ConceptTime:以預測驚奇構建可解釋時間序列分類器
Predictive Surprise as Self-Grounding Concept Bottleneck for Interpretable Time Series
ConceptTime 以凍結的機率預測器產生的預測驚奇切分時間序列,將片段聚類為動態概念,再按概念序列分類,並在證據不足時棄權。在 17 個公開基準上,它的準確度與黑盒最先進方法相若,勝過所有可解釋基準;在 Epilepsy 上僅用 1% 標籤仍維持接近 98% 準確度。整個方法無需標註、語言模型呼叫或領域專家。
Published on Nov 7
Authors:
,
,
,
Abstract
A self-supervised method segments time series by predictive surprise from a frozen forecaster, clusters segments into dynamical concepts, and classifies with reliability-aware abstention, achieving competitive accuracy and interpretability without annotations.
Time-series classifiers deployed in risk-sensitive domains must be (1) accurate, (2) interpretable, and (3) able to abstain when evidence is weak. Existing methods do not deliver all 3 at once, and the obstacle is structural, rather than logistical. Meaningful temporal patterns cannot be defined without knowing segment boundaries, yet meaningful boundaries cannot be placed without knowing which patterns to expect. Existing approaches break this circularity by using fixed windows or expert-annotated vocabularies, thereby sacrificing at least one of the 3 properties. We introduce ConceptTime, which closes the loop through a single self-supervised signal. A frozen probabilistic forecaster predicts a distribution over the next observation at every timestep of the sequence. We call the mismatch between its prediction and the realized signal the predictive surprise. This signal both locates segment boundaries and characterizes each resulting segment through the forecaster. Segments are thus grounded by how they behaved relative to expectation, not by what they look like. These summaries cluster into a vocabulary of dynamical regimes that we call concepts. A lightweight head classifies from the concept sequence, and the distance to the nearest concept yields reliability by design. Across 17 public benchmarks, ConceptTime is competitive with state-of-the-art black-box accuracy, beats every interpretable baseline, and retains near-98% accuracy on Epilepsy with 1% of labels. It further remains competitive with softmax, prototype-distance, and SHAP/TimeX/TimeX++ on OOD detection and deletion faithfulness, without a single annotation, language model call, or domain expert. Code is available at: https://github.com/Sachithx/ConceptTime.git.
View arXiv page View PDF Project page GitHub 1 Add to collection
Get this paper in your agent:
hf papers read 2611.12955
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash
Models citing this paper 0
No model linking this paper
Cite arxiv.org/abs/2611.12955 in a model README.md to link it from this page.
Datasets citing this paper 1
sachithabey/ConceptTime
Spaces citing this paper 0
No Space linking this paper
Cite arxiv.org/abs/2611.12955 in a Space README.md to link it from this page.
Collections including this paper 1
來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co