Federated Learning (FL) is a distributed machine learning paradigm used for decentralized training on a large number of endpoints. Each end-device stores data locally and collaboratively learns a shared predictive model. The security of user data and the unlimited number of devices involved are the main advantages of this approach. FL can be used in healthcare, autonomous driving, and IoT systems where the number of connected devices varies from thousands to millions.
AI Accelerator Architecture
In recent years, artificial intelligence (AI) and machine learning techniques have been used in a variety of technologies such as communications, autonomous driving, and smart industry. Current GPU platforms are not suitable for low-power applications, such as edge applications. To enable faster and more power-efficient processing of AI workloads, a new accelerator architecture is required. FPGA are the most promising hardware platforms due to their flexibility and parallelization capabilities.
|Deployment of deep neural networks onto embedded devices for processing radar data||Bachelorarbeit||ab 03 / 2023|
Topko, B.; Topko, Y.; Khabarov, S.; Zamyatin, N.; Zubarev, E.
2022. IEEE Transactions on Nuclear Science, 69 (1), 98–104. doi:10.1109/TNS.2021.3136944
Topko, Y.; Khabarov, S.; Topko, B.; Kovalev, Y.; Zamyatin, N.; Tarasov, O.; Zubarev, E.
2022. IEEE Transactions on Nuclear Science, 69 (3), 634–638. doi:10.1109/TNS.2022.3150753
Topko, Y.; Topko, B.; Khabarov, S.; Zamyatin, N.
2022. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 1033, Artkl.Nr.: 166680. doi:10.1016/j.nima.2022.166680
BM@N Collaboration; Patsyuk, M.; Kahlbow, J.; Laskaris, G.; Duer, M.; Lenivenko, V.; Segarra, E. P.; Atovullaev, T.; Johansson, G.; Aumann, T.; Corsi, A.; Hen, O.; Kapishin, M.; Panin, V.; Piasetzky, E.; Abraamyan, K.; Afanasiev, S.; Agakishiev, G.; Alekseev, P.; Ivanova, Y.; et al.
2021. Nature Physics, 17 (6), 693–699. doi:10.1038/s41567-021-01193-4