Predicting the accuracy of candidate neural architectures is an important capability of NAS-based solutions. When a candidate architecture has properties that are similar to other known architectures, the prediction task is rather straightforward using off-the-shelf regression algorithms. However, when a candidate architecture lies outside of the known space of architectures, a regression model has to perform extrapolated predictions, which is not only a challenging task, but also technically impossible using the most popular regression algorithm families, which are based on decision trees. In this work, we are trying to address two problems. The first one is improving regression accuracy using feature selection, whereas the other one is the evaluation of regression algorithms on extrapolating accuracy prediction tasks. We extend the NAAP-440 dataset with new tabular features and introduce NAAP-440e, which we use for evaluation. We observe a dramatic improvement from the old baseline, namely, the new baseline requires 3x shorter training processes of candidate architectures, while maintaining the same mean-absolute-error and achieving almost 2x fewer monotonicity violations, compared to the old baseline's best reported performance. The extended dataset and code used in the study have been made public in the NAAP-440 repository.
翻译:预测候选神经结构的准确性是NAS 基础解决方案的重要能力之一。 当候选结构具有与其他已知结构相似的特性时, 预测任务使用现成回归算法是相当直截了当的。 但是, 当候选结构位于已知建筑空间之外时, 回归模型必须进行外推预测, 这不仅是一项具有挑战性的任务, 而且技术上也不可能使用以决策树为基础的最受欢迎的回归算法家庭。 在这项工作中, 我们试图解决两个问题。 第一个是使用特征选择来提高回归精确性, 而另一个是使用外推精确预测任务的回归算法评估。 我们用新的表格功能扩展NAAP-440数据集, 并引入用于评估的NAAP-440e。 我们观察到, 与旧基线相比, 新的基线需要3x更短的候选结构培训过程, 同时保持相同的平均偏差-, 并且比旧的基线报告的最佳性能少了近2x单一度违反率。 我们观察到, NAAP 440 中所使用的扩展的数据设置和代码比旧的运行方式更精确。