Thibault Doutre

Thibault Doutre

New York, New York, United States
5K followers 500+ connections

Activity

Join now to see all activity

Experience

  • Point72 Graphic

    Point72

    New York, United States

  • -

    New York, United States

  • -

    San Francisco

  • -

    Berkeley

  • -

    Paris Area, France

  • -

    Paris Area, France

  • -

    Paris Area, France

Education

Volunteer Experience

  • Camara Education Ltd. Graphic

    Volunteer

    Camara Education Ltd.

    - 3 months

    Poverty Alleviation

    • Repaired the hardware and software of unwanted corporate computers in London so they could be donated to schools in Africa.

  • President of the Swim Team

    Ecole Centrale Paris

    - 10 months

    • Led a team of 50 swimmers
    • Organized swimming competitions in multiple engineering schools in France

Courses

  • Advanced Probability

    STAT 201A

  • Advanced Statistics

    STAT 201A

  • Capstone Projects

    STAT 222

  • Convex Optimization

    -

  • Decision Aids

    -

  • Discrete Optimization

    -

  • Learning and Decision Making

    CS 281B

  • Linear Models

    STAT 230

  • Machine Learning

    -

  • Non Parametric Methods in Statistics

    -

  • Probability

    -

  • Statistical Computing

    STAT 243

  • Statistical Learning Theory

    CS 281A

  • Statistics

    -

Projects

  • Adaptive rejection sampling for Gibbs Sampling

    - Built an R-package used for adaptive-rejection sampling, described in Gilks et al. (1992)

    See project
  • An Approximate Algorithm for Sparse Hidden Network Inference on Censored Non-Gaussian Observations

    - Present

    - Created an algorithm coded in R to analyze gene expression data and discover when the expression of one gene affected the expression of another gene in brain tissue.
    - Designed a latent-variable graphical model to fit the discrete gene expression data through an EM approach.
    - Used the Expectation Propagation framework (Minka, 2001) along with a Gaussian Kernel approximation to approximate the term likelihoods.

    Abstract:
    We propose a novel algorithm for inference on graphical…

    - Created an algorithm coded in R to analyze gene expression data and discover when the expression of one gene affected the expression of another gene in brain tissue.
    - Designed a latent-variable graphical model to fit the discrete gene expression data through an EM approach.
    - Used the Expectation Propagation framework (Minka, 2001) along with a Gaussian Kernel approximation to approximate the term likelihoods.

    Abstract:
    We propose a novel algorithm for inference on graphical models with sparse multivariate Gaussian latent structure, but non-Gaussian observations. We extend recent ideas in graph- ical model structure learning to the latent variable case through an EM approach, allowing for the use of more realistic and non-Gaussian statistical models. Our most significant contribution lies in the E-step, where we make use of the Expectation Propagation framework (Minka, 2001) along with a Gaussian Kernel approximation to the term likelihoods. The M- step then reduces to the familiar problem of the estimation of a sparse inverse covariance matrix, explored in great depth in the Machine Learning literature. Due to the latent variable structure, we are able to perform inference on complex models where standard Gaussian structure learning fails, and are able to scale to larger data sets than MCMC (Markov Chain Monte Carlo) approaches.

  • Analysis of mouse lifestyles

    -

    - Worked in a group of 26 people on a mouse behavioral data set from the Tecott Lab at UCSF.
    - Collaborated on building a python package using issues and pull requests on GitHub.
    - Revealed a set of fundamental principles of behavioral organization
    - Classified mice by genotype using behavioral features

    See project
  • Analysis of a wireless sensor network

    -


    - Performed data Analysis on wireless sensors designed to capture spatial and temporal information (e.g., temperature, humidity) of California’s trees in Redwoods.
    - Performed data cleaning on raw data from sensors.
    - Presented new findings using data visualization tools in R.

    See project
  • Optimization of parallel parking

    -

    - Led individual research in Lie Algebra, flows and Partial Differential Equations
    - Observed an analogy between Lie brackets of vector fields and movements of parallel-parking
    - Presented work to a jury of professors and professionals

Honors & Awards

  • UC Berekeley Fellowship

    Department of Statistics

  • French Academy Excellence Award

    Académie Française

  • Ecole Centrale Paris Excellence Prize

    Centrale Paris

Languages

  • English

    Native or bilingual proficiency

  • French

    Native or bilingual proficiency

  • Spanish

    Elementary proficiency

  • Chinese

    Elementary proficiency

Organizations

  • Centrale Swim Club

    President of the school Swim Club

    -
  • Gala Equinoxe - Centrale Paris

    Corporate Relationship

    -
  • Centrale-Supélec Career Fair

    Communication

    -

    One of the biggest career fairs in Europe, bringing together over 200 companies and over 3200 students.

More activity by Thibault

View Thibault’s full profile

  • See who you know in common
  • Get introduced
  • Contact Thibault directly
Join to view full profile

Explore collaborative articles

We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.

Explore More

Add new skills with these courses