Quantifying the Pre-training Dividend: Generative versus Latent Self-Supervised Learning for Time Series Foundation Models

• URL: http://arxiv.org/abs/2605.19462v1
• Date: Tue, 19 May 2026 07:13:20 GMT
• Title: Quantifying the Pre-training Dividend: Generative versus Latent Self-Supervised Learning for Time Series Foundation Models
• Abstract summary: Self-supervised learning (SSL) in vision and NLP has motivated its rapid adoption for time series.<n>We establish a framework to evaluate the "pre-training dividend": the value added by SSL across diverse temporal tasks.<n>Our analysis reveals that the pre-training dividend is highly asymmetric. SSL yields gains of up to 375% for anomaly detection and classification, yet remains marginal for forecasting.
• Score: 5.3565231231826695
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The success of self-supervised learning (SSL) in vision and NLP has motivated its rapid adoption for time series. However, research has focused primarily on Generative paradigms and forecasting tasks, leaving the broader utility of learned representations unquantified. We establish a controlled framework to evaluate the "pre-training dividend": the value added by SSL across diverse temporal tasks. We systematically compare Generative paradigms against Latent Alignment architectures, introducing adaptations of LeJEPA and DINO for time series. These adaptations utilize Discrete Wavelet Transform (DWT) augmentations to enforce invariance to local fluctuations. Our analysis reveals that the pre-training dividend is highly asymmetric: SSL yields gains of up to 375% for anomaly detection and classification, yet remains marginal for forecasting. We demonstrate that representational utility is non-universal, governed by a precision-invariance trade-off where the specific signal resolution required by the task must align with the objective. Finally, we show that representation quality is largely independent of data origin and saturates at moderate architectural depths, suggesting a path to scaling via massive synthetic generation. Our code is available at: https://github.com/noammajor/Models

