Fix the charts
The pretrained encoder and source-trained linear head stay frozen. k-means on standardized source embeddings defines fixed nearest-centroid charts.
Continual adaptation · 2026
Recurrent Domain Incremental Adaptation of Time-Series Foundation Models
A frozen-host method that corrects distribution shift only where it appears in the embedding space, while keeping an explicit path through the original model for conditions that recur.
Motivation
A fault-diagnosis model trained under one machine load or speed will meet new conditions during its service life. The same happens across users in wearable sensing or subjects in physiological monitoring. Each new condition arrives with only a small labeled calibration set, and earlier conditions, including the source, keep recurring at inference without anyone saying which condition produced a sample.
Fine-tuning shared parameters recovers the new condition but damages earlier ones. Strong regularization limits that damage but also limits adaptation. PACT starts from a measurement on frozen TSFM embeddings: condition shift is highly non-uniform across the embedding space. Some local regions degrade substantially; others barely change, and a single global correction needlessly moves embeddings that were already classified correctly.
Method
The pretrained encoder and source-trained linear head stay frozen. k-means on standardized source embeddings defines fixed nearest-centroid charts.
For each arriving condition, PACT learns an identity-initialized rank-8 residual transport per chart from the small labeled calibration set, through the frozen head.
The new transport bank is frozen and added to the atlas. Raw series and past class labels are discarded; only source and calibration embeddings are kept for routing.
A per-chart MLP gate, refit after each arrival, softly mixes the source identity branch with all learned transports from the embedding alone.
A margin-based condition shows when the gated correction is too small to change the frozen model's prediction; identity routing reproduces it exactly. In practice this certifies 99.8% of initially correct source predictions on FD, 97.5% on HHAR, and 71.1% on UCI-HAR, and every certified prediction was preserved.
Performance
4 / 5
With frozen Sundial, PACT leads all 14 main-comparison baselines on FD, WISDM, HHAR, and UCI-HAR. On Sleep-EDF, LwF is ahead by 0.09 points.
+11.2 pts
Over the strongest baseline on FD, and +8.3 points on WISDM, while discarding past class labels after each update.
8 hosts
Works unchanged on classification-pretrained, forecasting-pretrained, and task-specific encoders. With Mantis, it reaches 0.985 on FD and 0.967 on WISDM.
| Dataset | Strongest baseline | Baseline AMF1 | PACT AMF1 | PACT RF |
|---|---|---|---|---|
| FD | DualCD | 0.8460 | 0.9575 | 0.0021 |
| WISDM | LwF | 0.7940 | 0.8774 | 0.0292 |
| HHAR | DualCD | 0.8296 | 0.8367 | 0.0210 |
| UCI-HAR | LwF | 0.7351 | 0.7864 | 0.0425 |
| Sleep-EDF | LwF | 0.6138 | 0.6129 | 0.0171 |
Baselines span test-time and source-free adaptation, fine-tuning, regularization, distillation, and domain-incremental methods (S-Prompts, DualCD). Against rehearsal, PACT beats embedding replay on FD and WISDM; replay methods lead on HHAR, UCI-HAR, and Sleep-EDF.
What the results show
Ungated local transports improve plasticity but still apply unnecessary corrections. Adding the learned gate gives the best average macro-F1 and the lowest performance-weighted forgetting among adaptive variants on all five datasets. Rank-8 transports match or beat full-rank maps, and source k-means charts beat random-projection partitions at equal capacity.
Across eight frozen backbones the mechanism transfers unchanged, but final quality depends on the host: UCI-HAR ranges from 0.517 with PatchTST to 0.922 with Mantis. An embedding-space correction can only recover what the representation separates.