Invited Lectures


Sweet Rupture

Laurent Sarrazin

Société Générale (F)

A good break, it does not hurt, at least if it is smooth, or sweet!
Becoming agile requires courage and discipline, it’s not a scoop, but do we realize that agile transformation often begins with a break, especially within large organizations operating in an opposite ‘cultural legacy’.
How to become agile in full consciousness, and to ensure that the break is accepted to become an accelerator? How to motivate the troops in a sustainable manner, and disseminate the agility at scale?
Laurent presents “Sweet Rupture”, and you will follow the tracks of a large-scale agile transformation, driven by a genuine fusion of ingredients more or less conventional. You will discover the “Fearless Change”, the “Radical Management”, “A Playful Growth”, the “Assholes Jar “, the “Mental Vitamins”, the “Sociocracy”, and even a dose of “Tribal Leadership”.  And this “pragmatic cocktail”, spiced by the curve of diffusion of innovations of Everett W. Rogers, will surprise you with its efficiency, humor and fun, or even the desire it provides.

Telling Machines about the World:

From START to Watson and Beyond 

Boris Katz

Massachusetts Institute of Technology

Computer Science and Artificial Intelligence Laboratory (CSAIL)

The Stata Center

32 Vassar Street

Cambridge, MA 02139, USA

It has been argued that question answering can be used as a test that might help us define what it means for a computer to exhibit intelligence.  In this talk, I will describe and demonstrate natural language tools which make it possible to:

  • · provide machines with new knowledge: given a set of natural language assertions, create a knowledge base of machine-readable structures representing the input
  • · explain computer actions: given a fragment of the knowledge base, generate a natural language sentence
  • · test computer understanding: given a query, match the representation of the query against the knowledge base and generate an answer or execute an action


These tools have been created as part of a larger effort aimed at modeling human intelligence, including narrative understanding, and scene and activity recognition.  The tools are in daily use by several research groups at MIT for tasks such as providing machines with new knowledge and querying them about it.  Computers draw on this knowledge to reason, learn, and explain their actions and conclusions.