Related papers

• Self-Supervised Representation-Guided Generative Dataset Distillation [55.27114962330541]
  We propose self-supervised representation-guided generative dataset distillation (SRG)<n>SRG translates the SSL geometry into diffusion guidance.<n>It consistently outperforms evaluated generative baselines across multiple datasets and IPC settings.
  arXiv  Detail & Related papers  (2026-08-04T06:51:45Z)
• Learning by Shifting: Temporal View Construction for Time Series Contrastive Learning [4.7318594191739765]
  Supervised learning demands large quantities of labeled data, a bottleneck that is expensive and reliant on domain-specific expertise.<n>Self-supervised learning, particularly contrastive learning, has emerged as a compelling alternative, enabling rich representation learning directly from unlabeled data.<n>Existing approaches for time series rely on hand-crafted augmentations and maskings that embed strong domain assumptions.<n>We present a systematic analysis of contrastive learning dynamics in time series settings, examining the effects of batch size and the number of negatives on downstream performance.
  arXiv  Detail & Related papers  (2026-06-20T09:07:30Z)
• SAW: Stage-Aware Dynamic Weighting for Multi-Objective Reinforcement Learning in Large Language Models [66.92079232778707]
  We propose Stage-Aware Dynamic Weighting (SAW) as a lightweight, algorithm-agnostic dynamic weighting mechanism.<n>SAW reweights each dimension's reward or advantage contribution by its relative informativeness within the batch.<n> Experiments on tool-calling and text summarization tasks demonstrate that SAW consistently improves both training efficiency and final performance.
  arXiv  Detail & Related papers  (2026-06-05T10:00:19Z)
• Self-Supervised Learning via Flow-Guided Neural Operator on Time-Series Data [57.85958428020496]
  Flow-Guided Neural Operator (FGNO) is a novel framework combining operator learning with flow matching for SSL training.<n>FGNO learns mappings in functional spaces by using Short-Time Fourier Transform to unify different time resolutions.<n>Unlike prior generative SSL methods that use noisy inputs during inference, we propose using clean inputs for representation extraction while learning representations with noise.
  arXiv  Detail & Related papers  (2026-02-12T18:54:57Z)
• Large Language Models as Discounted Bayesian Filters [14.164508061248775]
  We introduce a Bayesian filtering framework to evaluate online inference in Large Language Models (LLMs)<n>We find that while LLM belief updates resemble Bayesian posteriors, they are more accurately characterized by an exponential forgetting filter with a model-specific discount factor smaller than one.<n>Although inherent priors are often miscalibrated, the updating mechanism itself remains structured and principled.
  arXiv  Detail & Related papers  (2025-12-20T19:56:39Z)
• Exploring Structural Degradation in Dense Representations for Self-supervised Learning [84.52554180480037]
  We observe a counterintuitive phenomenon in self-supervised learning (SSL): longer training may impair the performance of dense prediction tasks.<n>We refer to this phenomenon as Self-supervised Dense Degradation (SDD) and demonstrate its consistent presence across sixteen state-of-the-art SSL methods.<n>We introduce a Dense representation Structure Estimator (DSE) composed of a class-relevance measure and an effective dimensionality measure.
  arXiv  Detail & Related papers  (2025-10-20T08:40:16Z)
• How Different from the Past? Spatio-Temporal Time Series Forecasting with Self-Supervised Deviation Learning [15.102926671713668]
  We propose ST-SSDL, a Spatio-Temporal series time forecasting framework.<n>It discretizes latent space using learnable prototypes that represent typicaltemporal patterns.<n>Experiments show that ST-SSDL consistently outperforms state-of-the-art baselines across multiple metrics.
  arXiv  Detail & Related papers  (2025-10-06T15:21:13Z)
• Enhancing Semantic Segmentation with Continual Self-Supervised Pre-training [11.897717409259492]
  Self-supervised learning (SSL) has emerged as a central paradigm for training foundation models.<n>We propose GLARE, a novel continual self-supervised pre-training task designed to enhance downstream segmentation performance.
  arXiv  Detail & Related papers  (2025-09-22T14:11:02Z)
• ViRN: Variational Inference and Distribution Trilateration for Long-Tailed Continual Representation Learning [6.253882111488726]
  ViRN is a novel framework that integrates variational inference with distributional trilateration for robust long-tailed learning.<n> evaluated on six long-tailed classification benchmarks, including speech (e.g., rare acoustic events, accents) and image tasks.<n> achieves a 10.24% average accuracy gain over state-of-the-art methods.
  arXiv  Detail & Related papers  (2025-07-23T10:04:30Z)
• Inference-Time Scaling of Diffusion Language Models with Particle Gibbs Sampling [70.8832906871441]
  We study how to steer generation toward desired rewards without retraining the models.<n>Prior methods typically resample or filter within a single denoising trajectory, optimizing rewards step-by-step without trajectory-level refinement.<n>We introduce particle Gibbs sampling for diffusion language models (PG-DLM), a novel inference-time algorithm enabling trajectory-level refinement while preserving generation perplexity.
  arXiv  Detail & Related papers  (2025-07-11T08:00:47Z)
• Federated Class-Incremental Learning with Hierarchical Generative Prototypes [10.532838477096055]
  Federated Learning (FL) aims at unburdening the training of deep models by distributing computation across multiple devices (clients)<n>Our proposal constrains both biases in the last layer by efficiently finetuning a pre-trained backbone using learnable prompts.<n>Our method significantly improves the current State Of The Art, providing an average increase of +7.8% in accuracy.
  arXiv  Detail & Related papers  (2024-06-04T16:12:27Z)
• Self-Ensembling GAN for Cross-Domain Semantic Segmentation [107.27377745720243]
  This paper proposes a self-ensembling generative adversarial network (SE-GAN) exploiting cross-domain data for semantic segmentation. In SE-GAN, a teacher network and a student network constitute a self-ensembling model for generating semantic segmentation maps, which together with a discriminator, forms a GAN. Despite its simplicity, we find SE-GAN can significantly boost the performance of adversarial training and enhance the stability of the model.
  arXiv  Detail & Related papers  (2021-12-15T09:50:25Z)
• Exploring Complementary Strengths of Invariant and Equivariant Representations for Few-Shot Learning [96.75889543560497]
  In many real-world problems, collecting a large number of labeled samples is infeasible. Few-shot learning is the dominant approach to address this issue, where the objective is to quickly adapt to novel categories in presence of a limited number of samples. We propose a novel training mechanism that simultaneously enforces equivariance and invariance to a general set of geometric transformations.
  arXiv  Detail & Related papers  (2021-03-01T21:14:33Z)
• Unsupervised Controllable Generation with Self-Training [90.04287577605723]
  controllable generation with GANs remains a challenging research problem. We propose an unsupervised framework to learn a distribution of latent codes that control the generator through self-training. Our framework exhibits better disentanglement compared to other variants such as the variational autoencoder.
  arXiv  Detail & Related papers  (2020-07-17T21:50:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.