIO) of Mines ParisTech, Institut Curie and INSERM.
- Large p, small n: Feature selection with few samples in high dimension.
- Motivated by personalized medicine applications, I will discuss the difficulties of doing feature selection in high-dimensional settings, and present recent work on this topic.
- Sebastien Treguer,
- AutoML Challenge, what's that? AutoML, Successes & Challenges ( Bayesian optimisation, Ensemble construction, Sensible allocation of computation)
- Aurore Li, Composing Music with LSTM
- Kiran Varanasi, Max Planck Institut Informatik
Modeling Human Faces: What can machine learning methods learn from Da Vinci ? Da Vinci was probably the first person to study human facial anatomy in a scientific manner. Each human face is unique and shows expressions in its own particular way, which we can identify with remarkable precision. So modeling a virtual human face that can be posed to an arbitrary expression is an extremely hard task, which is typically done in the industry using high quality sensors and a lot of manual labor. Can we use machine learning for achieving comparable quality from videos taken from the web ? I will describe two regression problems that we solve for getting there: (1) mapping person-specific fine scale 3D geometry to a common face model parameterization (2) mapping facial dialogue between two different people speaking two different languages. I will end with some reflections on why considering domain specific knowledge in facial anatomy and psychology might be necessary for machine learning approaches synthesizing virtual faces.
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.