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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Mixed-precision quantization is a powerful tool to enable memory and computesavings of neural network workloads by deploying different sets of bit-widthprecisions on separate compute operations. In this work, we present a flexibleand scalable framework for automated mixed-precision quantization thatconcurrently optimizes task performance, memory compression, and computesavings through multi-objective evolutionary computing. Our framework centerson Neuroevolution-Enhanced Multi-Objective Optimization (NEMO), a novel searchmethod, which combines established search methods with the representationalpower of neural networks. Within NEMO, the population is divided intostructurally distinct sub-populations, or species, which jointly create thePareto frontier of solutions for the multi-objective problem. At eachgeneration, species perform separate mutation and crossover operations, and arere-sized in proportion to the goodness of their contribution to the Paretofrontier. In our experiments, we define a graph-based representation todescribe the underlying workload, enabling us to deploy graph neural networkstrained by NEMO via neuroevolution, to find Pareto optimal configurations forMobileNet-V2, ResNet50 and ResNeXt-101-32x8d. Compared to the state-of-the-art,we achieve competitive results on memory compression and superior results forcompute compression. Further analysis reveals that the graph representation andthe species-based approach employed by NEMO are critical to finding optimalsolutions.

Neuroevolution-Enhanced Multi-Objective Optimization for Mixed-Precision Quantization