Lightweight Spectrum Prediction Based on Knowledge Distillation
| dc.contributor.author | Cheng, R. | |
| dc.contributor.author | Zhang, J. | |
| dc.contributor.author | Deng, J. | |
| dc.contributor.author | Zhu, Y. | |
| dc.coverage.issue | 4 | cs |
| dc.coverage.volume | 32 | cs |
| dc.date.accessioned | 2024-01-09T14:20:49Z | |
| dc.date.available | 2024-01-09T14:20:49Z | |
| dc.date.issued | 2023-12 | cs |
| dc.description.abstract | To address the challenges of increasing complexity and larger number of training samples required for high-accuracy spectrum prediction, we propose a novel lightweight model, leveraging a temporal convolutional network (TCN) and knowledge distillation. First, the prediction accuracy of TCN is enhanced via a self-transfer method. Then, we design a two-branch network which can extract the spectrum features efficiently. By employing knowledge distillation, we transfer the knowledge from TCN to the two-branch network, resulting in improved accuracy for spectrum prediction of the lightweight network. Experimental results show that the proposed model can improve accuracy by 19.5% compared to the widely-used LSTM model with sufficient historical data and reduces 71.1% parameters to be trained. Furthermore, the prediction accuracy is improved by 17.9% compared to Gated Recurrent Units (GRU) in the scenarios with scarce historical data. | en |
| dc.format | text | cs |
| dc.format.extent | 469-478 | cs |
| dc.format.mimetype | application/pdf | en |
| dc.identifier.citation | Radioengineering. 2023 vol. 32, č. 4, s. 469-478. ISSN 1210-2512 | cs |
| dc.identifier.doi | 10.13164/re.2023.0469 | en |
| dc.identifier.issn | 1210-2512 | |
| dc.identifier.uri | https://hdl.handle.net/11012/244200 | |
| dc.language.iso | en | cs |
| dc.publisher | Společnost pro radioelektronické inženýrství | cs |
| dc.relation.ispartof | Radioengineering | cs |
| dc.relation.uri | https://www.radioeng.cz/fulltexts/2023/23_04_0469_0478.pdf | cs |
| dc.rights | Creative Commons Attribution 4.0 International license | en |
| dc.rights.access | openAccess | en |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | en |
| dc.subject | Spectrum prediction | en |
| dc.subject | knowledge distillation | en |
| dc.subject | temporal convolutional network | en |
| dc.subject | lightweight networks | en |
| dc.subject | few-shot learning | en |
| dc.title | Lightweight Spectrum Prediction Based on Knowledge Distillation | en |
| dc.type.driver | article | en |
| dc.type.status | Peer-reviewed | en |
| dc.type.version | publishedVersion | en |
| eprints.affiliatedInstitution.faculty | Fakulta eletrotechniky a komunikačních technologií | cs |
Files
Original bundle
1 - 1 of 1
Loading...
- Name:
- 23_04_0469_0478.pdf
- Size:
- 873.5 KB
- Format:
- Adobe Portable Document Format
- Description: