English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Neural construction models have shown promising performance for VehicleRouting Problems (VRPs) by adopting either the Autoregressive (AR) orNon-Autoregressive (NAR) learning approach. While AR models producehigh-quality solutions, they generally have a high inference latency due totheir sequential generation nature. Conversely, NAR models generate solutionsin parallel with a low inference latency but generally exhibit inferiorperformance. In this paper, we propose a generic Guided Non-AutoregressiveKnowledge Distillation (GNARKD) method to obtain high-performance NAR modelshaving a low inference latency. GNARKD removes the constraint of sequentialgeneration in AR models while preserving the learned pivotal components in thenetwork architecture to obtain the corresponding NAR models through knowledgedistillation. We evaluate GNARKD by applying it to three widely adopted ARmodels to obtain NAR VRP solvers for both synthesized and real-world instances.The experimental results demonstrate that GNARKD significantly reduces theinference time (4-5 times faster) with acceptable performance drop (2-3\%). Tothe best of our knowledge, this study is first-of-its-kind to obtain NAR VRPsolvers from AR ones through knowledge distillation.

Distilling Autoregressive Models to Obtain High-Performance Non-Autoregressive Solvers for Vehicle Routing Problems with Faster Inference Speed