Article
Computer Science, Interdisciplinary Applications
Maarten J. Van Strien, Adrienne Gret-Regamey
Summary: The identification of landscape classes is crucial for implementing planning strategies. Existing unsupervised clustering techniques often rely on categorical data and have limitations in quantifying landscape patterns. This study proposes a new unsupervised deep learning method (DCEC) to generate a landscape typology for Switzerland, which successfully distinguishes 45 landscape classes using continuous spatial data.
ENVIRONMENTAL MODELLING & SOFTWARE
(2022)
Article
Engineering, Aerospace
O. Bektas
Summary: The purpose of this paper is to group flight data phases based on distinctive sensor readings and represent the input space as a two-dimensional cluster map. The research design uses a self-organising map framework to provide organised representations of flight signal features. The findings show a significant correlation between monitored flight data signals and flight phases, and the clusters of flight regimes can be determined and observed on the maps. The contribution of this research is the grouping of real data flows for aircraft monitoring and visualising the evolution of monitored signals on a real aircraft.
AERONAUTICAL JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Rui Zhao, Jianfei Ruan, Bo Dong, Li Meng, Weizhan Zhang
Summary: This paper proposes a spatial consistency-based clustering (SCC) method to improve unsupervised image clustering performance. By retaining the alignment of representations learned from instance and class levels and effectively selecting reliable samples, SCC outperforms current state-of-the-art methods on benchmark datasets.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Water Resources
Sabrine Derouiche, Cecile Mallet, Abdelwaheb Hannachi, Zoubeida Bargaoui
Summary: This study analyzes the variability of precipitation in northern Tunisia using a variable time step and the concept of rain events. Through clustering methods, different rainfall patterns are identified, and the characteristics of wet spells and regionalization results are obtained. The findings of this study are important for understanding the precipitation patterns and water resource management in the region.
JOURNAL OF HYDROLOGY-REGIONAL STUDIES
(2022)
Article
Engineering, Environmental
Emmanouil A. Varouchakis, Dimitri Solomatine, Gerald A. Corzo Perez, Seifeddine Jomaa, George P. Karatzas
Summary: Successful modelling of groundwater level variations in complex aquifer systems requires integration of geostatistics and machine learning approaches. This study focuses on cases with large and randomly distributed datasets in different aquifer types. Self-Organizing Maps are used to identify locally similar data inputs and substitute the uncertain correlation length of the variogram model. Transgaussian Kriging is then applied to estimate the bias-corrected spatial distribution of groundwater level. The proposed methodology shows a significant improvement compared to classical geostatistical approaches.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2023)
Article
Computer Science, Information Systems
Li Zhang, Tong Qiao, Ming Xu, Ning Zheng, Shichuang Xie
Summary: This paper proposes a novel unsupervised detection method for identifying deepfake videos, which achieves good detection performance by extracting PRNU fingerprints and noise features from the videos for clustering analysis.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Information Systems
Matthew Coulson, Christos Ferles, Simon Winberg, Kevin J. Naidoo
Summary: The Growing Hierarchical Self-Organising Representation Map (GHSORM) is a model that combines denoising autoencoder and Growing Hierarchical Self-Organising Map algorithms to represent datasets and cluster input data. It shows the ability to subgroup clusters that cannot be fully separated by a single SOM. The model is applied to clustering handwritten digits and complex digital gene expression data, outperforming linear methods and its constituent algorithms.
INFORMATION SCIENCES
(2023)
Article
Biochemistry & Molecular Biology
Jorge Fernandez-de-Cossio-Diaz, Guido Uguzzoni, Andrea Pagnani
Summary: A computational method is developed to study the relationship between genotype and fitness, trained on sequencing samples from multiple rounds of screening experiments and tested on large-scale mutational scans. The inferred fitness landscape is robust and exhibits high generalization power.
