guest ::
| Home > Publications database > Exploring Quantum Annealing for Neuroscience with the L2L Framework: Network analytics in brain connectomes |
| Home > Publications database > Exploring Quantum Annealing for Neuroscience with the L2L Framework: Network analytics in brain connectomes |
| Master Thesis | FZJ-2025-04155 |
2025
This record in other databases:
Please use a persistent id in citations: doi:10.34734/FZJ-2025-04155
Abstract: Computational Neuroscience is the field of study that seeks to understand the complexity of the brain by analyzing brain images, networks, and simulations of neural systems. Classic computing resources, such as high-performance computing systems, are reaching their limits. Therefore, the exploration of alternative computing resources, such as quantum computing, is recommended for the field of computational neuroscience. Quantum annealing shows great potential for accelerating the resolution of optimization and sampling problems. D-Wave, a pioneering company in the field, has developed the first quantum annealers with industrial applications. These annealers feature a relatively high number of available qubits, a crucial aspect of their functionality. However, the success of this process depends on the careful selection of several key parameters. Futhermore it is a straightforward process for users to submit their problems to the D-Wave cloud platform, where they can be executed. Therefore, the utilization of such a D-Wave annealer is examined for its application in the domain of computational neuroscience. <br>This thesis presents a method for automated hyper-parameter optimization for quantum annealing-related hyper-parameters to perform community detection for brain connectomes. The L2L framework was utilized for this purpose, and the quantum annealer was incorporated at the backend for the optimization. To validate the results, several experiments were conducted. These experiments included the karate club data set and connectivity matrices obtained from a study conducted by the Uniklinik Aachen. L2L’s ability to identify suitable parameters for hyper-parameters was demonstrated, leading to an reliable and reproducible approach to use quantum annealing for community detection. The modularity, which was used to measure the performance of the community detection, was nearly similar for the classical and quantum annealing algorithms. Additionally, a comparison was made of the specific compute times.
|
The record appears in these collections: |
PDF