Self-Supervised Contrastive Pre-Training For Time Series

Time series datasets present a unique challenge for pre-training because of the potential mismatch between pre-training and fine-tuning that can lead to poor performance. Domain adaptation methods can mitigate unwanted distribution shifts, including shifts in dynamics, varying trends, and long-range and short-range cyclic effects. To do so, these methods require access to data points in the target (fine-tuning) domain, meaning they must access target data points early on during pre-training, which is not possible in the real world.

To address this challenge, we put forward a principle: time-based and frequency-based representations of the same time series and its local augmentations must produce consistent latent representations and predictions. This generalizable property, called Time-Frequency Consistency (TF-C), is theoretically rooted in signal processing theory and leads to effective pre-training.

Time series are relevant for many areas, including medical diagnosis and healthcare settings. While representation learning has considerably advanced analysis of time series, learning broadly generalizable representations for time series remains a fundamentally challenging problem. Being able to generate such representations can directly improve pre-training. The central problem is transferring knowledge from a time series dataset to other different datasets during fine-tuning. The difficulty due to distribution shifts and other issues is compounded by the complexity of time series: large variations of temporal dynamics across datasets, varying semantic meaning, system factors, etc.

Supervised pre-training requires large annotated datasets, creating an obstacle to deep learning deployment. For example, in medical domains, experts often disagree on ground-truth labeling, which can introduce unintended biases (e.g., ECG signals falling outside of normal and abnormal rhythms).

We introduce a generalizable concept called Time-Frequency Consistency (TF-C) and use it to develop a broadly generalizable pre-training strategy.

Motivation for TF-C

The TF-C approach uses self-supervised contrastive learning to transfer knowledge across time series domains and pre-train models. The approach builds on the fundamental duality between time and frequency views of time signals. TF-C embeds time-based and frequency-based views learned from the same time series sample such that they are closer to each other in the joint time-frequency space; the embeddings are farther apart if they are associated with different time series samples.

The broad scope of the TF-C principle indicates that the approach promotes positive transfer across different time-series datasets. That is true even when datasets are complex in terms of considerable variations of temporal dynamics across datasets, varying semantic meaning, irregular sampling, and systemic factors.

The following figure gives an overview of the TF-C approach. Given a time series sample, the resulting time and frequency based embeddings are close to each other in a latent time-frequency space (panel a). TF-C expands the scope and applicability of pre-training; it can transfer pre-trained models to diverse scenarios (panel b).

TF-C approach

TF-C has four neural network components (as shown in the following figure):

  • Contrastive time encoder,
  • Contrastive frequency encoder, and
  • two cross-space projectors that map time-based and frequency-based representations to the same time-frequency space.

To position time-based and frequency-based embeddings close together, TF-C specifies the following constrastive learning objective:

  1. We first apply time-domain augmentations to the input time series sample. The embeddings of the original and the augmented views produced by the time-domain encoder are used to compute a term in the contrastive loss.

  2. We transform the input time series to its frequency spectrum. Then, similarly to before, we apply frequency-domain augmentations to the input time series sample, produce the embeddings, and compute another term in the contrastive loss.

  3. TF-C requires the consistency between time-domain and frequency-domain representations, which is achieved by a well-designed consistency loss item measured in the time-frequency space. As shown in the figure, our total contrastive loss is a weighted sum of these three terms.

  4. Finally, during the fine-tuning phase, we concatenate the time-domain and frequency-domain embeddings to form the sample embedding. This learned sample embedding can be fed into a downstream task such as classification.

Attractive properties of TF-C

  • A generalizable assumption (TF-C) for time series: It is grounded in the signal theory that a time series can be represented equivalently in either the time or frequency domain. Hence, our assumption is very reasonable that TF-C is invariant to different time-series datasets and can guide the development of effective pre-training models.

  • Self-supervised pretraining for diverse scenarios: The TF-C model can be used with pre-training dataset that are entirely unlabelled. Results show that TF-C provides effective transfer across four scenarios involving time series datasets with different kinds of sensors and measurements and still achieve strong performance when the domain gap is large (e.g., when pre-training and fine-tuning data points are given in different feature spaces or have different semantic meaning).


