Stratum presents a paper about our neural network-based machine learning system and its application at the Jundee orogenic gold deposit, Yandal belt in Western Australia
Kriging frequently struggles to predict high variation and nuggety gold systems due to its linear interpolation nature of estimation. We design a neural network-based machine learning system to predict gold mineralization as a replacement to traditional domaining and kriging and apply it on an underground narrow vein gold deposit at a producing asset in the Northern Goldfields region in Western Australia. Our results show the machine learning can substantially improve mine planning and drill targeting for complex narrow vein gold deposits.