Two 24-months post-doctoral positions concerned with the application of Machine Learning methods to Search and Optimization are opened. The positions will be held at the Microsoft-INRIA joint laboratory at Saclay, (20km south of Paris), France.

Topic

The context of the work is to learn efficient solving strategies on top of existing Constraint-based or Meta-heuristics optimization algorithms.
A key challenge today for Constraint solvers is to be able to reformulate an existing CSP efficiently until the problem becomes obviously solvable or unsolvable. One possible topic of the post-doc will be to turn the reformulation of a given CSP into a reinforcement learning problem: define the state and actions spaces, and investigate the feasibility of learning a good reformulation strategy.
Another possible research direction is to identify the context of competence of specific heuristics within the search, and to use the gained knowledge to define heuristics selection strategies.

Profile

We are looking for PhD computer scientists with strong knowledge in at least one the following fields: Machine Learning and Data Mining (Reinforcement Learning, Monte-Carlo Tree Search, ...), Constraint Solving (exact and stochastic Optimization, Meta-Heuristics, ...). Strong programming skills are mandatory. A sound experience of existing CSP solvers will be appreciated.

Duration and remuneration

• 12+12 months contract, starting from September 2011.
• net salary: approx. 2100 Euros per month (health insurance included).

Application:

Please send your application (PDF format) as soon as possible.
Screening of applications starts immediately and continues until the position is filled.
Send cover letter including names of at least two references, CV and links to PhD dissertation (or draft) and up to three most relevant publications to the following emails:

• youssefh[at]microsoft[dot]com
• marc[dot]schoenauer[at]inria[dot]fr
• michele[dot]sebag[at]lri[dot]fr