Nazari Soheila, Keyanfar Alireza, Van Hulle Marc M
Faculty of Electrical Engineering, Shahid Beheshti University, Tehran, Iran.
Department of Neurosciences, Laboratory for Neuro- and Psychophysiology, KU Leuven - University of Leuven, 3000 Leuven, Belgium.
Cogn Neurodyn. 2023 Dec;17(6):1649-1660. doi: 10.1007/s11571-022-09917-9. Epub 2022 Dec 19.
McCulloch and Pitts hypothesized in 1943 that the brain is entirely composed of logic gates, akin to current computers' IP cores, which led to several neural analogs of Boolean logic. The current study proposes a spiking image processing unit (SIPU) based on spiking frequency gates and coordinate logic operations, as a dynamical model of synapses and spiking neurons. SIPU can imitate DSP functions like edge recognition, picture magnification, noise reduction, etc. but can be extended to cater for more advanced computing tasks. The proposed spiking Boolean logic platform can be used to develop advanced applications without relying on learning or specialized datasets. It could aid in gaining a deeper understanding of complex brain functions and spur new forms of neural analogs.
1943年,麦卡洛克和皮茨提出假设,认为大脑完全由逻辑门组成,类似于当前计算机的IP核,这导致了布尔逻辑的几种神经模拟。当前的研究提出了一种基于脉冲频率门和坐标逻辑运算的脉冲图像处理单元(SIPU),作为突触和脉冲神经元的动力学模型。SIPU可以模仿诸如边缘识别、图像放大、降噪等DSP功能,但可以扩展以满足更高级的计算任务。所提出的脉冲布尔逻辑平台可用于开发高级应用,而无需依赖学习或专门的数据集。它有助于更深入地理解复杂的大脑功能,并催生新形式的神经模拟。