MOLECULAR BIOLOGY AND EVOLUTION
(2021)
Article
Computer Science, Artificial Intelligence
Raymond Moodley, Francisco Chiclana, Fabio Caraffini, Mario Gongora
Summary: This paper introduces a prediction and containment model for pandemics and natural disasters, which combines selective lockdowns and protective cordons to rapidly contain hazards while keeping some regions economically active.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Shuo Li, Fang Liu, Licheng Jiao, Puhua Chen, Lingling Li
Summary: This paper proposes a deep learning-based clustering method called S(3)OCNet, which achieves joint learning of feature extraction and feature clustering through self-supervised learning, thus realizing a single-stage clustering method. The method significantly improves the performance on multiple image classification benchmarks and the effectiveness of the clustering results is validated through feature and image visualization.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2022)
Article
Computer Science, Information Systems
Kunkun Wu, Zhong Xie, Maosheng Hu
Summary: In this study, an unsupervised framework is proposed for extracting multilane roads from the OpenStreetMap dataset. By grouping and classifying road polygons, as well as applying post-processing techniques, multilane roads can be effectively extracted without the need for manually labeled data, achieving accuracy levels comparable to supervised methods.
INTERNATIONAL JOURNAL OF GEOGRAPHICAL INFORMATION SCIENCE
(2022)
Article
Biochemical Research Methods
Yang Xu, Rachel Patton McCord
Summary: The study introduces a new method called CoSTA, which uses convolutional neural network clustering to learn spatial relationships between gene expression matrices, emphasizing broader spatial patterns. CoSTA provides a quantitative measure of expression pattern similarity between each pair of genes, identifying narrower but biologically relevant sets of significantly related genes compared to other approaches.
BMC BIOINFORMATICS
(2021)
Article
Engineering, Electrical & Electronic
Yihong Cao, Hui Zhang, Xiao Lu, Yurong Chen, Zheng Xiao, Yaonan Wang
Summary: Unsupervised domain adaptation is an effective approach for solving the labeling difficulties in semantic segmentation tasks. A novel clustering-based method is proposed, which uses an adaptive refining-aggregation-separation framework to learn discriminative features for different domains and features. This method does not require tunable thresholds and includes techniques such as adaptive refinement, feature evaluation, and different losses for improving segmentation performance. Experimental results on benchmark datasets show that the proposed method outperforms existing state-of-the-art methods.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
Mohammadreza Khajeh Hosseini, Alireza Talebpour, Saipraneeth Devunuri, Samer H. Hamdar
Summary: This research proposes a new method for collecting trajectory data and uses clustering analysis to identify vehicle trajectories with similar adaptive cruise control behavior.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Keyla Gonzalez, Siddharth Misra
Summary: This study developed an unsupervised-learning-based visualization method for subsurface CO2 plume that can adapt and scale based on the data without assuming a geophysical model. A multi-level clustering approach was used to accurately differentiate CO2-bearing regions from non-CO2 bearing regions and further classify regions with different levels of CO2 content. The proposed method was validated in a real-world project and demonstrated high quality and reliability.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Geochemistry & Geophysics
Shawn B. Hood, Matthew J. Cracknell, Michael F. Gazley, Anya M. Reading
Article
Geochemistry & Geophysics
Cassady L. Harraden, Matthew J. Cracknell, James Lett, Ron F. Berry, Ronell Carey, Anthony C. Harris
Article
Geochemistry & Geophysics
Daniel D. Gregory, Mathew J. Cracknell, Ross R. Large, Peter McGoldrick, Stephen Kuhn, Valew V. Maslennikov, Michael J. Baker, Nathan Fox, Ivan Belousov, Maria C. Figueroa, Jeffrey A. Steadman, Adrian J. Fabris, Timothy W. Lyons
Article
Geology
Stephen Kuhn, Matthew J. Cracknell, Anya M. Reading
ORE GEOLOGY REVIEWS
(2019)
Article
Geochemistry & Geophysics
Esmaeil Eshaghi, Anya M. Reading, Michael Roach, Mark Duffett, Daniel Bombardieri, Matthew J. Cracknell, John L. Everard, Grace Cumming, Stephen Kuhn
Article
Geochemistry & Geophysics
Stephen Kuhn, Matthew J. Cracknell, Anya M. Reading, Stephanie Sykora
Article
Geology
Esmaeil Eshaghi, Anya M. Reading, Michael Roach, Mark Duffett, Daniel Bombardieri, Matthew J. Cracknell, John L. Everard
ORE GEOLOGY REVIEWS
(2020)
Article
Geochemistry & Geophysics
Rocky D. Barker, Shaun L. L. Barker, Matthew J. Cracknell, Elizabeth D. Stock, Geoffrey Holmes
Summary: In this study, LWIR spectra of hydrothermally altered carbonate rock core samples were analyzed using Random Forest machine learning approach to predict mineral species and abundances. The Random Forest models showed comparable accuracy to traditional spectral unmixing techniques, providing a more robust and meaningful interpretation of LWIR spectra. This new approach has the potential to improve the accuracy and speed of infrared data interpretation for various deposit types.
