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

TL;DR

ViterbiPlanNet introduces a novel framework that explicitly incorporates procedural knowledge into action sequence prediction for instructional videos, improving efficiency and robustness through a differentiable Viterbi layer.

Contribution

The paper presents ViterbiPlanNet, a framework that embeds a Procedural Knowledge Graph within a differentiable Viterbi layer, enabling explicit procedural knowledge integration and end-to-end training.

Findings

Achieves state-of-the-art performance on CrossTask, COIN, and NIV datasets.

Uses significantly fewer parameters than diffusion- and LLM-based planners.

Demonstrates improved sample efficiency and robustness to shorter horizons.

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.