3 dagen geleden - Radboud Universiteit (RU) - Nijmegen
PhD Position in Probabilistic Program Induction
Challenge: Overcome brittleness of artificially intelligent agents
Change: Use probabilistic programs
Impact: Trustworthy intelligent agents in the wild
- Mekelweg, Delft, Zuid-Holland
- Tijdelijk contract / Tijdelijke opdracht
- Uren per week:
- 36 - 40 uur
- € 2395 - € 3061 per maand
Despite the remarkable progress artificial intelligence has made over the decades, it still lacks many features we attribute to human intelligence. Perhaps the most startling deficiency is the flexible use of knowledge humans exhibit. Once they understand a concept, humans can use it in new situations, find it new purposes, generate exemplars, and compositionally assemble new concepts. Artificially intelligent agents, in contrast, are brittle and use knowledge only in narrowly defined contexts. Humans are so adaptive and flexible because they are adept at both learning and reasoning. When new challenges arise, they stay on top of their game by learning new skills and information. By reasoning about core properties of skills, humans transfer them to new challenges, with or without modification. With machine learning and probabilistic and logical reasoning, artificial intelligence has made significant progress in both directions independently. However, what we are missing is a framework that unifies them.
This PhD project will investigate the use of probabilistic programs as a unifying paradigm for learning and reasoning. Probabilistic programs extend programming languages with stochastic execution primitives. This simple idea turns probabilistic programs into a lingua franca for artificially intelligent agents: they support logical reasoning by using programming languages and probabilistic modelling through stochastic execution. While their reasoning capabilities are well understood, learning probabilistic programs from observations is still a major challenge - you will have the opportunity to tackle this problem. We envision that the outcomes of this research will have a big impact on artificial intelligence and its applications.
You will perform groundbreaking research within the Algorithmics group at TU Delft. The Algorithmics group aims to design and understand fundamental properties of planning and coordination algorithms for intelligent decision making in real-world applications, such as for coordinating electrical loads within network constraints, or logistic processes on a shunting yard or container terminal. To realise these objectives, the group focuses on the following research topics: model-based reinforcement learning, multi-party optimisation, constraint reasoning, and meta-heuristics
- You hold an MSc degree or a similar degree with an academic level equivalent to a two-year Master's degree in Computer Science or a closely related field;
- You have a strong background and demonstrable experience in artificial intelligence and machine learning. Background in probabilistic programming is not needed, but desirable;
- You have strong programming and algorithm developments skills;
- You have a strong interest or experience with empirical and theoretical research. More importantly, you enjoy doing research;
- You have good communication skills in English, demonstrated by an MSc thesis or other relevant writing;
● You are a team player
TU Delft offers PhD-candidates a 4-year contract, with an official go/no go progress assessment after one year. Salary and benefits are in accordance with the Collective Labour Agreement for Dutch Universities, increasing from € 2395 per month in the first year to € 3061 in the fourth year. As a PhD candidate you will be enrolled in the TU Delft Graduate School. The TU Delft Graduate School provides an inspiring research environment with an excellent team of supervisors, academic staff and a mentor. The Doctoral Education Programme is aimed at developing your transferable, discipline-related and research skills.
The TU Delft offers a customisable compensation package, discounts on health insurance and sport memberships, and a monthly work costs contribution. Flexible work schedules can be arranged. For international applicants we offer the Coming to Delft Service and Partner Career Advice to assist you with your relocation.
For more information about this vacancy and the application procedure, please contact Sebastijan Dumancic, Assistant Professor (S.Dumancic@tudelft.nl)