New L-GTA model enhances time series data augmentation

Researchers have developed L-GTA, a novel latent generative model for time series data augmentation. This model, built on a Variational Autoencoder with a Bi-LSTM backbone and temporal self-attention, learns latent representations and applies controlled perturbations. L-GTA aims to improve consistency between latent space and data space transformations, resulting in augmented samples with predictable signatures. Evaluations show L-GTA outperforms existing methods like TimeGAN and Diffusion-TS in downstream forecasting tasks, reducing prediction error by up to 27% compared to original data. AI

IMPACT This new augmentation technique could improve the accuracy of time series forecasting and anomaly detection models.

RANK_REASON The cluster contains a research paper detailing a new generative model for time series augmentation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Bi-LSTM-AMSM: bidirectional long short-term memory network and attention mechanism with semantic mining for e-commerce web page recommendation
• Diffusion-TS
• L-GTA
• Luis Roque
• TimeGAN
• TimeVAE
• variational auto-encoder

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →