Research areas on Machine Learning
Sustainable Machine Learning
We aim to balance the potential of Machine Learning with sustainability challenges in the development of AI through resource-efficient solutions.
Machine learning (ML) is a critical component of artificial intelligence (AI) development, allowing for complex tasks to be accomplished by learning from data. However, the energy and resource consumption of digital infrastructure, such as data centers and networks, may lead to adverse environmental effects. It has been estimated that 5-9% of the world’s total electricity production is used by information and communication technology, which may rise to 20% in 2030. For example, the development of large language models has been associated with a doubling External link. of the computing requirements every two months. It is important to balance the benefits of ML with its sustainability challenges and prioritize environmentally and socially responsible AI development. More resource-efficient ML hardware, algorithms and software engineering concepts are required to make the development of AI sustainable and widely accessible for the welfare of society.
Projects
MSc theses
- Andreas Ternstedt (2017), Pattern recognition with spiking neural networks and the ROLLS low-power online learning neuromorphic processor, link to thesis
External link..
- Mattias Nilsson (2018), Monte Carlo Optimization of Neuromorphic Cricket Auditory Feature Detection Circuits in the Dynap-SE Processor, link to thesis
External link.
- Oskar Öberg (2019), Critical Branching Regulation of the E-I Net Spiking Neural Network Model, link to thesis
External link., collaboration
External link.
- Olof Johansson (2021), Training of Object Detection Spiking Neural Networks for Event-Based Vision, link to thesis
External link.
- Kim Petersson Steenari (2022), A neuromorphic approach for edge use allocation, link to thesis
External link., collaboration
External link.
PhD theses
- Mattias Nilsson (2023), Event-Driven Architectures for Heterogeneous Neuromorphic Computing Systems, link to thesis
External link.
Education
- Neural Networks and Learning Machines
- Neuromorphic Computing
Brain Analysis Lab
Brain Analysis is an active area of research in the medical and neuroscience community. Analysing brain data can assist people that are in medical need. For example, with inner speech detection, people with Locked-In-Syndrome (LIS) who are not able to speak have a possibility to communicate with their environment using their inner speech. Patients who suffer from neurological disorders and have limited communication with the real world- if we are able to help them to communicate with the real world, it will be a worthy contribution.
(AI) Innovation in (AI) Education
Teaching Philosophy
Our group has embraced problem-based and active learning as the fundamental principles of the education process. To implement these principles, we have adopted a flipped classroom approach, where live sessions are conducted with active participation and substantial group and plenum discussions. Additionally, we assign projects and challenges in their labs and coursework. The teaching process of the ML group incorporates many pedagogical principles to achieve the goal of effective education.
Pedagogical Principles
Bank of micro-modules: The concept of a "bank of micro-modules" means that the course can be customized for various groups of students by selecting specific micro-modules from a pre-prepared collection. This approach will decrease the load that the teacher needs to spend on preparing individual modules or designing multiple courses for different levels. Each micro-module is made up of a brief recorded video or lecture (lasting between 5 and 10 minutes), a quiz with immediate feedback, a peer-reviewed assignment, and a reflection question to assess knowledge.
Teach-back method: We have created an initial implementation of the teach-back method, in conjunction with the use of study groups. The core of this implementation is a set of key concepts, and related questions that students divide among themselves on a weekly basis in their respective group. Each student then sets out to read more about their concept and explain it to other members of their group during a meeting. We set out to achieve the following goals with this exercise: i) enhance student engagement, and motivation ii) improve student comprehension and retention on key concepts, while iii) keeping the teacher hours at a reasonable level. This approach was adopted partly in response to evidence demonstrating the limitations of instructor-led explanations for promoting conceptual understanding, as well as to comply with the university's guidelines. We are currently in the process of improving the implementation.
Inclusion as intention and practice together: Inclusive education is the effort to ensure that everyone has equal access to education. The approach taken by LTU acknowledges that there may be conflicts when inclusive education principles clash with the values and traditions of the education system. This means that they have made significant efforts to address academic structures, traditions, and practices that contribute to the marginalization and exclusion of students. Inclusive education is not only focused on special needs students but aims to benefit all students who are at risk of being marginalized.
Major Contribution
Master Programme in Applied AI
Master Programme in Engineering Physics and Electrical Engineering
MOOCs: Production of 2 MOOC courses
– Introduction to artificial intelligence in tourism (link External link.)
– Computer vision, Image understanding for efficient business and industry (link External link.)
Universeh (EU funding- European Space University for Earth and Humanity)
– Joint course with AGH (Machine learning in robotics and edge devices for space) (link External link.)
Modularization – Life-long learning
– Lifelong learning, Practical introduction to data science
EU project- AI4EDU (AI conversational assistant for students) (link)
Joint course with Orebro university: Natural language processing
Updated:
Page author: Personnel