cs.AINov 19, 2026

MambaLIFT: Reference-Based Mechanism-Aware Attribution for VMamba

Authors: Jintu Kakati, Niladri B. Puhan

Organizations: School of Electrical and Computer Sciences, Indian Institute of Technology Bhubaneswar, Bhubaneswar 752050, India

Abstract

Attributing predictions of visual state-space models remains challenging because input-dependent selective dynamics couple feature selection, recurrent state propagation, and shared computational pathways. We introduce MambaLIFT, a reference-based, multiplier-first attribution framework for VMamba that propagates contributions through its computational graph using interaction-aware rules for selective-scan injection, recurrent memory, LayerNorm, and shared pathways. MambaLIFT decomposes the reference-relative class-logit difference, fc(x)−fc(x0)f_c(x)-f_c(x^0), by spatial position, VSS block, scan direction, and computational mechanism. Experiments on ImageNet-1K evaluate VMamba-T and VMamba-B, with hyperparameters selected on a calibration set and faithfulness assessed on a disjoint held-out evaluation set. MambaLIFT achieves numerical conservation to floating-point precision across audited attribution interfaces in both configurations. Under mean-colour pixel flipping, it achieves significantly higher mean faithfulness scores than MambaLRP on both configurations and the highest scores among the evaluated methods, including Integrated Gradients with matched references. Under blurred replacement, MambaLIFT is non-inferior to MambaLRP on VMamba-T and equivalent on VMamba-B, indicating that its advantage depends on the replacement protocol. Ablations on VMamba-T show that interaction allocation, temporal propagation, and pathway composition affect attribution maps. Mechanism-level analyses further show that exact relevance conservation does not imply additive effects under finite interventions: the mechanism decomposition accounts for the reference-relative logit difference but does not estimate intervention effects or establish causal validity.

Figures & tables

Explore similar work

CardsList
  1. Size-Matched Nulls Reveal What Block Ablation Measures in MambaVision

    Nov 14, 2026Tahmina Kousar, Tehseen ZiaNeural Network InterpretabilityHybrid Mamba-Transformer Architectures

  2. A Good Initialization is All You Need for Faithful Visual Attribution

    Jul 7, 2026Zihan Gu, Jiayu Wang, Hua Zhang +1Feature AttributionPerturbation-Based Feature Attribution

  3. READ: Reading the Selective Scan for Faithful Video Attribution

    Oct 9, 2026Magdalena Trędowicz, Piotr Kubaty, Arkadiusz Lewicki +2Feature AttributionSelective SSMs