PaperlayerView exact revision on arXiv

Source-linked AI summary

Towards A Unified Information Bottleneck Framework for Time Series Explanations

Xu Zheng, Zichuan Liu, Zhuomin Chen, Mayur Akewar, Janki Bhimani, Jason Liu, Mo Sha, Jingchao Ni, Wei Cheng, Dongsheng Luo

arXiv:2608.25897v1cs.LGcs.AI

TL;DR

Time-series explainability remains divided between attribution and counterfactual methods, with the former lacking causal validation and the latter producing unstable, adversarial-like changes. TimeX++ unifies both through an Information Bottleneck objective, generating in-distribution attributions and stable counterfactuals, and reports state-of-the-art performance across synthetic and real-world datasets.

  • Problem

    Attribution and counterfactual explanations are studied as disconnected tasks, leaving attributions without causal verification and counterfactuals without semantic grounding.

    Source: §I · p. 1

  • Method

    TimeX++ uses an Information Bottleneck-based unified objective to generate in-distribution attribution instances and semantically grounded counterfactuals.

    Source: §I · pp. 1–2

  • Results

    TimeX++ achieves state-of-the-art performance for both attribution and counterfactual explanation tasks on synthetic and real-world datasets.

    Source: §I · p. 2

  • Takeaways & Limitations

    The framework provides faithful attributions and robust counterfactual explanations while reflecting complex behaviors of pretrained time-series classifiers.

    Source: §VII · p. 12

Abstract

from arXiv · show

Explaining deep learning models operating on time series data is crucial in various applications that require transparent and interpretable insights into model behavior. {Existing explanation methods generally fall into two categories: attribution-based explanations, which identify the temporal regions most responsible for a prediction, and counterfactual explanations, which reveal how an input should be modified to alter the model's decision.} {Despite valuable insights, these two fields are largely studied independently. This disconnect leaves attribution methods lacking causal validation, while counterfactual methods suffer from severe instability, producing adversarial-like noise instead of meaningful explanations.} In this work, we revisit time-series explainability from an information-theoretic perspective and show that existing explainers are vulnerable to trivial solutions and distributional shifts. To address these limitations, we propose a unified objective function for explainable time series learning that bridges attribution and counterfactual reasoning within a single framework. Building upon the Information Bottleneck principle, our formulation explicitly prevents trivial explanations and out-of-distribution counterfactuals. {Based on this objective function, we introduce {\modelname}, a novel explanation framework that learns a parametric transformation network to construct explanation-embedded instances, where preserved information yields attribution explanations and controlled information removal produces stable counterfactual explanations.} We evaluate {\modelname} on synthetic and real-world benchmarks against state-of-the-art baselines. Extensive quantitative and qualitative results show that {\modelname} consistently outperforms competing methods, yielding faithful attributions and stable counterfactual explanations.

I. INTRODUCTION

Time-series explainability is split between attribution and counterfactual methods, leaving causal validation and semantic stability insufficiently addressed. The paper unifies both through an Information Bottleneck objective and proposes TimeX++ to produce in-distribution attributions and stable counterfactuals.

Motivation

  • Attribution methods identify salient temporal regions, but commonly use ad-hoc objectives and can produce out-of-distribution samples during evaluation.Masking or substituting salient features may violate the data manifold, making classifier-based evaluations unreliable.Source: §I · p. 1, §II · p. 3
  • Treating attribution and counterfactual explanations separately leaves attributions without causal verification and counterfactuals vulnerable to adversarial noise.Counterfactuals lack semantic grounding when label-flip searches modify the entire time series without attribution guidance.Source: §I · p. 1

Unified framework

  • The Information Bottleneck principle connects explanations by balancing compactness, I(X; X′), against informativeness, I(X′; Y ).Attribution is framed as retaining label-sufficient information while discarding redundant temporal dynamics.Source: §I · pp. 1–2
  • TimeX++ generates in-distribution explanation-embedded instances for attribution and constrains counterfactual perturbations to meaningful feature dimensions.The framework uses an attribution bottleneck to support semantically grounded counterfactual generation.Source: §I · p. 2

Results

  • TimeX++ achieves state-of-the-art performance on attribution and counterfactual tasks across synthetic and real-world datasets.The paper evaluates both explanation types against existing methods and reports their effectiveness.Source: §I · p. 2

III. PROBLEM FORMULATION

The paper formulates time-series attribution and counterfactual explanation as related information-bottleneck problems. It defines their objectives, identifies computational and distributional challenges, and replaces intractable mutual-information terms with tractable constraints.

