Acta Optica Sinica, Volume. 45, Issue 3, 0320001(2025)
Design and Analysis of Optical Extreme Learning Machine Based on Free Space Propagation
Fig. 1. Experimental designs of digital ELM and optical ELM. (a) Design of digital ELM; (b) design of optical ELM
Fig. 3. Numerical experiment results of digital ELM for different sizes of the inputs, the number of hidden nodes is fixed at 4000. (a) Test accuracy of digital ELM as a function of the input size; (b) train time of digital ELM as a function of the input size; (c) memory requirements of digital ELM as a function of the input size
Fig. 4. Numerical experiment results of digital ELM for different number of hidden nodes, the size of the input is fixed at 40000. (a) Test accuracy of digital ELM as a function of the number of hidden nodes; (b) train time of digital ELM as a function of the number of hidden nodes; (c) memory requirements of digital ELM as a function of the number of hidden nodes
Fig. 5. Experimental characterization of optical ELM based on free space propagation. (a) Accuracy as a function of the number of hidden nodes when the propagation distance is fixed at 10 µm; (b) accuracy as a function of propagation distance when the number of hidden nodes is fixed at 4000; (c) training time as a function of the number of hidden nodes; (d) memory requirements as a function of different numbers of hidden nodes
Fig. 6. Accuracy of optical ELMs with different numbers of hidden nodes as a function of the propagation distance
Get Citation
Copy Citation Text
Zhihong Xu, Schoenhardt Steffen, Xi Chen, Min Gu, Goi Elena. Design and Analysis of Optical Extreme Learning Machine Based on Free Space Propagation[J]. Acta Optica Sinica, 2025, 45(3): 0320001
Category: Optics in Computing
Received: Oct. 21, 2024
Accepted: Nov. 14, 2024
Published Online: Feb. 20, 2025
The Author Email: Goi Elena (elenagoi@usst.edu.cn)
CSTR:32393.14.AOS241671