Rethinking Time Series Tokenization from a Frequency Perspective

Key points

• Tokenization that partitions time series into subsequences has become a foundational paradigm in modern time series modeling.1
• However, we find that current tokenization strategies inherently introduce spectral distortions, forcing the learned representations to diverge from the true underlying patterns such as trend, periodicity, and seasonality, thereby impairing model performance.1
• Through theoretical analysis from a frequency perspective, we derive the boundary conditions under which such distortions occur.1
• TimeFT obtains tokens through frequency-domain partitioning, frequency shifting, and Nyquist sampling.1

Sources (1)

• [1]Rethinking Time Series Tokenization from a Frequency Perspective
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 11:51 PM

  To overcome this distortion, we propose TimeFT, a nearly distortion-free frequency-based tokenizer for time series data.

  Tokenization that partitions time series into subsequences has become a foundational paradigm in modern time series modeling.

Extractive summary: sentences quoted from the sources.