III. PROBLEM FORMULATION

  • The task concerns post-hoc, instance-level explanations for a black-box time-series classifier operating on multivariate or univariate inputs.Inputs are sequences X ∈ R^T×D with labels from a finite label space.Source: §III · p. 3
  • Attribution seeks a masked sub-instance X′ = X ⊙ M that explains the classifier’s prediction.The mask M is binary over temporal positions and features.Source: §III · p. 3
  • A differentiable extractor maps X to mask probabilities and samples a mask through independent Bernoulli variables.The extractor produces M ∈ [0,1]^T×D probabilities before discrete sampling.Source: §III · p. 3
  • Counterfactual explanation seeks a minimally perturbed X′′ classified as a target label, with perturbations restricted to the attribution mask’s semantic support.The objective requires f(X′′) = Y′′ while minimizing d(X, X′′).Source: §III · p. 3

IV. EXPLAINING TIME SERIES LEARNING VIA INFORMATION BOTTLENECK

  • The Information Bottleneck view balances label informativeness against compactness by optimizing I(X′; Y) − αI(X; X′).The paper replaces direct informativeness estimation with a tractable label-consistency measure.Source: §IV · p. 3
  • Direct application is challenged by intractable mutual-information estimation, out-of-distribution attribution inputs, and adversarially unstable counterfactuals.Masked instances can violate the data manifold, while unconstrained counterfactual optimization can exploit classifier vulnerabilities.Source: §IV · p. 3

A. Tractable Compactness Bound

  • The compactness term is replaced by a variational upper bound that penalizes mask size and entropy through a Bernoulli prior.This yields a KL-based tractable constraint that compresses the bottleneck without complex mutual-information estimators.Source: A. Tractable Compactness Bound · pp. 3–4

B. Distribution-Preserving and Bounded Generation

TimeX++ evaluates masked explanations through generated in-distribution instances and constrains counterfactual changes to the shared attribution bottleneck. Its unified objective combines compactness, label consistency, distribution preservation, and structural bounding.

B. Distribution-Preserving and Bounded Generation

A. Attribution Bottleneck Extractor

B. Conditioned Generators

The conditioned generators transform attribution masks into explanation-embedded instances and generate counterfactual perturbations confined to masked regions. Separate generator designs address OOD attribution artifacts and adversarial counterfactual noise.

B. Conditioned Generators

  • Conditioned generators translate the structural bottleneck M into generated explanation-embedded instances in the data space.ψa targets attribution OOD issues, while ψcf targets counterfactual robustness.Source: B. Conditioned Generators · p. 5
  • TimeXa++ combines the masked input with a dataset-sampled background and uses ψa to generate a smoothed instance.Distribution preservation and a reference-distance constraint regulate the generated attribution instance.Source: B. Conditioned Generators · p. 5
  • TimeXcf++ confines counterfactual perturbations to the semantic bottleneck to avoid exploiting classifier null spaces.The method contrasts with unconstrained perturbation in continuous temporal space.Source: B. Conditioned Generators · p. 5
  • Training computes compactness, reference, distribution, and label-consistency losses before updating the extractor and active generators.The pipeline alternates generator-specific optimization according to the selected explanation mode.Source: B. Conditioned Generators · p. 5
  • The counterfactual generator produces a target-driven perturbation matrix E from X and M.The structural form uses the input, mask, perturbation matrix, and auxiliary variation.Source: B. Conditioned Generators · p. 5
  • A variation function incorporates dynamics from a training instance with the target label, while inference removes the auxiliary noise term.During inference, the counterfactual simplifies to X̃cf = X + M ⊙ E.Source: B. Conditioned Generators · p. 5
  • The bottleneck constraint penalizes deviations from the original reference in non-bottleneck background regions.This acts as a hard anchor against imperceptible adversarial changes in redundant features.Source: B. Conditioned Generators · p. 5

C. End-to-End Optimization

TimeX++ shares one architecture and mathematical objective across attribution and counterfactual modes. Switching the objective label and active generator enables end-to-end optimization for either explanation type.

C. End-to-End Optimization

  • The entire framework is optimized end-to-end using a shared architecture and unified objective.Both TimeXa++ and TimeXcf++ minimize the same overarching formulation.Source: C. End-to-End Optimization · p. 6
  • Attribution and counterfactual training alternate by toggling the objective label and selecting the corresponding generator.The procedure is summarized in Algorithm 1.Source: C. End-to-End Optimization · p. 6

VI. EXPERIMENTS

The experiments evaluate TimeX++ across attribution and counterfactual explanation tasks using diverse datasets, baselines, and quantitative metrics.

