Belle实验中神经元网络和Fisher判别方法应用于压制连续过程本底的比较

摘要

在Belle实验B±→K±ωφ道的物理分析中,对神经元网络法和Fisher判别法在压制连续过程本底应用中的性能进行了比较研究.应用神经元网络法可以获得对B(B)信号事例42.3%的选择效率和99.2%的连续过程本底排除水平;应用Fisher判别法可以获得57.9%的信号事例选择效率和98.6%的连续过程本底排除水平.%The neural networks method and Fisher discriminant method for continuum background suppression were applied to the analysis of B±→K±ωφ decay respectively.A comparison was conducted of the effects of the two methods.With neural network,42.3% of the B(B)events can be preserved while 99.2% of continuum background is rejected.With Fisher discriminant,57.9% of the signal can be preserved while 98.6% of the continuum background is rejected.