A work developed at the IUMA, first national prize for the design of electronic systems

The Final Degree Project entitled “Design of an FPGA hardware accelerator for machine learning using a virtual platform” by student Mario Guanche Hernández and developed at the University Institute of Applied Microelectronics (IUMA) has been awarded first prize in the XV edition of the Future of Telecommunications Awards, in the category for the best Final Degree Projects carried out by Graduates in Telecommunications Engineering in Spain in 2021. This Final Degree Project has been developed under the direction of Professor Dr. Pedro Pérez Carballo of the SICAD Division and the researcher Ms. Sonia Raquel León Martín of the DSI Division.

The work is the result of an extensive study on different implementation techniques of hardware accelerators for artificial intelligence algorithms based on Machine Learning. Latest generation FPGA devices have been used, which have heterogeneous architectures with multiprocessing capacity (MPSoC). These techniques, which were applied for the detection of skin cancer using hyperspectral image processing techniques, drastically reduce processing and classification time. This has been possible thanks to the study of advanced analysis and design methodologies, as well as design and prototyping tools provided by the IUMA.

The results show that an acceleration factor of x189 is achieved for the computation of the images obtained, obtaining their classification of the different types presented in less than 10 milliseconds. This work has highlighted the need for collaboration between different disciplines to find efficient solutions to real problems, as has been done at the IUMA.

The results achieved culminate various student training actions under coordinated direction, including having obtained a scholarship for introduction to research from the Ministry of Education and Vocational Training and the completion of Business Practices at the IUMA during the studies of Degree. In this period it has been possible to carry out prior training in the methodologies and techniques necessary to be able to undertake quality work like the award-winning one.

Mario has continued his training in these techniques, completing the Master's Degree in Applied Electronics and Telecommunications at the IUMA. This training includes an in-depth study of algorithms and Artificial Intelligence processing techniques, knowledge of advanced programming techniques, reconfigurable and embedded computing systems, among other topics. He is currently working on his Master's Thesis to include new processing algorithms in the designed system, with their corresponding hardware accelerators.

These works follow lines of research consolidated at the IUMA: the use of hyperspectral images in the field of medicine, in which the researcher Sonia Raquel León Martín carries out her doctoral thesis, as well as the development of hardware accelerators based on FPGA using methodologies of high-level design, in which Professor Dr. Pedro Pérez Carballo develops his research.

For more information:

https://www.ulpgc.es/noticia/2022/03/16/cuatro-egresados-eite-premios-nacionales-xv-edicion-premios-futuro
https://www.telecos.zone/index.php/actualidad/noticias-coitt/909-estos-son-los-ganadores-de-los-premios-coitt-2021-futuro-de-las-telecomunicaciones