Place: Room Isabel Rodriguez (Room 215)
The impressive advancements in artificial intelligence that we all witness are software-based, biologically inspired implementations. Artificial neural networks (a tiny part of machine learning) are, in turn, one of the most attractive concepts behind these astonishing achievements. It is therefore counterintuitive that a purely analog and spike-based mathematical approach is materially implemented using the same CMOS gates early envisioned and masterly improved to implement fully
digital Boolean logic. Being able to keep pace with increasing the processing capacity depends on continuing to increase the number of CMOS gates (reducing each transistor size until physically prohibitive limits and/or dealing with higher volume and power consumption) or seeking new computational strategies.
In-materio signal processing is an attempt to project into hardware some of the computational operations currently implemented in software. It requires exploring alternative substrates to CMOSbased ones capable of computing. In this framework, although it is far from clear what this means, one aspect reveals the key: if complexity is desired, scale is important. Top-down assembly strategies in which trillions of structures must be individually defined and interconnected tens of thousands of times, is a difficult task to achieve. Self-assemblies appear as an interesting contender to mitigate such a difficulty. Defined as collective structures that assemble spontaneously, they are comprised of multiple, simple, and imperfect units whose nature depends on the intrinsic properties of the material or substrate of interest. Silver nanowire networks, ferroelectric domain walls, and ferromagnetic domains are examples of these objects. In this talk, I will comment on the recent progress we achieved on the simulation front[1, 2] and on those three different types of experimental self-ssemblies [3, 4, 5].
 F. Di Francesco and col., “Spatiotemporal evolution of resistance state in simulated memristive networks,” Applied Physics Letters, vol. 119, p. 193502, Nov. 2021.
 G. A. Sanca and col., “Collective electrical response of simulated memristive arrays using SPICE,” 2022. 10.48550/arXiv.2201.06984.
 J. I. Diaz Schneider and col., “Resistive Switching of Self-Assembled Silver Nanowire Networks Governed by Environmental Conditions,” Advanced Electronic Materials, vol. 8, no. 11, p. 2200631, 2022.
 J. L. Rieck and col., “Ferroelastic Domain Walls in BiFeO3 as Memristive Networks,” Advanced Intelligent Systems, vol. 5, no. 1, p. 2200292, 2023.
 C. P. Quinteros and col., “Evolution of ferromagnetic stripes in FePt films at low temperature,” 2023. 10.48550/arXiv.2302.09102.