MantisV2: Closing the Zero-Shot Gap in Time Series Classification with Synthetic Data and Test-Time Strategies

TL;DR

MantisV2 introduces synthetic data training, architectural improvements, and test-time strategies to significantly enhance zero-shot time series classification, outperforming prior models across multiple benchmarks.

Contribution

The paper presents MantisV2, a lightweight encoder trained on synthetic data with new test-time methods, achieving state-of-the-art zero-shot performance in time series classification.

Findings

MantisV2 outperforms prior models on multiple benchmarks.

Synthetic training data improves zero-shot capabilities.

Enhanced test-time strategies boost classification accuracy.

Abstract

Developing foundation models for time series classification is of high practical relevance, as such models can serve as universal feature extractors for diverse downstream tasks. Although early models such as Mantis have shown the promise of this approach, a substantial performance gap remained between frozen and fine-tuned encoders. In this work, we introduce methods that significantly strengthen zero-shot feature extraction for time series. First, we introduce Mantis+, a variant of Mantis pre-trained entirely on synthetic time series. Second, through controlled ablation studies, we refine the architecture and obtain MantisV2, an improved and more lightweight encoder. Third, we propose an enhanced test-time methodology that leverages intermediate-layer representations and refines output-token aggregation. In addition, we show that performance can be further improved via self-ensembling and cross-model embedding fusion. Extensive experiments on UCR, UEA, Human Activity Recognition (HAR) benchmarks, and EEG datasets show that MantisV2 and Mantis+ consistently outperform prior time series foundation models, achieving state-of-the-art zero-shot performance.