ChatPPG.
Join free
First page or project page of “CHORD: Multi-resolution Compression for Multimodal Time Series” (opens in a new tab)

PaperarXiv:2611.11913

CHORD: Multi-resolution Compression for Multimodal Time Series

PreprintDated Nov 9, 2026Added Nov 10, 2026

  • BU

Authors: Bibhus Luitel; Payal Mohapatra; Subrata Biswas; Ekata Poudel; Sunwoong Kim; Qi Zhu; Bashima Islam (UMass Amherst; Arm; Worcester Polytechnic Institute; Northwestern University)

arXiv preprint (cs.AI)

PDF (opens in a new tab) arXiv2611.11913 (opens in a new tab) Cite
CC BY 4.0
First page or project page of “CHORD: Multi-resolution Compression for Multimodal Time Series” (opens in a new tab)

Luitel, Bibhus et al., "CHORD: Multi-resolution Compression for Multimodal Time Series", arXiv:2611.11913, CC BY 4.0

TL;DR

UMass Amherst / Arm / WPI / Northwestern coarse-to-fine learned compressor for multimodal sensor streams: one payload decodes at several temporal resolutions with a bit rate fixed in advance. At an 8x target, downstream classification stays within 0.62 points of uncompressed on average, the smallest gap of the compressed methods tested. Lossy; no code stated.

Abstract

Wearables, mobile devices, and IoT platforms generate multimodal time-series data continuously and at scale, making efficient storage and transfer a primary concern. The same recordings are reused after collection by many downstream models, often at different temporal resolutions, so a compressed representation should be small, useful across resolutions, and produced at a bit rate known in advance. Prior work meets parts of this: symbolic and error-bounded compressors reconstruct accurately but fix resolution at compression time, learned codecs target one signal domain, and discrete tokenizers reconstruct without accounting for rate. We introduce CHORD, a coarse-to-fine compressor for heterogeneous multimodal time series. A residual cascade orders each payload from broad to fine temporal scales and synthesizes corrections through learned shapelet-inspired atoms. A shared pathway supplies a reusable synthesis basis across modalities, while private pathways retain capacity for sensor-specific morphology. Finite scalar quantization fixes the bit width of every modality and scale, so payload size is known before encoding, and scale dropout trains each coarse-to-fine prefix to be a valid reconstruction operating point. Across seven wearable-sensing and forecasting benchmarks, CHORD stays within 0.62 points of uncompressed classification accuracy on average at the 8x target, the smallest gap of any compressed method, with all resolutions decoded from one payload.

Verbatim from arXiv (opens in a new tab) · arXiv metadata (abstracts are released under CC0)

Results on ChatPPG leaderboards

All leaderboards

No results from this paper on ChatPPG leaderboards yet. Browse the leaderboards.

Tasks & signals

Tasks

Activity recognition

Signals

PPGECGIMUEDA

Devices

Empatica E4RespiBAN chest band

Keywords

CompressionLearned CodecMulti ResolutionFinite Scalar QuantizationMultimodalWearableRate ControlWesad

Datasets used

How it works

How Method works: Setup → Method → TargetSetup: Signals (ppg, ecg, eda, accelerometer, imu). Method: Method (Transformer). Target: Activity recognition.SetupSignals: ppg, ecg, eda, accelerometer, imuSignalsppg, ecg, eda, accelerometer, imuMethod×NMethodTransformerTargetActivity recognitionActivity recognition

Cite this paper

Cite this paper with the fields recorded here. Check it against the publisher before submitting.

Luitel, B., Mohapatra, P., Biswas, S., Poudel, E., Kim, S., Zhu, Q., & Islam, B. (2026). CHORD: Multi-resolution Compression for Multimodal Time Series. arXiv preprint. https://arxiv.org/abs/2611.11913
Luitel, Bibhus, Payal Mohapatra, Subrata Biswas, Ekata Poudel, Sunwoong Kim, Qi Zhu, and Bashima Islam. 2026. "CHORD: Multi-resolution Compression for Multimodal Time Series." arXiv preprint. https://arxiv.org/abs/2611.11913.
Luitel, B., Mohapatra, P., Biswas, S., Poudel, E., Kim, S., Zhu, Q. and Islam, B. (2026) 'CHORD: Multi-resolution Compression for Multimodal Time Series', arXiv preprint. Available at: https://arxiv.org/abs/2611.11913
Luitel, Bibhus, et al. "CHORD: Multi-resolution Compression for Multimodal Time Series." arXiv preprint, 2026. https://arxiv.org/abs/2611.11913.
B. Luitel et al., "CHORD: Multi-resolution Compression for Multimodal Time Series," arXiv preprint, 2026, [Online]. Available: https://arxiv.org/abs/2611.11913.
@misc{luitel2026chord,
  title         = {{CHORD: Multi-resolution Compression for Multimodal Time Series}},
  author        = {Luitel, Bibhus and Mohapatra, Payal and Biswas, Subrata and Poudel, Ekata and Kim, Sunwoong and Zhu, Qi and Islam, Bashima},
  year          = {2026},
  howpublished  = {arXiv preprint},
  eprint        = {2611.11913},
  archivePrefix = {arXiv}
}

Source metadata: 2026

Record updated Nov 10, 2026

Added to the ChatPPG research hub

Same tasks, data and authors

Papers

Models