How to build a new kind of information processing. You will no longer need programming languages as we know them.
During my PHD Thesis (2003-2007), supervised by Pr. Alain Cardon in LIP6 from P6, I developed a system based on an ontology of making self-assessment decisions, that allows one to generate models in real-time. This discovery is as a master algorithm to control machines at a knowledge level (A. Newell). It is a possible core to the development of an artificial consciousness.
I address the artificial consciousness problem with a philosophical and constructivist approach. The goal is not to copy the human nervous system or implement a specific cognitive model but to work on this question: what are humans capable of doing thanks to their consciousness? The main goal is to build a system capable of generating questions itself, and behaving according to its assumptions in the real world. To resume, build an artificial consciousness, not a Human artificial consciousness.
Why to build an artificial consciousness ? We have cognitive system services thanks to IBM and Microsoft, we have deep learning tools thanks to Google and a batch of brillant scientists. But these discoveries don't solve the fact that a human is able to transposing a solution of a problem to a different problem, a human doesn't need to learn 30 millions of patterns to do something. That is to say there is a way to use less energy to solve problems and there is a way to fully control in real-time what an autonomous system does and what will it do instead of a black box like a deep neural network.
26 papers including 3 re-editions in journals or books.
Journals with blind review
Conferences with blind peer review
Workshops with blind peer review
PHD thesis and Misc
I created and developed, with great teams, some technologies, this is a selection..
A robot in the cloud (2011-2015)
Daneel is a cognitive system built with a goal-oriented organizational memory. The entire mechanism of this system follows the laws of an auto-adaptive system. These laws give plasticity and dynamism to the intelligent system. These features are crucial for managing different abstraction layers in a symbolic system. Multiple contexts and different meanings are taken in account when users discuss with Daneel. Whatever the user is asking, Daneel uses its own knowledge to answer or to discuss with the user using a natural language. We are in an industrial transition: in this new era, developers will program some behaviour, which will integrate in a cognitive system, instead of programming some applications put in a global market. That’s mean that data are merged with algorithms: we work at the knowledge level.
Create some crazy projects and work with awesome students (2002-2008)
Courses: Artificial Intelligence (expert systems, neural networks, multi-agent system, genetic algorithm and game theory), Concurrent Programming and Kernel Programming.
Manage birds birth in real-time (1998-2001)
The AIS is an Artificial Incubation Software to manage birth of multiple species in real-time. Plugged to a calliper, a balance and an ontology, this software allowed the expert to control the birth rate of multiple species.