BEGIN:VCALENDAR VERSION:2.0 PRODID:-//linuxsoftware.nz//NONSGML Joyous v1.4//EN BEGIN:VEVENT SUMMARY:Towards Brain-Inspired Computing: Memristor-Based Single Layer Neu ral Networks DTSTART:20231122T160000Z DTEND:20231122T180000Z DTSTAMP:20260622T043842Z UID:54c69cbe-604e-494f-8ce7-80e63e93da26 SEQUENCE:1 CREATED:20231121T180434Z DESCRIPTION:This master thesis delves into the realm of neuromorphic compu ting\, a promising paradigm shift poised to overcome the limitations of tr aditional von Neumann computer architectures in the field of neural networ k algorithms. The focus is squarely placed on the memristor\, a non-volati le memory device that emulates the behavior of synapses in the human brain \, offering a scalable and efficient alternative for computing systems. Th rough rigorous exploration of design\, fabrication\, characterization\, an d simulation aspects\, this dissertation evaluates the practicality of inc orporating memristive devices into neuromorphic computing systems.The rese arch is framed around a multi-faceted approach that includes the innovatio n of tools and computational frameworks\, with a highlight being the devel opment of a pioneering software platform that synergistically integrates t he capabilities of LT-Spice and Python. A characterization software\, &quo t\;PyCharMem"\;\, has been developed to streamline the characterizatio n process\, making it more accessible and scalable. Additionally\, a porta ble probe station was conceived and manufactured\, addressing space constr aints.A total of 19 distinct memristive devices were subjected to electric al characterization\, facilitating the extraction of their critical proper ties. This exhaustive analysis laid the groundwork for their emulation usi ng SPICE models\, navigating the complexities through a genetic optimizati on algorithm tailored to extract intricate parameters. The research culmin ates in the execution of single-layer neural network simulations\, employi ng a gradient descent learning algorithm through a reputable SPICE simulat or\, ensuring a comprehensive and reliable evaluation of sneak paths in me mristor crossbars. LAST-MODIFIED:20231121T180434Z LOCATION:Online URL:http://df.vps.tecnico.ulisboa.pt/pt/eventos/towards-brain-inspired-com puting-memristor-based-single-layer-neural-networks/ X-ALT-DESC;FMTTYPE=text/html:

This master thesis delves into the realm of neuromorphic computing\, a promising paradigm shi ft poised to overcome the limitations of traditional von Neumann computer architectures in the field of neural network algorithms. The focus is squa rely placed on the memristor\, a non-volatile memory device that emulates the behavior of synapses in the human brain\, offering a scalable and effi cient alternative for computing systems. Through rigorous exploration of d esign\, fabrication\, characterization\, and simulation aspects\, this dis sertation evaluates the practicality of incorporating memristive devices i nto neuromorphic computing systems.

The research is framed around a multi-faceted approach that includes the innovation of tools and computational frameworks\, with a highlight being the development of a pioneering software platform that synergistically int egrates the capabilities of LT-Spice and Python. A characterization softwa re\, "\;PyCharMem"\;\, has been developed to streamline the charac terization process\, making it more accessible and scalable. Additionally\ , a portable probe station was conceived and manufactured\, addressing spa ce constraints.

A total of 19 disti nct memristive devices were subjected to electrical characterization\, fac ilitating the extraction of their critical properties. This exhaustive ana lysis laid the groundwork for their emulation using SPICE models\, navigat ing the complexities through a genetic optimization algorithm tailored to extract intricate parameters. The research culminates in the execution of single-layer neural network simulations\, employing a gradient descent lea rning algorithm through a reputable SPICE simulator\, ensuring a comprehen sive and reliable evaluation of sneak paths in memristor crossbars.

END:VEVENT END:VCALENDAR