{"slug": "orbit-autoregressive-time-series-generation-via-next-frequency-band-prediction", "title": "ORBIT: Autoregressive Time Series Generation via Next Frequency-Band Prediction", "summary": "Researchers introduced ORBIT, a deep generative framework for time series that predicts frequency bands autoregressively, combining a Frequency-Hierarchical VQ-VAE (FH-VQ-VAE) with a Frequency-Band Autoregressive Transformer (FBAR). The FH-VQ-VAE decomposes latent temporal representations into frequency bands and applies residual quantization with increasing token budgets, while the FBAR generates from low to high frequency, building global structure before fine-grained detail. In experiments on multiple real-world benchmarks, ORBIT reported state-of-the-art overall performance with better inference efficiency than advanced baselines, according to the arXiv paper 2610.27000v1.", "body_md": "arXiv:2610.27000v1 Announce Type: new \nAbstract: Deep generative models provide an effective modern solution to alleviating the problem of data scarcity by synthesizing realistic time series. However, despite the considerable progress achieved by existing deep time series generators, two key challenges remain: (1) learning hierarchical representations for heterogeneous temporal patterns; and (2) coordinating the generation process of heterogeneous temporal dynamics. To address these key bottlenecks, in this work, we propose ORBIT, a unified deep generative framework for time series. Specifically, we develop two key components: the Frequency-Hierarchical VQ-VAE (FH-VQ-VAE) and the Frequency-Band Autoregressive Transformer (FBAR). The former learns hierarchical representations by decomposing latent temporal representations into different frequency bands and applying residual quantization with increasing token budgets. The latter establishes a novel low-to-high frequency autoregressive direction, yielding a time series generation process that first constructs global temporal structures and then incrementally incorporates fine-grained temporal details. These compact designs not only disentangle heterogeneous temporal patterns for more effective representation learning but also mitigate potential interference among different temporal dynamics during generation. Extensive experiments on multiple challenging real-world benchmarks demonstrate that ORBIT achieves state-of-the-art overall performance while exhibiting superior inference efficiency compared with advanced baselines.", "url": "https://wpnews.pro/news/orbit-autoregressive-time-series-generation-via-next-frequency-band-prediction", "canonical_source": "https://arxiv.org/abs/2610.27000", "published_at": "2026-10-21 04:00:00+00:00", "updated_at": "2026-10-21 04:21:18.418139+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "artificial-intelligence"], "entities": ["ORBIT", "Frequency-Hierarchical VQ-VAE", "Frequency-Band Autoregressive Transformer", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/orbit-autoregressive-time-series-generation-via-next-frequency-band-prediction", "markdown": "https://wpnews.pro/news/orbit-autoregressive-time-series-generation-via-next-frequency-band-prediction.md", "text": "https://wpnews.pro/news/orbit-autoregressive-time-series-generation-via-next-frequency-band-prediction.txt", "jsonld": "https://wpnews.pro/news/orbit-autoregressive-time-series-generation-via-next-frequency-band-prediction.jsonld"}}