  • The evaluation targets faithful attributions, effective counterfactuals, reduced out-of-distribution shift, and improved inference efficiency.Source: §VI · p. 6
  • The experimental setup combines diverse datasets, established baselines, and strict evaluation metrics.Source: A. Experimental Setup · p. 6

1) Datasets:

The study uses ten synthetic and real-world time-series datasets, compares eleven explainers, and evaluates attribution and counterfactual quality with complementary metrics.

Datasets

  • Ten datasets span synthetic benchmarks with ground-truth explanations and real-world scenarios across multiple application domains.Source: 1) Datasets · p. 6

Baselines

  • TimeX++ is benchmarked against eleven state-of-the-art explainers spanning attribution and counterfactual paradigms.Source: 2) Baselines · p. 6

Attribution Metrics

Counterfactual Metrics

  • Counterfactuals are evaluated by validity, confidence, sparsity, Proximity-L1, and Proximity-L2, with direction-specific preferences.Source: 3) Evaluation Metrics · pp. 6–7

B. Efficacy of Unified Explanations

TimeX++ is evaluated through a unified training process for attribution and counterfactual explanations, with strong attribution results and Pareto-based counterfactual comparison.

Unified Evaluation

Attribution

Counterfactuals

C. Resilience to Out-of-Distribution Shift

The resilience analysis addresses distributional shift in attribution and adversarial degeneration in counterfactual explanations through distributional metrics, localized perturbations, and noise testing.

Distributional Alignment

Counterfactual Robustness

E. Runtime Results

TimeX++ is evaluated for runtime across attribution and counterfactual explanation tasks, with efficient training and especially fast inference. Additional experiments indicate strong performance across classifier architectures and counterfactual objectives.

  • Training each experiment takes approximately 3 to 15 minutes per fold, depending on dataset volume.Source: E. Runtime Results · p. 10
  • TimeXa++ requires less overall attribution training time than TIMEX because it directly perturbs the black-box instead of training a white-box consistency model.Source: E. Runtime Results · p. 10
  • TimeX++ is the most expedient model during inference across both attribution and counterfactual tasks.Source: E. Runtime Results · p. 10
  • TimeXcf++ is over 100 times faster than the fastest counterfactual baseline, AB-CF, by using a single forward inference pass.Source: E. Runtime Results · p. 10
  • Under CNN predictors, TimeXa++ achieves the highest AUPRC and AUP on SeqComb-MV and ECG datasets.Source: F. Flexibility Across Different… · p. 11
  • Across CNN and LSTM backbones, TimeXcf++ balances validity, confidence, sparsity, and proximity, indicating generalizability across architectures.Source: F. Flexibility Across Different… · p. 11

G. Ablation Study and Hyperparameter Analysis

Ablation and hyperparameter analyses show that TimeX++ depends on discrete masking, distributional and structural losses, and careful control of sparsity. Visualizations further indicate faithful attribution regions and compact, pattern-related counterfactual modifications.

  • 39.78% higher AUPRC, 22.04% higher AUP, and 3.10% higher AUR are achieved on ECG when TimeXa++ uses STE against leading baselines.Across datasets, STE also increases AUPRC by an average of 8.87% versus continuous masking.Source: G. Ablation Study and Hyperparameter… · p. 11
  • Removing LKL widens the distributional gap and reduces attribution performance, while removing Ldr causes the explainer to fail.Source: G. Ablation Study and Hyperparameter… · p. 11
  • Removing LLC dramatically degrades counterfactual validity because the generator loses its primary signal for targeted label flipping.Source: G. Ablation Study and Hyperparameter… · p. 11
  • Removing Ldr substantially worsens counterfactual proximity, while removing Lbound deteriorates sparsity and proximity by scattering perturbations across the sequence.Source: G. Ablation Study and Hyperparameter… · p. 12
  • Attribution performance stabilizes for r between 0.4 and 0.7, whereas increasing r improves counterfactual validity and confidence but worsens sparsity and proximity.Source: G. Ablation Study and Hyperparameter… · p. 12
  • On FreqShapes, TimeXa++ focuses on peaks matching ground-truth explanations, while TimeXcf++ makes fewer, pattern-related modifications than competing counterfactual methods.IG highlights unnecessarily large regions, TIMEX is affected by distribution bias, and several counterfactual baselines modify large input portions or fail validity.Source: H. Case Study · p. 12
  • The conclusion identifies dataset-dependent hyperparameter control over explanation quantifiers as a key limitation.Source: §VII · p. 12

Related papers

Loading 2608.25897v1…