Nuclide identification algorithm based on K–L transform and neural networks

Chen, Liang; Yixiang, Wei

doi:10.1016/j.nima.2008.09.035

Cited by 39 publications

(13 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In one of the earliest works, reported in 2001 [8], a neural network algorithm was employed in combination with a peak search procedure to reduce the network input data size, a limitation due to the reduced computational power at the time. Later works in 2008 [7] and 2009 [9] employed fully connected networks to perform nuclide identification, while a more recent work [10] claimed to have implemented a neural network algorithm that can determine the relative activities of radioisotopes in a large data set containing a mixture of multiple radioisotopes. All neural network techniques used for radioisotope identification are based on a variety of methods except for convolutional methods.…”

Section: Cnl Nuclear Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Machine-Learning Algorithm for Shielded Special Nuclear Materials Detection

Masala

Blomeley

2019

CNL Nuclear Review

View full text Add to dashboard Cite

A machine-learning algorithm has been implemented by use of a neural network as a preliminary study on the applicability of this method to special nuclear materials detection. The algorithm predicts the presence of the 238U isotope when learning from a gamma spectrum data measured with a high-purity germanium detector from a sample of depleted uranium. In this work, both a fully connected neural network and a convolutional neural network have been implemented, and the performance of different configurations of the network has been studied. The use of convolutional network showed better performance over the fully connected network, with cost function and success rate values supporting a better prediction while avoiding overfitting. Furthermore, implemented network features such as filtering, max-pooling, dropout regularization, and momentum optimization also showed improved prediction performance.

show abstract

Section: Cnl Nuclear Reviewmentioning

confidence: 99%

“…The efficiency of this approach may depend on the resolution of data available from detectors, but the general approach will be applicable to this or other scanning regimes at borders or elsewhere. Machine learning for radioisotope identification has been attempted already based on neural networks [7][8][9][10].…”

Section: Introductionmentioning

confidence: 99%

Machine-Learning Algorithm for Shielded Special Nuclear Materials Detection

Masala

Blomeley

2019

CNL Nuclear Review

View full text Add to dashboard Cite

show abstract

“…ANNs have been applied to peak fitting [31], isotope identification [2,3], and activity estimation [2,32]. Many of this work rely on ROI methods [33], feature extraction [34], high-resolution gamma-ray spectra as the input to the ANN [4], small libraries of isotopes, and assume perfectly calibrated detectors. ANN training methods created for high-resolution gamma-ray spectra may not perform well when trained using low-resolution spectra given the large discrepancy in resolution.…”

Section: Existing Neural Network Applications To Isotope Identificatimentioning

confidence: 99%

Automated Isotope Identification Algorithm Using Artificial Neural Networks

Kamuda

Stinnett

Sullivan

2017

IEEE Trans. Nucl. Sci.

View full text Add to dashboard Cite

There is a need to develop an algorithm that can determine the relative activities of a mixture of many isotopes in a low-resolution gamma-ray spectrum. While techniques for this task exist, they require a human operator and are too slow to use on very large datasets of spectra. Pattern recognition algorithms such as neural networks are prime candidates for automated isotope identification using low-resolution gamma-ray spectra. While algorithms based on feature extraction such as peak finding or ROI algorithms work well for well calibrated high resolution detectors, for low-resolution detectors it may be more beneficial to use algorithms that incorporate more abstract features of the spectrum. This is especially true when analyzing a mixture of isotopes where peak overlap and Compton continuum effects occlude features of interest. To solve this, an artificial neural network (ANN) was trained to predict the presence and relative activities of isotopes from a mixture of many isotopes. The

show abstract

“…ANN eliminates the limitations of classical approaches by extracting the desired information from the input data. Applying ANN to a spectrometry system needs sufficient input and output data instead of mathematical equations for performing the fit to nuclear spectra, including X-, gamma-ray and alpha-particles spectra (Baeza et al, 2011;Basheer and Hajmeer, 2000;Keller et al, 1995;Yoshida et al, 2002;Kangas et al, 2008;Chen and Wei, 2009;Medhat, 2012;Miranda et al, 2009;Doostmohammadi et al, 2010). For each nuclear spectrum, such as alpha spectrum, up to 2048 data points are selected as inputs.…”

Section: Introductionmentioning

confidence: 99%