File size: 10,158 Bytes
4478ec0
 
c6d52d6
 
4478ec0
c6d52d6
 
4478ec0
c6d52d6
 
 
 
 
 
 
 
 
 
 
 
4478ec0
 
 
 
 
 
 
 
 
 
c6d52d6
a47e85c
 
 
 
4478ec0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c6d52d6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
---
license: apache-2.0
size_categories:
- 10K<n<100K
task_categories:
- other
pretty_name: 'ConceptTime: Datasets and Checkpoints (NeurIPS 2026)'
tags:
- time-series
- classification
- interpretability
- concept-bottleneck
- self-supervised
- eeg
- human-activity-recognition
- sleep-staging
- uea
- fault-diagnosis
- neurips-2026
- reproducibility
viewer: false
---

# 🧩 ConceptTime: Datasets & Checkpoints

**Predictive Surprise as Self-Grounding Concept Bottleneck for Interpretable Time Series** (NeurIPS 2026)

Sachith Abeywickrama, Emadeldeen Eldele, Min Wu, Xiaoli Li, Chau Yuen

[![Code](https://img.shields.io/badge/GitHub-Sachithx%2FConceptTime-181717?logo=github)](https://github.com/Sachithx/ConceptTime)
[![arXiv](https://img.shields.io/badge/arXiv-2611.12955-b31b1b?logo=arxiv)](https://arxiv.org/abs/2611.12955)
[![Venue](https://img.shields.io/badge/NeurIPS-2026-4b44ce)](https://neurips.cc/virtual/2026/loc/sydney/poster/153183)
[![Project Page](https://img.shields.io/badge/Project-Website-2176BC?logo=GoogleChrome)](https://sachithx.github.io/research/concepttime.html)
[![HF Daily Paper](https://img.shields.io/badge/HF%20Daily%20Paper-2611.12955-FFD21E.svg?logo=huggingface)](https://huggingface.co/papers/2611.12955)
[![HF Dataset](https://img.shields.io/badge/HuggingFace-Datasets-FFD21E.svg?logo=huggingface)](https://huggingface.co/datasets/sachithabey/ConceptTime)
[![License](https://img.shields.io/badge/License-Apache%202.0-green)](https://github.com/Sachithx/ConceptTime/blob/main/LICENSE)

This repository holds the data and trained checkpoints that go with the
[ConceptTime code on GitHub](https://github.com/Sachithx/ConceptTime). There are two parts:

- **Ready-to-evaluate assets** for **UCI-HAR** and **Sleep-EDF**: the test split plus both trained checkpoints.
  `download_data.py` fetches them to the exact paths the evaluation script expects.
  Epilepsy, the third headline dataset, ships directly in the GitHub repo.
- **An `archive/`** with **all 17 benchmarks** (full train/val/test splits) and **all trained checkpoints**,
  including ablations and sweeps. Use it to retrain, run ablations, or explore beyond the headline results.

## What is ConceptTime?

Interpretable time-series classifiers face a chicken-and-egg problem. You can't define meaningful temporal
concepts without knowing where segments begin and end, and you can't place meaningful boundaries without
knowing which concepts to look for. ConceptTime breaks this loop with a single self-supervised signal,
**predictive surprise**.

1. A frozen probabilistic **density model** (a causal transformer) predicts a distribution over the next value at every timestep.
2. The mismatch between that prediction and the actual value is the **surprise**. It places segment boundaries, and it describes each segment by *how it behaved relative to expectation* rather than by *what it looks like*.
3. The segment summaries cluster into a vocabulary of dynamical regimes, called **concepts**. A lightweight transformer head then classifies from the sequence of concepts.
4. The distance to the nearest concept gives a **reliability score by design**, which supports calibrated abstention and OOD detection.

This needs no human concept annotations, no language model calls, and no domain experts. Across 17 public benchmarks,
ConceptTime is competitive with state-of-the-art black-box classifiers, beats every interpretable baseline, and keeps
about 98% accuracy on Epilepsy with only 1% of the labels.

## Repository contents

```
dataset/                       test splits for the headline datasets
β”œβ”€β”€ HAR/test.pt
└── SLeep-EDF/test.pt
checkpoints/                   trained checkpoints for the headline datasets
β”œβ”€β”€ HAR/{density_model_best,pipeline_best}.pt
└── SLeep-EDF/{density_model_best,pipeline_best}.pt
archive/                       everything (~4.2 GB)
β”œβ”€β”€ datasets/<Dataset>/{train,val,test}.pt          (FD-X: {train,val,test}_x.pt)
└── checkpoints/<Dataset>/
    β”œβ”€β”€ gaussian_entropy_best.pt                    density model
    β”œβ”€β”€ pipeline/entropy/pipeline_M<K>_*.pt         ConceptTime-Dynamic (surprise-based segmentation)
    β”œβ”€β”€ pipeline/static/pipeline_M<K>_*.pt          ConceptTime-Static (fixed-window ablation)
    └── sweep/...                                    hyperparameter sweeps (some datasets)
```

Each dataset needs **two checkpoints** for evaluation:

- `density_model_best.pt` is the frozen density model. It drives segmentation and the surprise signatures.
  In `archive/` this file is called `gaussian_entropy_best.pt`, its older name.
- `pipeline_best.pt` holds the concept space, the signature standardizer, and the classifier.

The archive pipeline filenames record their configuration. For example, in
`pipeline_M32_surprise_surprise_full_gmm_transformer_..._ls0.6_lt0.5.pt`, `M32` is a vocabulary of 32 concepts,
the first `surprise` is the boundary mode, `surprise_full` is the signature family, `gmm` is the clustering method,
`transformer` is the classifier head, and the trailing values are loss weights.

