J. Rajvel · 2026 · International Academic Journal of Innovative Research
Paper
Neuromorphic computing is a revolutionary method of addressing the rising energy needs of artificial intelligence (AI) systems. The paper explores the application of memristive devices in neuromorphic architectures to improve energy and computing efficiency. The use of memristors, their non-volatile memory nature, and the ability to perform synaptic functioning have great benefits over the conventional semiconductor-based technologies, which consume high amounts of energy. Research’s new memristive-based architecture involves the use of memristive crossbar arrays to implement synaptic weights within artificial neural networks, to solve the shortcomings of traditional AI hardware. This architecture is tested using simulations on a series of benchmark AI tasks, such as image classification and pattern recognition, to determine its effectiveness relative to the traditional CMOS-based systems. In this case, the study findings show a 25% energy-saving without loss in performance, with an accuracy of 98% on standard datasets, equivalent to the current AI hardware solution. Moreover, the architecture has a higher speed and scale compared to earlier memristive-based designs, which can make it a valid consideration in the field of large-scale AI. Parallel processing brought about by the memristive synapses also makes computations fast; hence, real-time processing abilities are significantly enhanced. The paper also draws attention to the main problems, such as the variability of the devices and the problem of scalability, which should be solved to implement the practical work at the industrial level. This contribution to the design of next-generation AI hardware by offering an in-depth examination of the possibilities of memristive devices in neuromorphic computing suggests that not only will such devices be energy efficient, but they will also be capable of delivering high computational performance.
Analysis
Preparing paper insights from the available abstract and paper details.
Discovery
Hanna Hamrell; Thomas Sjögren; Hannes Ovrén
Minsu Ko; Yongjin Byun; Sungjun Kim
Jnanaprakash Dash; Sandaap Sathyanarayana; Muhammed Sahad E; Saptarshi Bej; Bikas C. Das
Farooq F; Kaith P; Bera A
K DK; Ponnambalam M
Wang Z; Liu X; Xu C; Zhao K; Xu T; Wang Y; Guan Z; Wu C; Zhao J; Yan X
Source record