AutoSpace：人工干预较少的神经结构搜索
AutoSpace: Neural Architecture Search with Less Human Interference
摘要：
目前的神经结构搜索（NAS）算法仍然需要专家知识和努力来设计一个搜索空间的网络建设。在本文中，我们考虑自动化的搜索空间的设计，以尽量减少人为干扰，但面临着两个挑战：爆炸性的复杂性的探索空间和昂贵的计算成本，以评估不同的搜索空间的质量。为了解决这些问题，我们提出了一种新的可微进化框架AutoSpace，它通过以下新技术将搜索空间进化到最优空间：可微适应度评分函数，以有效地评估细胞的性能和参考架构，以加快进化过程，避免陷入次优解。该框架是通用的，并与额外的计算约束兼容，使得学习适合不同计算预算的专用搜索空间成为可能。与使用以前手动设计的空间相比，使用学习的搜索空间，最近的NAS算法的性能可以得到显着改善。值得注意的是，新搜索空间生成的模型在移动环境下（MAds <500 M）在ImageNet上实现了77.8%的top-1准确率，比之前的SOTA EfficientNet-B 0高出0.7%。所有代码都将公开。
英文摘要：
Current neural architecture search (NAS) algorithms still require expert knowledge and effort to design a search space for network construction. In this paper, we consider automating the search space design to minimize human interference, which however faces two challenges: the explosive complexity of the exploration space and the expensive computation cost to evaluate the quality of different search spaces. To solve them, we propose a novel differentiable evolutionary framework named AutoSpace, which evolves the search space to an optimal one with following novel techniques: a differentiable fitness scoring function to efficiently evaluate the performance of cells and a reference architecture to speedup the evolution procedure and avoid falling into sub-optimal solutions. The framework is generic and compatible with additional computational constraints, making it feasible to learn specialized search spaces that fit different computational budgets. With the learned search space, the performance of recent NAS algorithms can be improved significantly compared with using previously manually designed spaces. Remarkably, the models generated from the new search space achieve 77.8% top-1 accuracy on ImageNet under the mobile setting (MAdds < 500M), out-performing previous SOTA EfficientNet-B0 by 0.7%. All codes will be made public.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
