DynLP: Parallel Dynamic Batch Update for Label Propagation in Semi-Supervised Learning

TL;DR

DynLP introduces a GPU-optimized dynamic batch update method for label propagation in semi-supervised learning, significantly reducing computation time for incremental data updates.

Contribution

It presents a novel GPU-centric algorithm that efficiently updates labels in graph-based semi-supervised learning without full recomputation.

Findings

Achieves up to 102x speedup on large datasets.

Performs only necessary updates, saving computational resources.

Outperforms state-of-the-art approaches in efficiency.

Abstract

Semi-supervised learning aims to infer class labels using only a small fraction of labeled data. In graph-based semi-supervised learning, this is typically achieved through label propagation to predict labels of unlabeled nodes. However, in real-world applications, data often arrive incrementally in batches. Each time a new batch appears, reapplying the traditional label propagation algorithm to recompute all labels is redundant, computationally intensive, and inefficient. To address the absence of an efficient label propagation update method, we propose DynLP, a novel GPU-centric Dynamic Batched Parallel Label Propagation algorithm that performs only the necessary updates, propagating changes to the relevant subgraph without requiring full recalculation. By exploiting GPU architectural optimizations, our algorithm achieves on average 13x and upto 102x speedup on large-scale datasets compared to state-of-the-art approaches.