C3SAR (ref. PPJIA2023-025) is a research project funded by the University of Granada. Its objective is the creation of new algorithms for Electroencephalography and high-performance systems using efficient and sustainable computing as a bullet point, which will promote energy savings and the fight against climate change. The scope of the project covers the following topics:
- Parallel and distributed computing
- Mathematical prediction models
- Bioinspired algorithms for feature selection
- EEG classification
- Workload balancing strategies
- Electricity price-based computing
- Use of container technology
You can follow our progress on the news page.
Motivation
Energy-efficient computing is crucial given the exponential growth of bioinformatics and the increase in the size of biological datasets, such as EEG signals, which reflect the brain’s electrical activity. EEG data often suffer from a low signal-to-noise ratio, requiring sophisticated methods to extract meaningful information. Furthermore, these signals are non-periodic and non-stationary, demanding models that can accurately capture temporal dynamics. The scarcity of available data, due to the high experimental costs of EEG recording, exacerbates the issue, creating the so-called curse of dimensionality. This problem arises when the high number of features in EEG data overwhelms machine learning models, leading to inefficient processing and poor generalization. To address these challenges, techniques like feature selection are essential, reducing the size of datasets by eliminating redundant information. These approaches not only lower processing costs but also enhance the feasibility of solving complex problems within reasonable timeframes.
From an energy perspective, the project emphasizes the importance of time-energy efficiency, especially as HPC systems increasingly contribute to global electricity consumption. The Information and Communication Technology (ICT) sector already accounts for a significant portion of worldwide energy use, a figure projected to rise dramatically. This growth makes ICT systems a significant contributor to greenhouse gas emissions, raising concerns about their environmental impact. Addressing these concerns requires optimizing energy use in computational tasks, especially in resource-intensive fields like bioinformatics. Advanced HPC platforms with multicore CPUs and GPUs, while powerful, consume vast amounts of energy, highlighting the need for solutions that balance performance and sustainability.
