Master thesis: Adapting Neural Networks to New Tasks

Join our Team

About this opportunity:

In the Research Area Artificial Intelligence, which is part of Ericsson Research, we are pushing the technology frontiers in AI, combining machine learning and reasoning methods, tools, and techniques to drive intelligent autonomous operations in large complex telecom systems.

We are now looking for a talented and motivated student to join us for a study on adaptation of neural network architectures for changing machine-learning tasks.

In the telecommunication world, machine-learning problems are dynamic. Frequent changes in network conditions, available data, and operational objectives create distributional shifts, evolving feature spaces, and entirely new learning tasks. The neural network architecture tailored for an initial state of the problem may therefore become suboptimal over time, leading to degraded performance and increased maintenance costs.

In this thesis, you will investigate whether model architecture adaptation can maintain and improve the model performance over time. You will leverage and further refine ideas from Continual Learning and Neural Architecture Search to develop more flexible and resilient machine-learning systems.

What you will do:

• Conduct literature review in the areas of continual learning, neural architecture search, and machine-learning in telecommunications

• Evaluate existing solutions in the area to establish a baseline

• Develop and evaluate your own approach for adapting neural architectures for improved model performance

• Collaborate with your supervisor to define research directions

• Collaborate with research team to ensure technical feasibility

• Present findings through regular discussions and final thesis documentation

The skills you bring:

• Master’s student with most courses completed and strong academic performance

• Knowledge of continual learning, deep learning, or neural architecture search is a plus

• Proficiency in Python programming

• Hands-on experience with machine learning frameworks such as PyTorch

• Strong programming, debugging, and problem-solving skills, including effective use of generative AI development tools

• Excellent communication skills

• Excellent written and spoken English, ability to work as part of an international team

• Knowledge in telecommunication networks

As the work is research-oriented, we expect an analytical mindset, the ability to learn quickly, work independently, and identify problems and solutions.

What we offer

The project will be conducted in Kista during Spring 2027, as part of our activities at Ericsson Research. You will be offered mentorship from experienced researchers at Ericsson Research, access to industry tools and datasets, and an opportunity to support Ericsson initiatives towards an AI-native network. Further, outstanding results may contribute to scientific publications, patent applications, and future Ericsson research activities.