| Abstract Scope |
Conventional edge computing systems involve physically separated sensors, memory devices, and processors, each utilizing distinct manufacturing technologies. This results in increased delays, energy consumption and cost. In this talk, I will be discussing recent advances in multifunctional memory devices that can sense, store, and compute. Such devices enable more real-time data analysis with lower energy consumption and smaller footprint area. The memory devices I will discuss have been developed based on the mature charge trapping memory or emerging memristor structures, using various stimuli sensitive materials focused on solution-processable 2D materials. As a result, such devices have been able to show in-memory optical sensing with an excellent retention and endurance characteristics. Moreover, such devices have shown the ability to mimic synaptic features and to engage in low power neuromorphic computing processes similar to the human brain. This advancement signifies a departure from traditional computing models, moving towards a more brain-inspired approach. |