time-series-generation

[official] PyTorch implementation of TimeVQVAE from the paper ["Vector Quantized Time Series Generation with a Bidirectional Prior Model", AISTATS 2023]

[IJCAI 2025] Official implementation of "T2S: High-resolution Time Series Generation with Text-to-Series Diffusion Models"

Generative Models for Real-World Sensor Time Series

Synthetic financial time series generation with regime clustering

Official implementation of STDiffusion, published in ICDM 2025.

Unofficial implementation of Renewables Scenario Generation GAN (Chen et al., IEEE 2018) in PyTorch 2.

FourierFlow is a tool for generating synthetic financial time series data using deep generative models. This modified version is utilized for the Market Scenario Generator Hackathon: From Stability to Storms. It aims to provide effective synthetic data generation to model different financial scenarios.

Diffusion model for time series generation and time series decomposition through wavelet transform. Implementation of progressive distillation on the trained models for lower sampling time

A novel approach, named SamplerGAN, for generating high-quality labeled data

Industrial Production Index Time Series Forecasting using a range of models including Holt-Winters, ARIMA, SARIMA, LSTMs, and Facebook's Prophet. The project focuses on predicting production trends through model evaluation, tuning, and visualization of forecasted outcomes.

Unofficial implementation of TimeGAN (Yoon et al., NIPS 2019) in PyTorch 2.

A comprehensive collection of implementations for Deep Generative Models, including VAEs, Normalizing Flows, CycleGAN, EBMs, Score-Based Models, Diffusion (DDPM/DDIM), and Flow Matching. Features applications in Anomaly Detection, Disentanglement, Style Transfer, Subject-Driven Generation (DreamBooth), and Financial Time Series Synthesis.

Unofficial implementation of Renewables Scenario Generation GAN (Chen et al., IEEE 2018) in TensorFlow 2.

Generative modeling pipeline for artifact correction in MEG data, based on latent diffusion models

Unofficial implementation of TimeGAN (Yoon et al., NIPS 2019) in TensorFlow 2.

This is the code related to my MSc thesis at the Norwegian University of Science and Technology (NTNU). The MSc program is Electronic system design and innovation with a specialization in signal processing. The goal of this thesis is to explore and compare the state of the art solution to more traditional models for time series generation.

Python scripts to use the Azure SDKs, plus a few general machine learning and related scripts.

Generative model for rough volatility: log-signatures + a learned Besov-wavelet decoder reconstruct high-frequency texture via differentiable IDWT. Pluggable MLP/attention/transformer backbones, scale-weighted wavelet loss, and a 5-dataset multi-domain registry (fBM, rough Bergomi, Burgers turbulence, CHB-MIT EEG, ESC-50 audio).

Unemployment Rate Forecasting using Time Series techniques, leveraging Statsmodels, LSTMs, and Facebook's Prophet library to predict future unemployment trends. The project includes model comparison, hyperparameter tuning, and visualization of forecasted results.

Time-causal financial generative models: refactored TC-VAE baselines with causal VQ/RVQ tokenizers, token priors, S&P500/VIX, Hawkes/SVMHJD, multi-dimensional benchmarks, and path-risk diagnostics.