Library · Machine learning & AI
What K-Means Clustering got wrong, from 55 theses
Summary written by a model from the records below: The evaluated theses document various practical shortcomings and algorithmic limitations of k-means clustering across diverse scientific domains. Researchers frequently found that the algorithm struggles with non-spherical geometries, high-dimensional spaces, sensitivity to initialization, and complex temporal or hierarchical structures. These records come from graduate theses, 2021 to 2026, each labeled with its degree where the repository states it, and each links to its thesis. They were extracted by language models reading the full text, so treat each as a lead to read, not a verdict.
Plus 9 problems the authors left open, listed at the end.
Failure modes, by theses reporting them
Select a mode to jump to its records.
K-means underperforms hierarchical models, mixture models, and supervised alternatives
K-means achieved lower cluster separation, worse silhouette scores, and inferior accuracy when compared to hierarchical clustering, Gaussian mixture models, and supervised classifiers. It also exhibited slower execution times on large datasets and failed to improve upon hierarchical initializations during iterative optimization.
Tried and failed
k-means clustering initialized with hierarchical clustering applied to spatial tracking coordinate data. Outcome: worse than baseline. Reason: Lloyd iterations failed to improve upon the initial hierarchical assignments across all k values.
New methods in home-range overlap and clustering · Iowa State · PhD
Tried and failed
k-means clustering with local search heuristic applied to multi-period resource allocation optimization. Outcome: worse than baseline. Reason: produced inferior solution quality compared to structured branch-and-bound and column generation frameworks
Optimization of budgeted investment planning for multi-period, multi-facility, and multi-product systems · UT Austin · Doctorate
Tried and failed
cascade k-means clustering for unsupervised classification applied to remote sensing imagery classification. Outcome: worse than baseline. Reason: underperformed compared to x-means in identifying target spectral clusters accurately
Methods for quantifying changes to Lesser Prairie-Chicken lek connectivity and habitat · Texas Tech · Master's
Lost to a baseline
K-means, K-medoid, DBSCAN, and hierarchical clustering (single and Ward linkage) were beaten by level-set tree clustering (~95% cumulative accuracy on baseline gene OCR cluster recovery).
Categorizing the dispersion of open chromatin usage patterns of Treg related genes using an variational autoencoder based algorithm · Harvard · Master's
Lost to a baseline
k-means and k-medoids achieved lower silhouette scores (e.g. 0.0887 to 0.1703) on categorical taxonomy benchmarks compared to LSHFk-Centers and k-SCC (up to 0.3536)
Information systems and decision support for sustainable urban transport, energy systems, and emerging methods in research and practice · Leibniz Universität Hannover · PhD
Lost to a baseline
K-Means clustering on original features (KMC7) was beaten by PCA + GMM (GMM1/GMM2) on final pass smoothness and mean metrics
Study of retrofitted system for Intelligent Compaction Analyzer, a machine learning approach for Quality Control of Asphalt Pavement during Construction · University of Nevada, Reno · Master's
Lost to a baseline
K-means clustering took approximately 10x longer execution runtime than Agglomerative clustering (~1000 ns vs. ~100 ns) on large datasets.
Dynamic Decomposition and Deployment of the Virtual Network Functions with Microservices using Deep Reinforcement Learning · University College Dublin · PhD
Lost to a baseline
K-means and Gaussian Mixture Models (GMM) achieved lower cluster separation on SVD episode embeddings (silhouette scores of 0.53 and 0.25, respectively) compared to agglomerative hierarchical clustering (silhouette = 0.73).
Learning and Benchmarking Temporal Representations in Claims Data: Telehealth Trajectories in Buprenorphine Treatment for Opioid Use Disorder · Harvard · Master's
Considered and rejected
Considered and rejected: Rejected K-Means clustering (k=20) in favor of HDBSCAN because K-Means lacked ranked output of the top 10 most representative posts and exhibited lower clustering accuracy on large datasets.
Characterization of the promotion, adverse events, and regulation related to synthetic nicotine products on social media: a multiplatform content analysis using topic modeling · Boston University · Master's
Lost to a baseline
K-means unsupervised clustering underperformed supervised Random Forest, achieving only 38.2% to 82.4% accuracy compared to RF's 81.6% to 94.4%.
IDENTIFYING MIGRATION FATE AND FACTORS CONTRIBUTING TO MORTALITY OF ATLANTIC SALMON SMOLTS · Dalhousie · Master's
Lost to a baseline
On the TB-Xray image labeling task, baseline Gaussian Mixture Model clustering (76.70%) outperformed GOGGLES's generative inference method (76.89% vs baseline optimal cluster-class mapping where GMM scored 76.70% and K-Means scored 76.33% - wait, GOGGLES got 76.89%, but on GTSRB K-means achieved 70.74% outperforming GOGGLES's 70.51%).
Understanding, Fortifying and Democratizing AI Security · Georgia Tech · PhD
Considered and rejected
Considered and rejected: Rejected using standard k-means on raw handcrafted features alone because it lacks nonlinear feature optimization, yielding significantly lower reproducibility NMI than DEC.
In-vivo Human Superior Colliculus Functional Connectivity and Anatomy Segmentation · Tübingen · PhD
Hard Euclidean partitioning conflicts with continuous, temporal, and hierarchical data structures
Partitional k-means clustering failed when applied to data characterized by hierarchical lineages, continuous spatial axes, or temporal dynamics. Enforcing discrete Euclidean boundaries caused severe information loss and obscured subpopulation structures that required soft memberships or continuous modeling.
