Classification of Gold-Bearing Particles Using Visual Cues and Cost-Sensitive Machine Learning

    Research output: Contribution to journalArticle

    7 Citations (Scopus)

    Abstract

    © 2015, International Association for Mathematical Geosciences. Ore sorting increases the grade of an ore feed stream by separating very low-grade particles (‘waste’) from those containing higher concentrations of the desired mineral (‘ore’), thus economically reducing the amount of material processed in further mineral concentration steps. This paper reports a preliminary study that aims to develop an automated method for discriminating waste and gold-bearing particles. The study used both hyperspectral measurements and RGB images of waste and gold-bearing particles from the Sunrise Dam Gold Mine as input to the discriminating method. Advanced feature extraction methods were employed to capture visual cues such as texture and colour from the RGB images, which were combined with hyperspectral features to give nine types of representative features. Feature selection was applied to groups of the representative features and resulting feature subsets were evaluated using three machine learning algorithms, namely a support vector machine, a naïve Bayes classifier, and a majority decision table, to identify a highly informative subset of features. Cost-sensitive training was used to minimise the nominal profit lost due to sorting error based on real cost values from the milling process, with the aim of economically balancing the ore acceptance rate with the waste rejection rate. A cost-blind support vector machine achieved an ore acceptance rate of 84 % and a waste rejection rate of 87 %, which resulted in $0.98 nominal profit lost per tonne of crushed rock particles. Cost-sensitive training reduced the nominal profit lost to $0.34 per tonne, undercutting the costs associated with refining all particles by $0.24 per tonne.
    Original languageEnglish
    Pages (from-to)521-545
    JournalMathematical Geosciences
    Volume47
    Issue number5
    DOIs
    Publication statusPublished - Jul 2015

    Fingerprint

    Cost-sensitive Learning
    visual cue
    Gold
    Machine Learning
    gold
    Costs
    Categorical or nominal
    Profit
    cost
    Rejection
    Sorting
    sorting
    Support Vector Machine
    Bayes Classifier
    Decision Table
    Subset
    ore mineral
    gold mine
    extraction method
    Balancing

    Cite this

    @article{e1f37dc176da442e84ef659848bffd1f,
    title = "Classification of Gold-Bearing Particles Using Visual Cues and Cost-Sensitive Machine Learning",
    abstract = "{\circledC} 2015, International Association for Mathematical Geosciences. Ore sorting increases the grade of an ore feed stream by separating very low-grade particles (‘waste’) from those containing higher concentrations of the desired mineral (‘ore’), thus economically reducing the amount of material processed in further mineral concentration steps. This paper reports a preliminary study that aims to develop an automated method for discriminating waste and gold-bearing particles. The study used both hyperspectral measurements and RGB images of waste and gold-bearing particles from the Sunrise Dam Gold Mine as input to the discriminating method. Advanced feature extraction methods were employed to capture visual cues such as texture and colour from the RGB images, which were combined with hyperspectral features to give nine types of representative features. Feature selection was applied to groups of the representative features and resulting feature subsets were evaluated using three machine learning algorithms, namely a support vector machine, a na{\"i}ve Bayes classifier, and a majority decision table, to identify a highly informative subset of features. Cost-sensitive training was used to minimise the nominal profit lost due to sorting error based on real cost values from the milling process, with the aim of economically balancing the ore acceptance rate with the waste rejection rate. A cost-blind support vector machine achieved an ore acceptance rate of 84 {\%} and a waste rejection rate of 87 {\%}, which resulted in $0.98 nominal profit lost per tonne of crushed rock particles. Cost-sensitive training reduced the nominal profit lost to $0.34 per tonne, undercutting the costs associated with refining all particles by $0.24 per tonne.",
    author = "Tom Horrocks and Daniel Wedge and Eun-Jung Holden and Peter Kovesi and N. Clarke and John Vann",
    year = "2015",
    month = "7",
    doi = "10.1007/s11004-015-9597-7",
    language = "English",
    volume = "47",
    pages = "521--545",
    journal = "Mathematical Geosciences",
    issn = "1874-8953",
    publisher = "Springer",
    number = "5",

    }

    TY - JOUR

    T1 - Classification of Gold-Bearing Particles Using Visual Cues and Cost-Sensitive Machine Learning

    AU - Horrocks, Tom

    AU - Wedge, Daniel

    AU - Holden, Eun-Jung

    AU - Kovesi, Peter

    AU - Clarke, N.

    AU - Vann, John

    PY - 2015/7

    Y1 - 2015/7

    N2 - © 2015, International Association for Mathematical Geosciences. Ore sorting increases the grade of an ore feed stream by separating very low-grade particles (‘waste’) from those containing higher concentrations of the desired mineral (‘ore’), thus economically reducing the amount of material processed in further mineral concentration steps. This paper reports a preliminary study that aims to develop an automated method for discriminating waste and gold-bearing particles. The study used both hyperspectral measurements and RGB images of waste and gold-bearing particles from the Sunrise Dam Gold Mine as input to the discriminating method. Advanced feature extraction methods were employed to capture visual cues such as texture and colour from the RGB images, which were combined with hyperspectral features to give nine types of representative features. Feature selection was applied to groups of the representative features and resulting feature subsets were evaluated using three machine learning algorithms, namely a support vector machine, a naïve Bayes classifier, and a majority decision table, to identify a highly informative subset of features. Cost-sensitive training was used to minimise the nominal profit lost due to sorting error based on real cost values from the milling process, with the aim of economically balancing the ore acceptance rate with the waste rejection rate. A cost-blind support vector machine achieved an ore acceptance rate of 84 % and a waste rejection rate of 87 %, which resulted in $0.98 nominal profit lost per tonne of crushed rock particles. Cost-sensitive training reduced the nominal profit lost to $0.34 per tonne, undercutting the costs associated with refining all particles by $0.24 per tonne.

    AB - © 2015, International Association for Mathematical Geosciences. Ore sorting increases the grade of an ore feed stream by separating very low-grade particles (‘waste’) from those containing higher concentrations of the desired mineral (‘ore’), thus economically reducing the amount of material processed in further mineral concentration steps. This paper reports a preliminary study that aims to develop an automated method for discriminating waste and gold-bearing particles. The study used both hyperspectral measurements and RGB images of waste and gold-bearing particles from the Sunrise Dam Gold Mine as input to the discriminating method. Advanced feature extraction methods were employed to capture visual cues such as texture and colour from the RGB images, which were combined with hyperspectral features to give nine types of representative features. Feature selection was applied to groups of the representative features and resulting feature subsets were evaluated using three machine learning algorithms, namely a support vector machine, a naïve Bayes classifier, and a majority decision table, to identify a highly informative subset of features. Cost-sensitive training was used to minimise the nominal profit lost due to sorting error based on real cost values from the milling process, with the aim of economically balancing the ore acceptance rate with the waste rejection rate. A cost-blind support vector machine achieved an ore acceptance rate of 84 % and a waste rejection rate of 87 %, which resulted in $0.98 nominal profit lost per tonne of crushed rock particles. Cost-sensitive training reduced the nominal profit lost to $0.34 per tonne, undercutting the costs associated with refining all particles by $0.24 per tonne.

    U2 - 10.1007/s11004-015-9597-7

    DO - 10.1007/s11004-015-9597-7

    M3 - Article

    VL - 47

    SP - 521

    EP - 545

    JO - Mathematical Geosciences

    JF - Mathematical Geosciences

    SN - 1874-8953

    IS - 5

    ER -