Learning Boolean Circuits from Examples for Approximate Logic Synthesis
Learning Boolean Circuits from Examples for Approximate Logic Synthesis
复制标题
DOI：
10.1145/3394885.3431559
复制
发表时间：
2021-01
期刊：
影响因子：
--
通讯作者：
Sina Boroumand;C. Bouganis;G. Constantinides
中科院分区：
文献类型：
--
作者：
Sina Boroumand;C. Bouganis;G. Constantinides
许多计算应用程序天生具有容错性。因此，可以降低计算精度以实现面积、性能和/或能耗方面的更高效率。近年来，人们提出了一系列自动近似计算技术；然而，大多数这些技术需要完全了解一个精确的，或“黄金”电路描述。相反，最近人们对从实例中合成计算产生了浓厚的兴趣，这是一种监督学习的形式。在本文中，我们探讨了布尔电路的监督学习与现有的合成不完全指定函数的工作之间的关系。我们表明，当从机器学习的角度考虑时，后一种工作提供了良好的训练精度，但测试精度较差。我们将此与20世纪90年代的先前工作进行对比，该工作使用互信息来引导搜索过程，旨在实现良好的泛化。通过将这项早期工作与最近学习逻辑函数的方法相结合，我们能够在面积/延迟/测试错误权衡方面为布尔电路实现可扩展且高效的机器学习方法。
Many computing applications are inherently error resilient. Thus, it is possible to decrease computing accuracy to achieve greater efficiency in area, performance, and/or energy consumption. In recent years, a slew of automatic techniques for approximate computing has been proposed; however, most of these techniques require full knowledge of an exact, or ‘golden’ circuit description. In contrast, there has been significant recent interest in synthesizing computation from examples, a form of supervised learning. In this paper, we explore the relationship between supervised learning of Boolean circuits and existing work on synthesizing incompletely-specified functions. We show that when considered through a machine learning lens, the latter work provides a good training accuracy but poor test accuracy. We contrast this with prior work from the 1990s which uses mutual information to steer the search process, aiming for good generalization. By combining this early work with a recent approach to learning logic functions, we are able to achieve a scalable and efficient machine learning approach for Boolean circuits in terms of area/delay/test-error trade-off.