Streaming Structured Inference with Flash-SemiCRF

TL;DR

This paper introduces Flash-SemiCRF, a memory-efficient method for exact semi-Markov CRF inference on long sequences, enabling scalable segment-level labeling with principled uncertainty estimates.

Contribution

It replaces large edge potential tensors with on-the-fly prefix-sum evaluations and employs streaming algorithms to handle long sequences efficiently.

Findings

Reduces memory footprint by replacing edge tensors with prefix-sum lookup.

Enables exact inference on sequences exceeding 100,000 positions.

Provides a GPU-accelerated implementation with sublinear memory usage.

Abstract

Semi-Markov Conditional Random Fields (semi-CRFs) assign labels to segments of a sequence rather than to individual positions, enabling exact inference over segment-level features and principled uncertainty estimates at their boundaries. However, existing implementations must materialize a large edge potential tensor whose size grows with sequence length, maximum segment length, and label count, becoming prohibitive for speech-scale state spaces and intractable at genomic scales where sequences can exceed 100,000 positions. This memory bottleneck has limited the adoption of exact segment-level inference for long sequences and large label sets. We identify that the core inefficiency is materializing edge potentials that can instead be evaluated on-the-fly from a compact prefix-sum array, and make several improvements. First, replacing the stored edge tensor with prefix-sum lookup reduces the memory footprint by a factor proportional to the product of segment length and label count. Second, a streaming forward-backward pass with checkpoint-boundary normalization keeps working memory sublinear in sequence length while preserving exact gradients. Third, zero-centered cumulative scores control numerical drift and induce an adaptive duration prior under label imbalance. We integrate these ideas into Flash-SemiCRF, a fused Triton kernel that enables exact semi-CRF inference on previously intractable problem sizes. Available at https://github.com/biobenkj/flash-semicrf.