Pith. sign in

REVIEW 2 cited by

TimeSHAP: Explaining Recurrent Models through Sequence Perturbations

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2012.00073 v2 pith:XHTZS34P submitted 2020-11-30 cs.LG cs.AI

TimeSHAP: Explaining Recurrent Models through Sequence Perturbations

classification cs.LG cs.AI
keywords timeshapmodelpositivepredictionsrecurrentaccountattributionattributions
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Although recurrent neural networks (RNNs) are state-of-the-art in numerous sequential decision-making tasks, there has been little research on explaining their predictions. In this work, we present TimeSHAP, a model-agnostic recurrent explainer that builds upon KernelSHAP and extends it to the sequential domain. TimeSHAP computes feature-, timestep-, and cell-level attributions. As sequences may be arbitrarily long, we further propose a pruning method that is shown to dramatically decrease both its computational cost and the variance of its attributions. We use TimeSHAP to explain the predictions of a real-world bank account takeover fraud detection RNN model, and draw key insights from its explanations: i) the model identifies important features and events aligned with what fraud analysts consider cues for account takeover; ii) positive predicted sequences can be pruned to only 10% of the original length, as older events have residual attribution values; iii) the most recent input event of positive predictions only contributes on average to 41% of the model's score; iv) notably high attribution to client's age, suggesting a potential discriminatory reasoning, later confirmed as higher false positive rates for older clients.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. From Prediction to Practice: A Task-Aware Evaluation Framework for Blood Glucose Forecasting

    cs.LG 2026-05 unverdicted novelty 7.0

    A new evaluation framework shows that blood glucose forecasting models with high overall accuracy often fail at timely hypoglycemia detection in high-risk periods and at predicting effects of changed insulin doses.

  2. Direct optimization of the probability of lesion origin in proton treatment planning for low-grade glioma patients

    physics.med-ph 2025-06 unverdicted novelty 7.0

    A framework for directly optimizing proton treatment plans using an extended probability of lesion origin (POLO) model in low-grade glioma patients.