2022
DOI: 10.1016/j.forsciint.2021.111146
|View full text |Cite
|
Sign up to set email alerts
|

Recognition of gasoline in fire debris using machine learning: Part I, application of random forest, gradient boosting, support vector machine, and naïve bayes

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
7
0
1

Year Published

2022
2022
2024
2024

Publication Types

Select...
8
1

Relationship

0
9

Authors

Journals

citations
Cited by 22 publications
(8 citation statements)
references
References 12 publications
0
7
0
1
Order By: Relevance
“…Bogdal et al presented a two-part study on recognizing gasoline in fire debris using machine learning [ 53 , 54 ]. Part 1 focused on the development of a machine learning tool using various algorithms.…”
Section: Fire Debris Analysismentioning
confidence: 99%
“…Bogdal et al presented a two-part study on recognizing gasoline in fire debris using machine learning [ 53 , 54 ]. Part 1 focused on the development of a machine learning tool using various algorithms.…”
Section: Fire Debris Analysismentioning
confidence: 99%
“…Aliaño-González et al [65] applied PCA and LDA to GC data of firefighters' coats to discriminate those exposed to fire from the non-exposed ones, in order to identify toxic substances to which firefighters can be exposed. Two interesting works of Bogdal et al [66,67] focused on tracing gasoline residues in fire debris, with the aim of identifying arsons. These works tested the use of several machine learning methods to discriminate debris with the presence or absence of ignitable liquid trace.…”
Section: Gas Chromatography (Gc) and Chemometricsmentioning
confidence: 99%
“…Similar studies, but ones which only took into account one honey class (Quercus ilex honeydew and pine honeys) and focused on geographical origin, with samples coming from several regions of Greece and Turkey, respectively, were carried out by Karabagias et al [36] and Duru et al [37] by PCA and stepwise Machine learning is a "new" frontier of chemometrics in which models are computed in an iterative way, with the computation that "learns" from data: once a model has been computed, the calculation starts again using the previous results as starting points, instead of the original data, and this procedure is carried out iteratively until a satisfactory result, or a convergence, is reached. Bogdal et al tested random forest, gradient boosting, support vector machine, naïve bayes, logistic regression [66] and convolutional neural networks on GC-MS spectra converted to images [67]. Satisfactory results were obtained for most of these methods, except for logistic regression and naïve bayes, for which, probably, there were not enough samples.…”
Section: Gas Chromatography (Gc) and Chemometricsmentioning
confidence: 99%
“…The structure of this model focuses on a set of independently trained decision models, combining their predictions to generate the overall prediction, thus reducing the probability of errors in decision-making. Some examples of this type of algorithms are gradient boosting [80] and AdaBoosting [81].…”
Section: Decision Matrixmentioning
confidence: 99%