用群体进化方法优化判别器，提升半监督GAN的训练稳定性和分类准确率。
Population-Based Multi-Objective Training of Discriminators for Semi-Supervised GANs
- 将判别器训练转化为多目标优化，保留分类与真假区分的权衡
- 在少量标签下训练更稳定，精英策略实现最高分类准确率
- 适合关注半监督生成模型稳定性与性能的科研人员
半监督生成对抗网络（SSL-GAN）可在利用大量无标签数据的同时保留判别器中的分类器，但其训练常不稳定。本文提出一种基于种群的进化训练策略，将判别器学习建模为多目标优化问题。不将监督与非监督损失合并为单一标量损失，而是维护一个由帕累托支配关系排序的判别器种群，以探索分类准确率与真实/虚假样本区分能力之间的不同权衡。该方法旨在同时提升SSL-GAN的两个功能：学习高精度分类器，并训练能生成逼真样本的生成器。我们分析了多种变体，包括精英策略和单目标消融实验，以评估多目标选择的作用。在标签数量有限的MNIST数据集上，相比SSL-GAN和CE-SSL-GAN等先进基线，本方法显著提升了训练鲁棒性；其中精英策略始终达到最高分类准确率。
原文摘要 · Abstract (English)
Semi-supervised generative adversarial networks (SSL-GANs) can exploit large unlabeled datasets while retaining a classifier in the discriminator, but their training is often unstable. This paper proposes a population-based evolutionary training strategy in which discriminator learning is formulated as a multi-objective optimization problem. Instead of aggregating the supervised and unsupervised components of the SSL objective into a single scalar loss, the method maintains a population of discriminators ranked by Pareto dominance, enabling the exploration of different trade-offs between classification accuracy and real/fake discrimination. This formulation aims to improve both roles of SSL-GANs: learning accurate classifiers and training generators capable of producing realistic samples. We analyze several variants, including an elitist strategy and a mono-objective ablation, to assess the role of multi-objective selection. Experiments on MNIST with limited labels show improved training robustness compared to SSL-GAN and CE-SSL-GAN state-of-the-art baselines, while the elitist variant consistently achieves the highest classification accuracy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。