Technology Category
Machine Learning: Evaluation and AnalysisSearch and Optimization: Mixed Discrete/Continuous SearchIntelligent Robots: State Estimation
Pulse Shape Discrimination Algorithms: Survey and Benchmark
🤖 AI Summary
Existing pulse shape discrimination (PSD) algorithms for radiation detection lack standardized benchmarks and reproducible evaluation protocols. Method: We establish a comprehensive, standardized evaluation framework encompassing over 60 PSD methods—including statistical, prior-knowledge-based, time-frequency, classical machine learning (ML), and deep learning approaches—and propose a hybrid deep architecture integrating statistical features with neural regression. Contribution/Results: We identify critical limitations of the conventional Figure of Merit (FOM) metric and advocate a multidimensional evaluation suite incorporating F1-score, ROC-AUC, and other task-relevant metrics. Experimental results demonstrate that multilayer perceptron (MLP) and the proposed hybrid model significantly outperform traditional PSD methods. To foster transparency and reproducibility, we publicly release a Python/MATLAB toolbox and a curated benchmark dataset, enabling fair, rigorous, and replicable comparison of PSD algorithms.
Technology Category
Application Category
📝 Abstract
This review presents a comprehensive survey and benchmark of pulse shape discrimination (PSD) algorithms for radiation detection, classifying nearly sixty methods into statistical (time-domain, frequency-domain, neural network-based) and prior-knowledge (machine learning, deep learning) paradigms. We implement and evaluate all algorithms on two standardized datasets: an unlabeled set from a 241Am-9Be source and a time-of-flight labeled set from a 238Pu-9Be source, using metrics including Figure of Merit (FOM), F1-score, ROC-AUC, and inter-method correlations. Our analysis reveals that deep learning models, particularly Multi-Layer Perceptrons (MLPs) and hybrid approaches combining statistical features with neural regression, often outperform traditional methods. We discuss architectural suitabilities, the limitations of FOM, alternative evaluation metrics, and performance across energy thresholds. Accompanying this work, we release an open-source toolbox in Python and MATLAB, along with the datasets, to promote reproducibility and advance PSD research.
Problem
Research questions and friction points this paper is trying to address.
Survey and benchmark pulse shape discrimination algorithms for radiation detection
Evaluate algorithms using standardized datasets and multiple performance metrics
Compare deep learning models with traditional methods in PSD performance
Innovation
Methods, ideas, or system contributions that make the work stand out.
Comprehensive survey of PSD algorithms
Benchmark using standardized datasets and metrics
Open-source toolbox for reproducibility
🔎 Similar Papers
No similar papers found.