Self-Supervised Contrastive Pre-Training For Time Series via Time-Frequency Consistency
Xiang Zhang, Ziyuan Zhao, Theodoros Tsiligkaridis, and Marinka Zitnik
Proceedings of Neural Information Processing Systems, NeurIPS 2022 [arXiv]

title = {Self-Supervised Contrastive Pre-Training For Time Series via Time-Frequency Consistency},
author = {Zhang, Xiang and Zhao, Ziyuan and Tsiligkaridis, Theodoros and Zitnik, Marinka},
booktitle = {Proceedings of Neural Information Processing Systems, NeurIPS},
year      = {2022}


PyTorch implementation of TF-C and all datasets are available in the GitHub repository.


  • SleepEEG dataset of whole-night EEG recordings monitored by a sleep cassette

  • Epilepsy dataset of single-channel EEG measurements from 500 subjects

  • FD-A dataset of an electromechanical drive system monitoring the condition of rolling bearings

  • FD-B dataset of an electromechanical drive system monitoring the condition of rolling bearings

  • HAR dataset of human daily activities, including walking, walking upstaris, walking downstairs, sitting, standing, and laying

  • Gesture dataset of accelerometer measurements for eight hand gestures

  • ECG dataset of ECG measuruments across four underlying conditions of cardiac arrhythmias

  • EMG dataset of EMG recordings of muscular dystrophies and neuropathies


Latest News

Mar 2023:   New Paper in Nature Machine Intelligence

New paper with NASA in Nature Machine Intelligence on biomonitoring and precision health in deep space supported by artificial intelligence.

Mar 2023:   New Paper in Nature Machine Intelligence

Mar 2023:   TxGNN - Zero-shot prediction of therapeutic use

Mar 2023:   GraphXAI published in Scientific Data

Feb 2023:   Welcoming New Postdoctoral Fellows

A warm welcome to postdoctoral fellows Ruth Johnson and Wanxiang Shen. We are thrilled to have them joining us soon and look forward to working together.

Feb 2023:   New Preprint on Distribution Shifts

Feb 2023:   PrimeKG published in Scientific Data

Jan 2023:   GNNDelete published at ICLR 2023

Jan 2023:   New Network Principle for Molecular Phenotypes

Dec 2022:   Can we shorten rare disease diagnostic odyssey?

New preprint! Geometric deep learning for diagnosing patients with rare genetic diseases. Implications for using deep learning on sparsely-labeled medical datasets. Thankful for this collaboration with Zak Lab. Project website.

Nov 2022:   Can AI transform the way we discover new drugs?

Our conversation with Harvard Medicine News highlights recent developments and new features in Therapeutics Data Commons.

Oct 2022:   New Paper in Nature Biomedical Engineering

New paper on graph representation learning in biomedicine and healthcare published in Nature Biomedical Engineering.

Sep 2022:   New Paper in Nature Chemical Biology

Our paper on artificial intelligence foundation for therapeutic science is published in Nature Chemical Biology.

Sep 2022:   Self-Supervised Pre-Training at NeurIPS 2022

New paper on self-supervised contrastive pre-training accepted at NeurIPS 2022. Project page. Thankful for this collaboration with the Lincoln National Laboratory.

Sep 2022:   Best Paper Honorable Mention Award at IEEE VIS

Our paper on user-centric AI of drug repurposing received the Best Paper Honorable Mention Award at IEEE VIS 2022. Thankful for this collaboration with Gehlenborg Lab.

Sep 2022:   Multimodal Representation Learning with Graphs

Aug 2022:   On Graph AI for Precision Medicine

The recording of our tutorial on using graph AI to advance precision medicine is available. Tune into four hours of interactive lectures about state-of-the-art graph AI methods and applications in precision medicine.

Aug 2022:   Evaluating Explainability for GNNs

New preprint! We introduce a resource for broad evaluation of the quality and reliability of GNN explanations, addressing challenges and providing solutions for GNN explainability. Project website.

Jul 2022:   New Frontiers in Graph Learning at NeurIPS

Excited to organize the New Frontiers in Graph Learning workshop at NeurIPS.

Jul 2022:   AI4Science at NeurIPS

We are excited to host the AI4Science meeting at NeurIPS discussing AI-driven scientific discovery, implementation and verification of AI in science, the influence AI has on the conduct of science, and more.

Zitnik Lab  ·  Artificial Intelligence in Medicine and Science  ·  Harvard  ·  Department of Biomedical Informatics