**Data format.** Every `.pt` file is a dictionary:

```python
{"samples": FloatTensor[N, C, T], "labels": LongTensor[N]}
```

### Datasets

| Dataset | Domain | Train / Test | Channels | Length *T* | Classes |
|---|---|---|---:|---:|---|
| UCI-HAR | Wearable inertial sensing | 7,352 / 2,947 | 9 | 128 | 6 activities |
| Epilepsy | EEG | 9,200 / 2,300 | 1 | 178 | Seizure vs. non-seizure |
| Sleep-EDF | Sleep EEG (Fpz-Cz, 100 Hz) | 25,612 / 8,910 | 1 | 3,000 | 5 sleep stages |
| Fault Diagnosis (FD-A–D) | Bearing fault diagnosis, 4 operating conditions | 8,184 / 2,728 | 1 | 5,120 | 3 (healthy, inner fault, outer fault) |
| EthanolConcentration | UEA: Alcohol industry | 261 / 263 | 3 | 1,751 | 4 |
| FaceDetection | UEA: Face / MEG | 5,890 / 3,524 | 144 | 62 | 2 |
| Handwriting | UEA: Handwriting | 150 / 850 | 3 | 152 | 26 |
| Heartbeat | UEA: Heart sounds | 204 / 205 | 61 | 405 | 2 |
| JapaneseVowels | UEA: Speech | 270 / 370 | 12 | 29 | 9 |
| PEMS-SF | UEA: Transportation | 267 / 173 | 963 | 144 | 7 |
| SelfRegulationSCP1 | UEA: EEG | 268 / 293 | 6 | 896 | 2 |
| SelfRegulationSCP2 | UEA: EEG | 200 / 180 | 7 | 1,152 | 2 |
| SpokenArabicDigits | UEA: Speech | 6,599 / 2,199 | 13 | 93 | 10 |
| UWaveGestureLibrary | UEA: Gesture recognition | 120 / 320 | 3 | 315 | 8 |

For Epilepsy, the four non-seizure classes of the original five-class dataset are merged into one, which makes
it binary seizure detection. For the UEA datasets, the official train/test splits are used, and validation
examples are carved out of the training split. FD-A to FD-D are used for the cross-domain transfer experiments.

## Quick start

```bash
git clone https://github.com/Sachithx/ConceptTime && cd ConceptTime
pip install -r requirements.txt

python download_data.py                    # HAR + Sleep-EDF test splits and checkpoints
python download_data.py --datasets HAR     # just one

python src/evaluate.py --dataset HAR \
    --artifact checkpoints/HAR/pipeline_best.pt \
    --density  checkpoints/HAR/density_model_best.pt --split test
```

Each run prints accuracy and macro-F1 at 1%, 5%, and full label budgets. Use `--dataset Epilepsy` or `--dataset SLeep-EDF`
to evaluate the other headline datasets.

To fetch **everything**, including the archive, use `huggingface_hub`:

```python
from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="sachithabey/ConceptTime",
    repo_type="dataset",
    local_dir="ConceptTime-assets",
    allow_patterns=["archive/datasets/Heartbeat/*", "archive/checkpoints/Heartbeat/*"],  # or omit to get all ~4.3 GB
)
```

## Expected results

Results for ConceptTime-Dynamic, the paper's primary method. The numbers are test accuracy (%) under different
label budgets, averaged over 3 seeds (Table 1 of the paper). A single released checkpoint can differ slightly from the 3-seed mean.

| Dataset | 1% labels | 5% labels | Full |
|---|---:|---:|---:|
| UCI-HAR | 74.60 | 79.60 | 88.91 |
| Epilepsy | 97.91 | 98.01 | 98.39 |
| Sleep-EDF | 70.21 | 71.92 | 78.15 |
| UEA (10-dataset average) | 58.50 | 60.10 | 71.82 |

ConceptTime-Dynamic is the strongest interpretable method in every column. At 1% labels on Epilepsy, it also beats
every black-box baseline (97.91% compared with 95.41% for TSLANet).

## Source datasets and licensing

The checkpoints are released under the Apache 2.0 license. The datasets are preprocessed versions of public benchmarks.
They are included here only for convenience and reproducibility, and they remain subject to the terms of their original sources.
If you use them, please also cite the original authors:

- **UCI-HAR:** Anguita et al., *A Public Domain Dataset for Human Activity Recognition Using Smartphones*, ESANN 2013. [UCI repository](https://archive.ics.uci.edu/dataset/240/human+activity+recognition+using+smartphones)
- **Epilepsy:** Andrzejak et al., *Indications of nonlinear deterministic and finite-dimensional structures in time series of brain electrical activity*, Phys. Rev. E, 2001
- **Sleep-EDF:** Goldberger et al., *PhysioBank, PhysioToolkit, and PhysioNet*, Circulation, 2000. [PhysioNet](https://physionet.org/content/sleep-edfx/)
- **UEA archive:** Bagnall et al., *The UEA multivariate time series classification archive, 2018*, arXiv:1811.00075. [timeseriesclassification.com](https://www.timeseriesclassification.com/)
- **Fault Diagnosis:** Lessmeier et al., *Condition Monitoring of Bearing Damage in Electromechanical Drive Systems*, PHM Society European Conference, 2016

## Citation

If you find ConceptTime or these assets useful, please cite:

```bibtex
@inproceedings{sachith_concepttime_26,
  title     = {Predictive Surprise as Self-Grounding Concept Bottleneck for Interpretable Time Series},
  author    = {Abeywickrama, Sachith and Eldele, Emadeldeen and Wu, Min and Li, Xiaoli and Yuen, Chau},
  booktitle = {Advances in Neural Information Processing Systems},
  year      = {2026}
}
```

## Contact

For questions or issues, please open an issue on [GitHub](https://github.com/Sachithx/ConceptTime/issues).