Scholarlink

<p>Machine learning techniques are increasingly used in high-energy nuclear physics because they can exploit multivariate correlations more efficiently than conventional cut-based analyses. A central challenge is the construction of training samples that faithfully reproduce the detector response observed in data. Signal samples are usually derived from detector simulations; therefore, mismatches between simulation and data can degrade classifier performance and introduce systematic biases. This work presents two practical correction procedures, namely cumulative distribution function (CDF) mapping and a shift-and-scale transformation, to align simulated signal features with those measured in data. Their performance is demonstrated with <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(J/\psi\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>J</mi> <mo stretchy="false">/</mo> <mi>ψ</mi> </mrow> </math></EquationSource> </InlineEquation> yield measurements in <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(\sqrt{s_\textrm{NN}}=200\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msqrt> <msub> <mi>s</mi> <mtext>NN</mtext> </msub> </msqrt> <mo>=</mo> <mn>200</mn> </mrow> </math></EquationSource> </InlineEquation> GeV Ru+Ru and Zr+Zr collisions recorded by STAR. A set of self-consistency tests shows that these procedures substantially suppress the systematic bias associated with data-simulation discrepancies in machine-learning-based signal extraction.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reducing systematic bias in machine learning applications to J/ψ signal extraction in high-energy nuclear physics

  • 期刊论文

  • 发表日期：

    2026年5月20日

查看全文
  • Yan Wang,
  • Rong-Rong Ma,
  • Kai-Feng Shen,
  • Ze-Bo Tang,
  • Wang-Mei Zha

摘要

Machine learning techniques are increasingly used in high-energy nuclear physics because they can exploit multivariate correlations more efficiently than conventional cut-based analyses. A central challenge is the construction of training samples that faithfully reproduce the detector response observed in data. Signal samples are usually derived from detector simulations; therefore, mismatches between simulation and data can degrade classifier performance and introduce systematic biases. This work presents two practical correction procedures, namely cumulative distribution function (CDF) mapping and a shift-and-scale transformation, to align simulated signal features with those measured in data. Their performance is demonstrated with \(J/\psi\) J / ψ yield measurements in \(\sqrt{s_\textrm{NN}}=200\) s NN = 200 GeV Ru+Ru and Zr+Zr collisions recorded by STAR. A set of self-consistency tests shows that these procedures substantially suppress the systematic bias associated with data-simulation discrepancies in machine-learning-based signal extraction.