Alessandro Lupo (ULB)
Alessandro’s research project was centered on photonic reservoir computing based on frequency-multiplexing. For its nature, this project was multifaceted: it offered multiple possible paths of exploration, both theoretical and experimental. For Alessandro the POST-DIGITAL network provided him with valuable tools to pursue the research project, as described below.
When Alessandro started his PhD, his research group already had a preliminary experimental demonstration of photonic frequency-multiplexing reservoir computing (FM-RC). This was Alessandro’s first experimental training ground, which afterward led to a scientific publication [1]. Having acquired knowledge about the experimental implementation of frequency-multiplexing computing schemes, he has been able to design and realize a demonstrator for a frequency-multiplexing extreme learning machine (FM-ELM) [2]. As reported in the acknowledgments section of the article [2], this experiment benefited from the contribution of the network, both conceptually (contribution to the conception of the experiment) and practically (loaning of an experimental instrument).
The work on photonic frequency-multiplexing neuromorphic computing led to multiple research avenues, such as parallel computation [3], deep configurations [4] and computing based on optical solitons (preliminary results obtained, but not yet published).
One part of Alessandro’s project concerned the design and realization of photonic integrated chips, mainly constituting an output layer for the frequency-multiplexing computer schemes. When Alessandro started his PhD, the group at ULB lacked the technical capabilities to design and produce integrated photonic chips, but this has been entirely compensated by the collaborations within the POST-DIGITAL network. Alessandro spend time at VLC Photonics (Valencia, Spain) on secondment, where he acquired the knowledge necessary to design the chips, and at UGent (Ghent, Belgium), where he acquired the knowledge in testing photonic chips in high-speed setups. POST-DIGITAL network training events on photonic integration directly led to the design and production of two integrated photonic chips, with initial characterization gave a positive outlook, as reported in [5].
Key publications by Alessandro Lupo:
[1] Butschek, L., Akrout, A., Dimitriadou, E., Lupo, A., Haelterman, M., & Massar, S. (2022). Photonic reservoir computer based on frequency multiplexing. Optics Letters, 47(4), 782-785.
[2] Lupo, A., Butschek, L., & Massar, S. (2021). Photonic extreme learning machine based on frequency multiplexing. Optics express, 29(18), 28257-28276.
[3] Lupo, A., & Massar, S. (2021). Parallel extreme learning machines based on frequency multiplexing. Applied Sciences, 12(1), 214.
[4] Lupo, A., Picco, E., Zajnulina, M., & Massar, S. (2023). Fully analog photonic deep Reservoir Computer based on frequency multiplexing. arXiv preprint arXiv:2305.08892.
[5] Jonuzi, T., Lupo, A., Soriano, M. C., Massar, S., & Domenéch, J. D. (2023). Integrated programmable spectral filter for frequency-multiplexed neuromorphic computers. Optics Express, 31(12), 19255-19265.