Research

Paper

AI LLM March 04, 2026

ViterbiPlanNet: Injecting Procedural Knowledge via Differentiable Viterbi for Planning in Instructional Videos

Authors

Luigi Seminara, Davide Moltisanti, Antonino Furnari

Abstract

Procedural planning aims to predict a sequence of actions that transforms an initial visual state into a desired goal, a fundamental ability for intelligent agents operating in complex environments. Existing approaches typically rely on large-scale models that learn procedural structures implicitly, resulting in limited sample-efficiency and high computational cost. In this work we introduce ViterbiPlanNet, a principled framework that explicitly integrates procedural knowledge into the learning process through a Differentiable Viterbi Layer (DVL). The DVL embeds a Procedural Knowledge Graph (PKG) directly with the Viterbi decoding algorithm, replacing non-differentiable operations with smooth relaxations that enable end-to-end optimization. This design allows the model to learn through graph-based decoding. Experiments on CrossTask, COIN, and NIV demonstrate that ViterbiPlanNet achieves state-of-the-art performance with an order of magnitude fewer parameters than diffusion- and LLM-based planners. Extensive ablations show that performance gains arise from our differentiable structure-aware training rather than post-hoc refinement, resulting in improved sample efficiency and robustness to shorter unseen horizons. We also address testing inconsistencies establishing a unified testing protocol with consistent splits and evaluation metrics. With this new protocol, we run experiments multiple times and report results using bootstrapping to assess statistical significance.

Metadata

arXiv ID: 2603.04265
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-04
Fetched: 2026-03-05 06:06

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.04265v1</id>\n    <title>ViterbiPlanNet: Injecting Procedural Knowledge via Differentiable Viterbi for Planning in Instructional Videos</title>\n    <updated>2026-03-04T16:50:07Z</updated>\n    <link href='https://arxiv.org/abs/2603.04265v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.04265v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Procedural planning aims to predict a sequence of actions that transforms an initial visual state into a desired goal, a fundamental ability for intelligent agents operating in complex environments. Existing approaches typically rely on large-scale models that learn procedural structures implicitly, resulting in limited sample-efficiency and high computational cost. In this work we introduce ViterbiPlanNet, a principled framework that explicitly integrates procedural knowledge into the learning process through a Differentiable Viterbi Layer (DVL). The DVL embeds a Procedural Knowledge Graph (PKG) directly with the Viterbi decoding algorithm, replacing non-differentiable operations with smooth relaxations that enable end-to-end optimization. This design allows the model to learn through graph-based decoding. Experiments on CrossTask, COIN, and NIV demonstrate that ViterbiPlanNet achieves state-of-the-art performance with an order of magnitude fewer parameters than diffusion- and LLM-based planners. Extensive ablations show that performance gains arise from our differentiable structure-aware training rather than post-hoc refinement, resulting in improved sample efficiency and robustness to shorter unseen horizons. We also address testing inconsistencies establishing a unified testing protocol with consistent splits and evaluation metrics. With this new protocol, we run experiments multiple times and report results using bootstrapping to assess statistical significance.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-03-04T16:50:07Z</published>\n    <arxiv:comment>Accepted at CVPR 2026</arxiv:comment>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Luigi Seminara</name>\n    </author>\n    <author>\n      <name>Davide Moltisanti</name>\n    </author>\n    <author>\n      <name>Antonino Furnari</name>\n    </author>\n  </entry>"
}