P6-2. Neural Network Analysis of Biostratigraphically Important Conodonts
Project leader: Tim Denham (SoAA-CASS, ANU)
Industry partner: Dr. Patrick Mark Smith, Australian Museum/UNSW
- Conodonts are extremely useful biostratigraphic markers of rich mineral bearing fossil sequences.
- MicroCT scanning of diverse conodont taxa in Australian Museum collections provide dataset for training of neural network to discriminate taxa.
- First step in designing a new system to use computer-derived biostratigraphic data to correlate and identify important rock units.
Alignment within M3D Innovation:
- Use of multiscale 3D imaging and neural network to design new computer-aided technologies for industry.
- Development of technology and trained personnel.
- Current methods to discriminate significant conodont species labour-intensive and 19th century technology.
- MicroCT scan, visualise and analyse taxonomically identified conodonts in Australian Museum collection.
- Train neural network to discriminate conodont taxa, with a focus on mineralogically significant biostratigraphic markers.
- Develop computer-aided and potentially semi-automated workflows to inform mineral exploration.
- Bring study of conodont taxonomy and biostratigraphy into the 21st century.
- Create digital database of taxonomically identified conodonts from existing collections.
- Train neural network in interspecific discrimination.
- Design computer-aided workflow to rapidly and reliably correlate economically important rock units using microCT scanning, data visualization and neural network.