Register now After registration you will be able to apply for this opportunity online.
Diffusing Time Series in the Wavelet Domain
Diffusion models (DDPMs) have revolutionised generative modelling, surpassing GANs in images, advancing audio synthesis, and enabling de-novo protein design. Yet progress on time series lags behind early adversarial work. Recent studies highlight the benefits of spectral biases - FourierFlow and frequency-domain DDPMs.
In parallel, diffusion in the wavelet domain has emerged for images, offering a multi-resolution view well-suited to non-stationary signals. Wavelets capture localised, scale-dependent features, making them attractive for domains from finance to climate and biomedical data such as ECGs.
This project proposes the first DDPM framework operating directly in the wavelet domain for time series, aiming to improve generalisation, interpretability, and robustness across diverse sequential tasks.
Keywords: Diffusion Models, Time Series, Wavelet Domain
See attached PDF for more information on the project.
Please email **(1)** your CV, **(2)** transcripts, **(3)** a brief motivation paragraph, and **(4)** an indicative timeline/availability to
samuel.ruiperezcampillo@inf.ethz.ch and jorge.dasilvagoncalves@inf.ethz.ch.
**Incomplete applications may not be considered.** We look forward to hearing from you!
See attached PDF for more information on the project.
Please email **(1)** your CV, **(2)** transcripts, **(3)** a brief motivation paragraph, and **(4)** an indicative timeline/availability to samuel.ruiperezcampillo@inf.ethz.ch and jorge.dasilvagoncalves@inf.ethz.ch.
**Incomplete applications may not be considered.** We look forward to hearing from you!
1. Build and benchmark DDPMs on multiple time-series domains (biomedical, economics, weather).
2. Compare the impact of different wavelet morphologies within DDPMs.
3. Evaluate cross-domain generalisation and robustness.
4. Develop learnable or adaptive wavelets that optimise generative performance.
5. Explore multimodal DDPMs combining layered wavelet decompositions for richer representations.
1. Build and benchmark DDPMs on multiple time-series domains (biomedical, economics, weather). 2. Compare the impact of different wavelet morphologies within DDPMs. 3. Evaluate cross-domain generalisation and robustness. 4. Develop learnable or adaptive wavelets that optimise generative performance. 5. Explore multimodal DDPMs combining layered wavelet decompositions for richer representations.
M_thesis_short_proposal_diffusion_wavelet.pdf
