- Rating
- 1
- Confidence
- 3
- Informativeness
- 1
- Presentation
- 1
- Soundness
- 2
Read review
Paper summary
Time DeCode addresses noise through a discrete tokenizer that separates sparse irregular components and applies frequency-ordered residual quantization. A Transformer predicts future code indices using next multi-code prediction, and code-space prediction calibration aggregates overlapping soft predictions before decoding. Experiments cover seven forecasting benchmarks, ten classification datasets and controlled corruption settings. Forecasting training uses only each target dataset's training split; classification adapts the objective and attention architecture.
Critical strengths
- The design is motivated by the characteristics of time-series data, specifically observational noise. The targeted methodological choices and detailed analytical experiments strengthen the empirical credibility of the work.
- The codebook-based approach is relatively simple, with moderately coupled components that may offer good modular extensibility.
Critical weaknesses
- The methodological design is insufficiently explored. The Noise-Aware Discrete Tokenizer uses “a lightweight network” without specifying its architecture, and the overall structure may introduce representational bottlenecks. The frequency-related ablations do not independently establish the advantage of additive fusion.
- The main forecasting experiments cover relatively few datasets and domains, providing insufficient evidence of broad effectiveness. Recent baselines such as TimesFM-3 and Toto 2.0 are missing. Moreover, the short input context of 96 steps may substantially affect performance and raises concerns about the adequacy and fairness of the comparison.
Questions for authors
- The evaluated datasets are relatively small. Experience from other modalities suggests that their scale, complexity and pattern diversity may be insufficient to train comprehensive codebooks. Have the authors considered joint pretraining on substantially larger datasets?
- Could the method directly support time-series foundation models, or serve as a general-purpose embedding approach that improves their generalization and scalability? Larger-scale experiments would help assess these possibilities.