PhD position Statistical Physics of Learning in Neural Networks (1.0 FTE)

Applications are invited for a fully funded PhD in the area of the statistical physics based theory of neural networks. The position is part of an NWO-supported project with the title The Role of the Activation Function in Feed-forward Learning Systems (RAFFLES).
The aim of the project is to obtain a thorough theoretical understanding of learning processes in feed-forward neural networks, with emphasis on the role of the hidden unit activation functions. We will systematically investigate layered networks for regression and classification in well-defined model situations, for which we can compare different types of activation functions and training algorithms.

The PhD student will study a variety of models and learning scenarios by using (a) analytical tools from the statistical physics of disordered systems, (b) ODE-based descriptions of the learning dynamics, and (c) Monte Carlo simulations. In particular, the following fundamental questions will be addressed:

• What is the precise influence of the choice of activation functions on the performance of multilayered neural networks?
• To what extent is the training process governed by mathematical properties of the activation functions? Delayed learning due to the necessary hidden unit specialization will be in the center of interest.
• Is it possible to identify families (universality classes) of activation functions that share characteristic properties of the training process?
• Can the theoretical insights be exploited in the design of efficient network architectures and training algorithms?

The successful candidate should:

• hold a Master’s degree or equivalent in theoretical physics, computer science, mathematics, or another relevant field
• have a keen interest in the theory of machine learning and artificial neural networks
• ideally come with a background in statistical and computational physics, or with a basic knowledge and strong interest in these and related areas.

We offer you, following the Collective Labour Agreement for Dutch Universities:

• a salary of € 2,443 gross per month in the first year, up to a maximum of € 3,122 gross per month in the fourth and final year for a full-time
working week
• a holiday allowance of 8% gross annual income and an 8.3% year-end bonus
• a full-time position (1.0 FTE).

The successful candidate will first be offered a temporary position of one year with the option of renewal for another three years. Prolongation of the contract is contingent on sufficient progress in the first year to indicate that a successful completion of the PhD thesis within the next three years is to be expected. A PhD training programme is part of the agreement and the successful candidate will be enrolled in the Graduate School of Science and Engineering.

Founded in 1614, the University of Groningen enjoys an international reputation as a dynamic and innovative institution of higher education offering high-quality teaching and research. Flexible study programmes and academic career opportunities in a wide variety of disciplines encourage the 35,000 students and researchers alike to develop their own individual talents. As one of the best research universities in Europe, the University of Groningen has joined forces with other top universities and networks worldwide to become a truly global centre of knowledge.
Within the Faculty of Science and Engineering, a 4-years PhD position is available at the Bernoulli Institute for Mathematics, Computer Science and Artificial Intelligence with the topic of the statistical physics of learning in neural networks. The candidate will become a member of the Intelligent Systems Group of the Computer Science Department and will work under the supervision of Prof. Dr. Michael Biehl.