Postdoc: Symmetries and their Breaking in Deep Learning

University of Amsterdam; Amsterdam; Ibrido; EUR 41.000 - 64.000; Tempo pieno

Descrizione del lavoro

The University of Amsterdam's AMLab is seeking a researcher with a PhD in machine learning, physics, mathematics, or computer science to study neural networks from a physical perspective and to develop principled deep learning architectures for scientific applications.

You will lead and co-author papers, help shape the research agenda, mentor PhD and MSc students, and build your own research portfolio within a newly formed AMLab group led by Nabil Iqbal, starting Fall 2027.

Competenze

• PhD completed or near completion in ML, CS, physics, mathematics, or related field.
• Strong publication record in major journals or international conferences.
• Strong mathematical and programming skills (e.g. PyTorch or JAX) and ability to run experiments as well as build theoretical frameworks.
• Ability to drive research independently and to mentor PhD and MSc students.
• Professional command of English.

Mansioni

• Lead and co-author publications at top-tier venues and represent the lab at international meetings.
• Help shape the research agenda of the group and co-supervise PhD and MSc students.
• Build your own research portfolio within a new AMLab group.

Conoscenze

Publication record

Programming skills (PyTorch/JAX)

Independent research

Mentoring PhD/MSc students

English proficiency

Formazione

PhD in ML/CS/Physics/Math or related field

Strumenti

PyTorch

JAX

Descrizione del lavoro

Join Us!

Artificial neural networks models are made out of many individually simple constituents which assemble together into emergent entities with remarkable capabilities, including the ability to converse, reason, and accomplish difficult tasks. How can stacks of matrices and non-linearities accomplish such amazing feats? The task of developing a theoretical framework for this emergence – and to then using this understanding to construct more robust, efficient, controllable, and generally better machine learning models – is currently a focus of extremely intense study. Philosophically similar problems involving the emergence of non-trivial collective behavior from simple microscopic ingredients have been studied for decades in the neighbouring field of physics. In this project, you will take inspiration from those ideas – and in particular the general unifying principle of symmetry – to tackle these difficult problems.

This is what you will do

You will build a line of research that is oriented towards understanding the dynamics of modern neural architectures in a mathematically principled manner, and further you will use this new understanding to build new deep learning architectures with an eye towards scientific applications.

The precise directions will be chosen to align with your research interest but some potential ideas include:

• understanding the dynamics of spontaneous symmetry breaking in equivariant neural networks
• working towards a theoretical understanding of scaling laws in deep neural networks, either for language models or in geometric deep learning
• studying the dynamics of generative models using tools drawn from theoretical physics such as non-equilibrium and non-reciprocal thermodynamics, and using this approach to construct new generative paradigms
• applying modern generative techniques to problems of physical interest, such as new ways of sampling for many-body quantum physics

This list of topics is to give you an idea of the interests of the group and is by no means exclusive.

To accomplish these research goals you will:

• lead and co-author publications at top-tier venues and represent the lab at international meetings
• help shape the research agenda of the group, co-supervise PhD and MSc students, and build your own research portfolio.

You will be part of a new group within AMLab that is being started by Nabil Iqbal to study neural networks from a physical point of view.

What we ask of you

• a PhD (completed or near completion) in machine learning, computer science, physics, mathematics, or a related field.
• A strong publication record in major academic journals or international conferences
• Strong mathematical and programming skills (e.g. PyTorch or JAX) and ability to run experiments as well as build theoretical frameworks
• The ability to drive research independently and to mentor PhD and MSc students
• A professional command of English

If you have questions we encourage you to contact Nabil at [e-mail]. The start date is Fall 2027.

This is what we offer you

We offer a temporary employment contract 38 hours per week for a period of 24 months with a probationary period of two months. The preferred starting date is Fall 2027.

The gross monthly salary, based on 38 hours per week and dependent on relevant experience, ranges between €3,706 to € 5,760 (scale 10). This does not include 8% holiday allowance and 8,3% year-end allowance. The UFO profile Researcher 4 is applicable. A favourable tax agreement, the ‘30% ruling’, may apply to non-Dutch applicants. The Collective Labour Agreement of Universities of the Netherlands is applicable.

Curious about our extensive secondary benefits package? Read more about the benefits.

You will work in this team

The Faculty of Science has a student body of around 8,000, as well as 1,800 members of staff working in education, research or support services. Researchers and students at the Faculty of Science are fascinated by every aspect of how the world works, be it elementary particles, the birth of the universe or the functioning of the brain.

You will be part of the Amsterdam Machine Learning Lab (AMLab). AMLab conducts research in machine learning, artificial intelligence, and its applications to large scale data domains in science and industry. This includes the development of deep generative models, methods for approximate inference, probabilistic programming, Bayesian deep learning

Want to know more about our organisation? Read more about working at the University of Amsterdam.