arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2207.09390cs.LGstat.ML
收藏

用于特征选择的神经贪婪追踪

Neural Greedy Pursuit for Feature Selection

KTH Royal Institute of Technology(瑞典皇家理工学院);Weizmann Institute of Science(魏茨曼科学研究所)

Sandipan Das, Alireza M. Javid, Prakash Borpatra Gohain, Yonina C. Eldar, Saikat Chatterjee

AI总结:

提出神经贪婪追踪(NGP)算法,通过神经网络预测器在迭代损失最小化中逐一顺序选择特征,有效解决非线性预测中 $N \ll P$ 的特征选择问题,性能优于多种方法并展现相变行为。

摘要:

我们提出一种贪婪算法,针对非线性预测问题,从 $P$ 个输入特征中选出 $N$ 个重要特征。特征在迭代的损失最小化过程中被逐一顺序选出。我们在算法中使用神经网络作为预测器来计算损失,因此将该方法称为神经贪婪追踪(NGP)。当 $N \ll P$ 时,NGP 能高效选出 $N$ 个特征,并按顺序选择过程以降序提供特征重要性的概念。实验表明,NGP 的性能优于 DeepLIFT 和 Drop-one-out loss 等多种特征选择方法。此外,实验还展示了一种相变行为:当训练数据规模超过某阈值时,可以完美选出全部 $N$ 个特征且无误报。

英文摘要:

We propose a greedy algorithm to select $N$ important features among $P$ input features for a non-linear prediction problem. The features are selected one by one sequentially, in an iterative loss minimization procedure. We use neural networks as predictors in the algorithm to compute the loss and hence, we refer to our method as neural greedy pursuit (NGP). NGP is efficient in selecting $N$ features when $N \ll P$, and it provides a notion of feature importance in a descending order following the sequential selection procedure. We experimentally show that NGP provides better performance than several feature selection methods such as DeepLIFT and Drop-one-out loss. In addition, we experimentally show a phase transition behavior in which perfect selection of all $N$ features without false positives is possible when the training data size exceeds a threshold.

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。

↑