论文概要
研究领域: ML 作者: Romain Amigon 发布时间: 2026-07-13 arXiv: 2607.11826
中文摘要
神经架构搜索（NAS）自动化了深度学习模型设计，但传统上需要海量计算资源，常以数千GPU-天计量。本文提出一种节俭且模因式的NAS框架，旨在消费者级硬件上普及架构设计。该方法结合自回归Transformer控制器的全局宏观搜索能力（通过强化学习训练）和人工蜂群（ABC）算法的局部微观利用。为防止RL阶段过早收敛，引入动态熵机制，在检测到性能停滞时强制拓扑探索。在标准GPU（NVIDIA RTX 3060）上评估，该混合方法有效解决了元启发式固有的’冷启动’问题。通过算法惩罚网络深度，框架主动缓解模型膨胀：在CIFAR-10上发现以约174000参数达到84.85%准确率的高效架构（显著小于ResNet-20等标准基线），搜索时间仅3小时。还展示了灵活性，应用于信用卡欺诈检测，直接在高度不平衡表格数据上优化F1-Score，以约4600参数的紧凑网络达到F1-Score 0.71。
原文摘要
Neural Architecture Search (NAS) has automated the design of deep learning models but traditionally requires massive computational resources, often measured in thousands of GPU-days. In this paper, we propose a frugal and memetic NAS framework designed to democratize architecture design on consumer-grade hardware. Our approach combines the global macro-search capabilities of an autoregressive Transformer controller, trained via Reinforcement Learning (RL), with the local micro-exploitation of an Artificial Bee Colony (ABC) algorithm. To prevent premature convergence during the RL phase, we introduce a dynamic entropy mechanism that forces topological exploration upon detection of performance stagnation. Evaluated on a standard GPU (NVIDIA RTX 3060), our hybrid method effectively resolves t…
— 自动采集于 2026-07-15
#论文 #arXiv #ML #小凯