Massey Documents by Type

Permanent URI for this communityhttps://mro.massey.ac.nz/handle/10179/294

Browse

Start date: 2020Subject: search.filters.subject.Classification

Search Results

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Novel machine learning-driven comparative analysis of CSP, STFT, and CSP-STFT fusion for EEG data classification across multiple meditation and non-meditation sessions in BCI pipeline.
    (BioMed Central Ltd, 2025-02-08) Liyanagedera ND; Bareham CA; Kempton H; Guesgen HW
    This study focuses on classifying multiple sessions of loving kindness meditation (LKM) and non-meditation electroencephalography (EEG) data. This novel study focuses on using multiple sessions of EEG data from a single individual to train a machine learning pipeline, and then using a new session data from the same individual for the classification. Here, two meditation techniques, LKM-Self and LKM-Others were compared with non-meditation EEG data for 12 participants. Among many tested, three BCI pipelines we built produced promising results, successfully detecting features in meditation/ non-meditation EEG data. While testing different feature extraction algorithms, a common neural network structure was used as the classification algorithm to compare the performance of the feature extraction algorithms. For two of those pipelines, Common Spatial Patterns (CSP) and Short Time Fourier Transform (STFT) were successfully used as feature extraction algorithms where both these algorithms are significantly new for meditation EEG. As a novel concept, the third BCI pipeline used a feature extraction algorithm that fused the features of CSP and STFT, achieving the highest classification accuracies among all tested pipelines. Analyses were conducted using EEG data of 3, 4 or 5 sessions, totaling 3960 tests on the entire dataset. At the end of the study, when considering all the tests, the overall classification accuracy using SCP alone was 67.1%, and it was 67.8% for STFT alone. The algorithm combining the features of CSP and STFT achieved an overall classification accuracy of 72.9% which is more than 5% higher than the other two pipelines. At the same time, the highest mean classification accuracy for the 12 participants was achieved using the pipeline with the combination of CSP STFT algorithm, reaching 75.5% for LKM-Self/ non-meditation for the case of 5 sessions of data. Additionally, the highest individual classification accuracy of 88.9% was obtained by the participant no. 14. Furthermore, the results showed that the classification accuracies for all three pipelines increased with the number of training sessions increased from 2 to 3 and then to 4. The study was successful in classifying a new session of EEG meditation/ non-meditation data after training machine learning algorithms using a different set of session data, and this achievement will be beneficial in the development of algorithms that support meditation.
  • Thumbnail Image
    Item
    Inferring arsenic anomalies indirectly using airborne hyperspectral imaging – Implication for gold prospecting along the Rise and Shine Shear Zone in New Zealand
    (Elsevier B V, 2024-08-01) Chakraborty R; Kereszturi G; Pullanagari R; Craw D; Durance P; Ashraf S
    Well-exposed mineral deposits are scarce at a global level and presently potential mineral-rich sites are underlying vegetation cover and topsoil, which are suboptimal for direct remote sensing based exploration techniques. This study aims to implement an indirect approach to arsenic (As) distribution mapping using the surface manifestations of the subsurface geology and link it to the known gold mineralisation in the study area. Rise and Shine Shear Zone (RSSZ) in New Zealand is broadly a part of the Otago schist hosting lower to upper green-schist facies rocks manifesting mesothermal gold mineralisation. The area has several surficial geological imprints separating mineralised and non-mineralised zones, but these are dominated by topographic ruggedness, soil moisture and vegetation (mainly grass/tussock) spectra in the hyperspectral data. Initially, a band selection using Recursive Feature Elimination (RFE) was executed. The bands generated were tallied with the field and geological understanding of the area. The resultant 85 bands were then further put through Orthogonal Total Variation Component Analysis (OTVCA) to concise the information in 10 bands. OTVCA output was then classified using Random Forest classifier to map three levels of As concentration (<20 ppm, between 20 and 100 ppm and >100 ppm). The potentially high As concentration zones are likely to be related to the gold mineralisation. The geology of the area correlates with soil exposure which is captured by the classification in some parts, this increases the accuracy but also makes the classification analysis challenging.
  • Thumbnail Image
    Item
    An investigation of the imputation techniques for missing values in ordinal data enhancing clustering and classification analysis validity
    (Elsevier Inc, 2023-12) Alam S; Ayub MS; Arora S; Khan MA
    Missing data can significantly impact dataset integrity and suitability, leading to unreliable statistical results, distortions, and poor decisions. The presence of missing values in data introduces inaccuracies in clustering and classification and compromises the reliability and validity of such analyses. This study investigates multiple imputation techniques specifically designed for handling missing values in ordinal data commonly encountered in surveys and questionnaires. Quantitative approaches are used to evaluate different imputation methods on various datasets with varying missing value percentages. By comparing the performance of imputation techniques using clustering metrics and algorithms (e.g., k-means, Partitioning Around Medoids), the study provides valuable insights for selecting appropriate imputation methods for accurate data analysis. Furthermore, the study examines the impact of imputed values on classification algorithms, including k-Nearest Neighbors (kNN), Naive Bayes (NB), and Multilayer Perceptron (MLP). Results demonstrate that the decision tree method is the most effective approach, closely aligning with the original data and achieving high accuracy. In contrast, random number imputation performs poorly, indicating limited reliability. This study advances the understanding of handling missing values and emphasizes the need to address this issue to enhance data analysis integrity and validity.
  • Thumbnail Image
    Item
    Common spatial pattern for classification of loving kindness meditation EEG for single and multiple sessions
    (BioMed Central Ltd, Springer Nature, 2023-09-09) Liyanagedera ND; Hussain AA; Singh A; Lal S; Kempton H; Guesgen HW
    While a very few studies have been conducted on classifying loving kindness meditation (LKM) and non-meditation electroencephalography (EEG) data for a single session, there are no such studies conducted for multiple session EEG data. Thus, this study aims at classifying existing raw EEG meditation data on single and multiple sessions to come up with meaningful inferences which will be highly beneficial when developing algorithms that can support medita- tion practices. In this analysis, data have been collected on Pre-Resting (before-meditation), Post-Resting (after- meditation), LKM-Self and LKM-Others for 32 participants and hence allowing us to conduct six pairwise comparisons for the four mind tasks. Common Spatial Patterns (CSP) is a feature extraction method widely used in motor imaginary brain computer interface (BCI), but not in meditation EEG data. Therefore, using CSP in extracting features from medi- tation EEG data and classifying meditation/non-meditation instances, particularly for multiple sessions will create a new path in future meditation EEG research. The classification was done using Linear Discriminant Analysis (LDA) where both meditation techniques (LKM-Self and LKM-Others) were compared with Pre-Resting and Post-Resting instances. The results show that for a single session of 32 participants, around 99.5% accuracy was obtained for clas- sifying meditation/Pre-Resting instances. For the 15 participants when using five sessions of EEG data, around 83.6% accuracy was obtained for classifying meditation/Pre-Resting instances. The results demonstrate the ability to classify meditation/Pre-Resting data. Most importantly, this classification is possible for multiple session data as well. In addi- tion to this, when comparing the classification accuracies of the six mind task pairs; LKM-Self, LKM-Others and Post- Resting produced relatively lower accuracies among them than the accuracies obtained for classifying Pre-Resting with the other three. This indicates that Pre-Resting has some features giving a better classification indicating that it is different from the other three mind tasks.
  • Thumbnail Image
    Item
    Rhododendron taxonomy and diversity of ex situ collections for conservation : (subsection) Maddenia species as a case study : a thesis presented in the partial fulfilment of the requirements for the degree of Doctor of Philosophy (PhD) in Agriculture and Horticulture at Massey University (Manawatū campus], Palmerston North, New Zealand
    (Massey University, 2024-02-01) Hu, Ling
    In the ‘big genus’ Rhododendron of over 1,000 species, more than half of the species are threatened, at risk, or lacking data for biodiversity conservation. Ex situ collections, especially those from the wild, are crucial for safeguarding the diversity of species. However, lack of knowledge on existing wild diversity in botanic garden collections, and complex taxonomy, are two main problems in species assessments and conservation decision making. This research studied subsection Maddenia, a group of ~65 taxa encountering the two problems but seldom studied, as an exemplar to investigate species taxonomy and ex situ diversity. An ex situ conservation gap analysis was undertaken, using ecogeographical representation as a proxy for genetic representation in current botanic garden collections worldwide. Fifty-five of the total 65 taxa were found in cultivation, with over 86% of the living collections conserved in 66% of global botanic gardens. Half of the 18 threatened taxa, and nine of the 12 Data Deficient taxa require further wild collection to achieve a minimum level of ecogeographical representation in ex situ collections. Occurrence of ex situ collections in countries of origin is limited, and the distribution of ex situ collections worldwide is northern hemisphere centric. The results highlight the necessity of having more ex situ collections in the 10 native countries, and the importance of inter-institutional data sharing and robust documentation of collections. Determination of ploidy level of species was the second study, as the presence of polyploid samples may affect phylogenetic analysis. Ploidy levels were estimated for 263 accessions of 47 taxa (including 135 wild accessions) using flow cytometry. Meiotic chromosomes were counted for representative species of both diploids and polyploids to verify the flow cytometry results. This study showed that all taxa except one were diploid. The exception was that polyploids (2–8x, 12x) occur in the R. maddenii complex, where only seven of the 62 accessions tested were diploid while the rest were polyploid. This high level of polyploidy, combined with (i) the wide geographical distribution of the R. maddenii complex, and (ii) the previous ‘lumping’ of 12 taxa into the two subspecies, suggests the possibility of either some cryptic species or the need to re-evaluate some of the synonymized species. If new species were revealed, some may require conservation action. However, a greater number of wild-collected accessions and of different geographic origins are needed to explore this possibility. Following the ploidy study, molecular phylogeny of 40 taxa, including diploids and polyploids, was analysed using target capture sequencing. Phylogenetic trees from maximum likelihood and Bayesian analyses largely supported the morphological groupings of the Maddenii Series, Ciliicalyx Subseries and Megacalyx Subseries, but not the Ciliatum Subseries as classified by Davidian (1982). Of particular interest was the clustering in one clade of all of the R. maddenii complex, including all polyploid samples. This occurred irrespective of the method of analysis; however, there was no clear pattern of relationships to ploidy levels within the clade. The molecular phylogeny delimited several species and suggested a revision of the boundary of ‘subsection Maddenia’, although further research, to include a wider range of species, is needed to determine whether the new boundaries should be wider or narrower than before. The feasibility of using controlled pollination for safeguarding germplasm of prioritised species in ex situ collections was studied. Fruit set and seed germination identified the self- and cross-incompatibility of R. excellens (Vulnerable), which requires methods other than controlled pollination to conserve the intraspecific diversity in botanic gardens. R. dalhousiae var. dalhousiae (Least Concern), R. dalhousiae var. rhabdotum (Vulnerable), R. lindleyi (Least Concern), and R. nuttallii (Near Threatened) were both self- and cross-compatible, but the compatibility between self and cross pollinations differed from taxon to taxon and from accession to accession. These results suggest the choice of intraspecific pollination should be tested for each species before a programme of propagation is initiated. These aspects studied for subsection Maddenia can be immediately applied to conservation of this group of plants by working with the Global Conservation Consortium for Rhododendron. Meanwhile, the methods used here provide an exemplar for investigating other Rhododendron species or plant groups that encounter similar problems, to guide conservation efforts.