1. Vacatures
  2. Technische Universiteit Delft (TUD)
  3. PhD position in reliable deep reinforcement learning

Helaas, deze vacature staat inmiddels niet meer online

Kijk gerust verder naar andere vacatures.

PhD position in reliable deep reinforcement learning

Challenge: AI that generalizes reliably to new situations.
Change:     Make neural networks certain about their predictions.
Impact:      Help AI to leap from the simulator to the real world

ongeveer 2 maanden geleden

Arbeidsvoorwaarden

Standplaats:
Mekelweg, Delft, Zuid-Holland
Dienstverband:
Tijdelijk contract / Tijdelijke opdracht
Uren per week:
36 - 40 uur
Salarisindicatie:
€ 2434 - € 3111 per maand
Opleidingsniveau:
WO

Functieomschrijving

Are you interested in machine learning for intelligent decision making? Would you like to overcome fundamental challenges and bridge the gap to physical agents like robots, autonomous cars or automated factories? Create new cutting edge neural network architectures to empirically solve theoretical problems in artificial intelligence? Think about broader implications and practical applications?

Deep reinforcement learning (RL) has been at the core of many recent success stories in AI, in particular for playing strategic games like Go, Chess and StarCraft. Despite those spectacular breakthroughs, RL is rarely used in practice, as the learned control policies are generally not assumed to be reliable enough for deployed robots or autonomous cars. We want to change that!

During your PhD, you will develop new algorithms which generalize to situations that differ significantly from training. This will be possible by controlling a graph neural network's internal epistemic uncertainty, that is, how much the network trusts it's own computations. You will evaluate your work on multi-task RL benchmarks, where the agent learns one policy that is able to solve more than one task. Examples are simulated Mujoco robots or different levels of the same computer game. Your challenge will be to study, propose and empirically verify the properties that improve generalization in these benchmarks. You will be collaborating with other members of the Algorithmics group that work on related projects. There might also be an opportunity to transfer your work to real robots, if you are interested

Functie-eisen

You hold an MSc degree or a similar degree with an academic level equivalent to a two-year master's degree in either Computer Science, Artificial Intelligence, Mathematics or a similar area and have an interest in Deep Reinforcement Learning. You want to know more about fundamental questions in autonomous control and enjoy empirically validating theoretical predictions and hypotheses. You possess a strong background in machine learning, and have good programming and math skills. Familiarity with Python and a deep learning framework like PyTorch or TensorFlow (and potentially links to past projects) are appreciated. Prior research experience, in particular in thematically relevant or particularly interesting projects, is a bonus.

Conditions

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 € 2434 per month in the first year to € 3111 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.

Additional information

For more information about this vacancy, please contact Wendelin Boehmer (j.w.bohmer@tudelft.nl).