Besides, more electronics architectures have been also developed, such as Application-Specific Integrated Circuit (ASIC) and Field-Programmable Gate Array (FPGA) chips to increase the ANN computing speed and efficiency 6, 7, 8. As this demand continues, graphics processing unit (GPU) and even central processing unit (CPU)/GPU heterogenous architectures become attractive options for the ANN acceleration since they offer more computational parallelism than CPU 5. With the continuous advancement of ANN, the past decade has witnessed an exponential rise in the demand for high computing speed and low energy consumption 3, 4. The computational complexity of ANN in model iterations requires large computational ability for multiply-and-accumulate (MAC) operation 2. It consists of massive nodes, namely “neurons” connected to each other through synapses. Our accelerator constitutes a major step towards the realization of in-situ photonic neural network and pave the way for the integration of photonic integrated circuits (PIC).Īrtificial Neural Network (ANN) is a computational model for mimicking the human brain in information processing 1. Furthermore, we confirm its feasibility and capability for MNIST handwritten digit recognition, achieving a high accuracy of 91.6%. Our reconfigurable scheme enables implementation of a wide variety of nonlinear responses. Here, we demonstrate an efficient in-situ nonlinear accelerator comprising a unique solution-processed two-dimensional (2D) MoS 2 Opto-Resistive RAM Switch (ORS), which exhibits tunable nonlinear resistance switching that allow us to introduce nonlinearity to the photonic neuron which overcomes the linear voltage-power relationship of typical photonic components. Despite that the integrated Mach–Zehnder Interferometer (MZI) mesh can perform vector-matrix multiplication in photonic neural network, a programmable in-situ nonlinear activation function has not been proposed to date, suppressing further advancement of photonic neural network. Photonic neural network has been sought as an alternative solution to surpass the efficiency and speed bottlenecks of electronic neural network.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |