pith. sign in

Differentiable Dynamic Programming for Structured Prediction and Attention

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
abstract

Dynamic programming (DP) solves a variety of structured combinatorial problems by iteratively breaking them down into smaller subproblems. In spite of their versatility, DP algorithms are usually non-differentiable, which hampers their use as a layer in neural networks trained by backpropagation. To address this issue, we propose to smooth the max operator in the dynamic programming recursion, using a strongly convex regularizer. This allows to relax both the optimal value and solution of the original combinatorial problem, and turns a broad class of DP algorithms into differentiable operators. Theoretically, we provide a new probabilistic perspective on backpropagating through these DP operators, and relate them to inference in graphical models. We derive two particular instantiations of our framework, a smoothed Viterbi algorithm for sequence prediction and a smoothed DTW algorithm for time-series alignment. We showcase these instantiations on two structured prediction tasks and on structured and sparse attention for neural machine translation.

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

Regularized Large Neighborhood Search

cs.LG · 2026-06-01 · unverdicted · novelty 7.0

RLNS regularizes LNS to perform block Gibbs sampling under entropy, interpolating between pseudolikelihood and exact MLE for differentiable combinatorial optimization.

citing papers explorer

Showing 1 of 1 citing paper.

  • Regularized Large Neighborhood Search cs.LG · 2026-06-01 · unverdicted · none · ref 34 · internal anchor

    RLNS regularizes LNS to perform block Gibbs sampling under entropy, interpolating between pseudolikelihood and exact MLE for differentiable combinatorial optimization.