Conditional Generation of Periodic Signals with Fourier-Based Decoder
This work addresses a domain-specific problem for applications involving periodic signal generation, but it is incremental as it builds on Fourier series concepts for a known bottleneck in sequential modeling.
The paper tackles the problem of modeling periodic signals with conventional sequential models, which often fail to capture periodicity effectively, by introducing a Fourier-based decoder framework that decomposes signals into sine and cosine components for conditional generation. The model outperforms baselines in reconstruction, imputation, and conditional generation tasks, showing more stable and refined results.
Periodic signals play an important role in daily lives. Although conventional sequential models have shown remarkable success in various fields, they still come short in modeling periodicity; they either collapse, diverge or ignore details. In this paper, we introduce a novel framework inspired by Fourier series to generate periodic signals. We first decompose the given signals into multiple sines and cosines and then conditionally generate periodic signals with the output components. We have shown our model efficacy on three tasks: reconstruction, imputation and conditional generation. Our model outperforms baselines in all tasks and shows more stable and refined results.