The general policy of the institute regarding partnerships is the following. We generally do not have contracts involving exclusivity constraints, that are too application-specific and end up tying our innovations to IP belonging to a specific company. What we do, both in terms of code and algorithmic ideas, is put them in the public domain. In other words, we do not write patents for anything we do, and we publish our research.
The advantages for industrial partners are that:
- They get privileged access to the lab researchers, including students (who often end up being hired by one of the partners, either for internships or jobs)
- They get privileged access to the outputs of the lab researchers (e.g. even before stuff get published, we are open about what we do, along the way)
- Their engineers can learn from our researchers (sometimes through visits or collaborations)
- Our students can focus on tasks of interest to them and we can orient our research in directions that are both interesting scientifically for us and interesting for dealing with potential obstacles that our partners face
- Some of our staff can do consulting for the partner
- This kind of sponsorship can be good for visibility / publicity from the point of view of the partner, given that we are the biggest academic lab in the deep learning area and the other senior researchers have gone into exclusive deals with big companies.
This policy allows us to avoid having research being done in isolated silos associated with each industrial partner: information flows freely throughout the lab.