Rapid Wavelength Classification Using Deep Learning on Truncated Temporal Signals in Cu 2 O/Si Self-Powered Photodetector Arrays
Source abstract
Abstract
Conventional machine learning approaches for wavelength recognition using photodetectors typically rely on complete time-series photocurrent curves to extract critical temporal features such as rise and decay times. However, acquiring the full response curve can be time-consuming and impractical for real-time or edge applications. In this study, we propose a deep learning framework utilizing LSTM and BiLSTM networks to classify four distinct wavelengths (365, 465, 560, and 730 nm) based on truncated photocurrent signals from a self-powered Cu 2 O/Si photodetector array. The BiLSTM model achieved perfect classification accuracy (100%) with only the first 40 ms of the 150 ms signal when configured with 64 hidden units, demonstrating the feasibility of early-stage inference. In contrast, LSTM required the full temporal profile for comparable performance. The BiLSTM also exhibited strong robustness across varying training/test splits and random initializations, highlighting its generalization and reproducibility. These results demonstrate the potential of combining truncated temporal data with bidirectional deep learning to enable fast, efficient, and filter-free spectral sensing suitable for real-time edge-deployed applications.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Copper-based nanomaterials and applications · Thin-Film Transistor Technologies · Transition Metal Oxide Nanomaterials
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Photonics Group, Department of Engineering Science and Ocean Engineering
National Taiwan University
Department of Applied Materials and Optoelectronic Engineering
National Chi Nan University
Funding
National Science and Technology Council
References (37)
Show source-provided references
1. 10.1039/C8TC02188K10.1039/C8TC02188K ↗
2. 10.1016/j.nanoen.2023.10869610.1016/j.nanoen.2023.108696 ↗
3. 10.1021/acsaelm.0c0006610.1021/acsaelm.0c00066 ↗
4. 10.1109/TED.2023.327905410.1109/TED.2023.3279054 ↗
5. 10.1021/acsaelm.9b0036810.1021/acsaelm.9b00368 ↗
6. 10.1021/acsaelm.4c0098310.1021/acsaelm.4c00983 ↗
7. 10.1021/acsaelm.2c0007410.1021/acsaelm.2c00074 ↗
8. 10.1021/acsaelm.5c0077610.1021/acsaelm.5c00776 ↗
9. 10.1002/admt.20240174010.1002/admt.202401740 ↗
10. 10.1002/smll.20220792810.1002/smll.202207928 ↗
11. 10.1016/j.apsusc.2022.15421710.1016/j.apsusc.2022.154217 ↗
12. 10.1021/acsaelm.4c0170310.1021/acsaelm.4c01703 ↗
13. 10.35848/1347-4065/ada77f10.35848/1347-4065/ada77f ↗
14. 10.1021/acssensors.5c0118310.1021/acssensors.5c01183 ↗
15. 10.1016/j.jallcom.2025.18278410.1016/j.jallcom.2025.182784 ↗
16. 10.1002/adom.20180024910.1002/adom.201800249 ↗
17. 10.1016/j.cej.2023.14490810.1016/j.cej.2023.144908 ↗
18. 10.1016/j.neucom.2025.12971210.1016/j.neucom.2025.129712 ↗
19. 10.1039/D4TC03451A10.1039/D4TC03451A ↗
20. 10.1021/acsenergylett.0c0259910.1021/acsenergylett.0c02599 ↗
21. 10.1016/j.jallcom.2015.06.14710.1016/j.jallcom.2015.06.147 ↗
22. 10.1002/solr.20190033910.1002/solr.201900339 ↗
23. 10.1088/2053-1591/1/1/01500210.1088/2053-1591/1/1/015002 ↗
24. 10.1016/j.nanoen.2025.11123910.1016/j.nanoen.2025.111239 ↗
25. 10.1016/j.vacuum.2008.10.00310.1016/j.vacuum.2008.10.003 ↗
26. 10.1016/j.electacta.2013.08.03810.1016/j.electacta.2013.08.038 ↗
27. 10.1039/D2TC03856K10.1039/D2TC03856K ↗
28. 10.1021/acsaem.5c0189910.1021/acsaem.5c01899 ↗
29. 10.1021/acsanm.5c0197010.1021/acsanm.5c01970 ↗
30. 10.1016/j.vacuum.2022.11087610.1016/j.vacuum.2022.110876 ↗
31. 10.1016/j.sna.2022.11399210.1016/j.sna.2022.113992 ↗
32. 10.1021/ja304228c10.1021/ja304228c ↗
33. 10.1021/acsami.1c0513610.1021/acsami.1c05136 ↗
34. 10.1155/2022/614079610.1155/2022/6140796 ↗
35. 10.1007/s00521-023-08568-z10.1007/s00521-023-08568-z ↗
36. 10.1109/ACCESS.2020.298855010.1109/ACCESS.2020.2988550 ↗
37. 10.1088/1361-6463/adfbf910.1088/1361-6463/adfbf9 ↗