RocketPFN: Accurate Time Series Classification via In-Context Learning
Provides a strong, training-free baseline for evaluating zero-shot time series classification foundation models.
RocketPFN achieves state-of-the-art mean accuracy of 0.900 on 92 UCR datasets, matching the best published method HC2, with no training on target data and median inference time of 30 seconds per fold. It also significantly outperforms TSC foundation models like MOMENT, Mantis, and MantisV2.
We introduce RocketPFN, a training-free pipeline for time series classification that combines random convolutional feature extraction (Rocket) with in-context classification via a pretrained tabular foundation model (TabPFN v2.5). On 92 UCR datasets (30-resample protocol), RocketPFN matches HC2, the strongest published method on the archive, in mean accuracy (both 0.900, Wilcoxon p=0.50), with no training on the target data and a median inference time of 30 seconds per fold. It also significantly outperforms every individual classifier in the HC2 ensemble. On UEA (20 datasets) the difference is likewise not statistically significant. A separate comparison concerns TSC foundation models: when paired with the same downstream classifier, MOMENT, Mantis, and MantisV2 are all significantly outperformed by RocketPFN using fewer extracted features and no learned parameters (p<0.001 in each case). This holds even when the encoders were pretrained on corpora that include the UCR training samples. We propose this two-stage pipeline as a reference point for evaluating zero-shot TSC foundation models.