Article
Geology
Richen Zhong, Yi Deng, Wenbo Li, Leonid Danyushevsky, Matthew J. Cracknell, Ivan Belousov, Yanjing Chen, Lamei Li
Summary: Classification algorithms based on pyrite trace elements were developed using support vector machine (SVM) and artificial neural network (ANN) to distinguish the genesis of pyrites from different types of mineral deposits. The study demonstrated the effectiveness of these algorithms in decoding the geochemical data of pyrite, showcasing the potential of machine learning in geological research.
ORE GEOLOGY REVIEWS
(2021)
Article
Geology
Michael Gazley, Shawn B. Hood, Matthew J. Cracknell
Summary: The main goal of mineral exploration is to narrow down the search area, identify regions for further investigation, and drill potential targets. This study utilizes machine-learned map products to interpret a regional soil sample dataset and identify exploration targets.
ORE GEOLOGY REVIEWS
(2021)
Article
Geochemistry & Geophysics
Daniel Bombardieri, Mark Duffett, Andrew McNeill, Matthew Cracknell, Anya Reading
Summary: Over the past two decades, Mineral Resources Tasmania has developed regional 3D geological and geophysical models to improve the understanding of controls on ore-forming processes and prospectivity. These models, based on high-quality potential field data sets, are crucial for 3D modeling workflows that allow rapid hypothesis testing while satisfying the constraints of observed data.
Article
Agriculture, Multidisciplinary
Laura N. Sotomayor, Matthew J. Cracknell, Robert Musk
Summary: This research investigates the use of supervised machine learning to model and predict forest productivity across pine plantations in northern Tasmania, Australia. The study found that rainfall is the most important factor driving forest productivity.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Engineering, Chemical
Javier Merrill-Cifuentes, Matthew J. Cracknell, Angela Escolme
Summary: The balance between the valuable metal content and the cost of obtaining a commercial product from an ore body is crucial for its exploitation. Rock texture, although difficult to model and quantify, plays an important role in various stages of the business. This study applied the Mineral Co-Occurrence Probability Fields method to a large dataset of hyperspectral imaging to incorporate textural features into the predictive modeling of rock hardness and copper recovery. The results showed a significant improvement in precision and accuracy of the predictions, suggesting that acquiring hyperspectral imagery from drill cores can enhance the forecasting of metallurgical parameters reliant on rock textures.
MINERALS ENGINEERING
(2023)
Proceedings Paper
Geology
Daniel D. Gregory, Chao Liu, Shaunna M. Morrison, Robert M. Hazen, Mathew J. Cracknell, Ross R. Large, Peter McGoldrick, Stephen Kuhn, Michael J. Baker, Nathan Fox, Ivan Belousov, Jeffery A. Steadman, Adrian J. Fabris, Valery V. Maslennikov, Timothy W. Lyons, Maria C. Figueroa
LIFE WITH ORE DEPOSITS ON EARTH, PROCEEDINGS OF THE 15TH SGA BIENNIAL MEETING, 2019, VOLS 1-4
(2019)
Proceedings Paper
Geology
Sibele C. Nascimento, Anita Parbhakar-Fox, Matthew J. Cracknell, Wei Xuen Heng
LIFE WITH ORE DEPOSITS ON EARTH, PROCEEDINGS OF THE 15TH SGA BIENNIAL MEETING, 2019, VOLS 1-4
(2019)