Considered and rejected
Considered and rejected: Rejected treating ordinal survey data as interval and applying k-means clustering in favor of LCCA, citing poor performance of naive clustering on categorical data.
Method versatility in analysing human attitudes towards technology · Leibniz Universität Hannover · PhD
Considered and rejected
Considered and rejected: Partitional k-means clustering was rejected because biological relatedness is hierarchically nested
The Impact of Medieval and Early Modern Migrations on Dental Nonmetric Variation in Hungary · University of Nevada, Reno · PhD
Considered and rejected
Considered and rejected: K-means clustering for comment vectors was rejected in favor of agglomerative hierarchical clustering after dendrogram analysis.
Modelling Human Behaviour Based on Similarity Measurements Between Event Sequences · Queen's University · Master's
Considered and rejected
Considered and rejected: Rejected k-means clustering using Euclidean distance across all data, because it treats textual similarities poorly compared to categorical simple matching.
Optimizing Data Compression via Data Reordering Strategies · York University · Master's
Considered and rejected
Considered and rejected: Relying on k-means or NMF on plainly smoothed data: rejected because they fail to produce neighbor-graph-compatible latent embeddings for UMAP or trajectory inference.
Fundamental representations of regions and interactions in spatial transcriptomics · MIT · PhD
Tried and failed
k-means clustering on 1D spatial projections applied to continuous longitudinal anatomical functional patterns. Reason: Failed to preserve spatial continuity along the longitudinal axis
Spinal Cord fMRI: Functional Connectivity, Network Modeling, and Brain-Spine Interactions · EPFL · PhD
Considered and rejected
Considered and rejected: Rejected standard/global 1D clustering (e.g. k-means across all features/subjects) because it overgeneralizes and obscures localized, subpopulation-specific neuroimaging patterns
Granular factoring into neuroimaging dynamic across space, time, and modality · Georgia Tech · PhD
Considered and rejected
Considered and rejected: Rejected generic K-means clustering for weekly sub-dataset evaluation because it is not tailored to time-series data.
Clustering and dimensionality reduction for time-series service monitoring data · University of Regina · Master's
Considered and rejected
Considered and rejected: Rejected hard clustering algorithms (e.g., standard k-means) for textual analysis because they force discrete discriminant grouping rather than accommodating multidimensional cluster memberships.
El comportamiento del donante de sangre en España desde la perspectiva del marketing social · University of Las Palmas de Gran Canaria · PhD
Considered and rejected
Considered and rejected: Rejected K-means clustering because it uses cluster averages instead of actual data points as medoids and cannot naturally cluster negatively correlated variables.
Analysis of High-dimensional Data with Variable Clustering and Selection · Georgia Tech · PhD
Assumptions of spherical, isotropic, and uniformly sized clusters fail on complex data geometries
The algorithm assumes isotropic, convex clusters with uniform variances and densities, leading to failures on non-linearly separable shapes and varied spatial densities. Practitioners observed that these geometric assumptions degraded performance on trajectory benchmarks, biological cells, and multispectral data.
Lost to a baseline
Kernel K-Means failed to cluster trajectories in the Duffing oscillator and hand-drawn datasets (misclassifying 27.81%-51.15% of trajectories) where Spectral Clustering and the proposed velocity-augmented Laplacian method succeeded.
Geometric Learning: Leveraging differential geometry for learning and control · EPFL · PhD
Considered and rejected
Considered and rejected: Decided against standard K-means and density-based clustering for scRNA-seq because they falsely assume equal cluster sizes and uniform within-population variance.
Integrating Genomic and Biological Understandings of Disease to Improve Patient Outcome in Mature B Cell Lymphoma · Duke · PhD
Considered and rejected
Considered and rejected: Rejected k-means clustering for spatial cell clustering because it requires a predefined cluster count and forces geometrically equal partitions regardless of biological cluster density/noise.
Analysing the relationship between immune infiltration and tissue architecture in high grade serous ovarian cancer · Cambridge · PhD
Considered and rejected
Considered and rejected: Avoided standard k-means 1D splits in hierarchical bisection when cluster sizes/densities differ significantly, preferring KDE-based thresholding.
Algorithms for Data Fusion, Representation Learning, and Scalable Clustering based on Constrained Low-Rank Approximation · Georgia Tech · PhD
Tried and failed
k-means clustering instead of Gaussian mixture models applied to multispectral image binarization. Outcome: worse than baseline. Reason: Spherical cluster assumption in k-means fails to capture non-spherical spectral distribution variances.
Restoration of multispectral images of ancient documents · TU Wien · PhD
Tried and failed
standard k-means or k-medoids clustering applied to representative distributional data reduction. Reason: it over-samples sparse regions and under-samples dense regions instead of preserving distribution density
Representative Data and Models for Complex Aerospace Systems Analysis · Georgia Tech · PhD
Considered and rejected
Considered and rejected: Rejected K-means and its variants for identifying lower-level subnetwork islands because they struggle with clusters of highly varying sizes, shapes, and densities.
Synthetic models of distribution gas networks in low-carbon energy systems · Politecnico di Torino · PhD
Considered and rejected
Considered and rejected: Rejected K-Means and DBSCAN clustering because K-Means assumes convex isotropic clusters and DBSCAN cannot handle large spatial density variations across CFD meshes
A machine learning approach to develop turbulence closures using clustering and neural networks · Imperial · PhD
Considered and rejected
Considered and rejected: Rejected K-Means and DBSCAN clustering due to sensitivity to outliers and poor performance on non-linearly separable OCR character shapes.
Automated identification of press variants in old documents · De Montfort University · PhD
Sensitivity to centroid initialization causes unstable clustering and convergence failures
K-means exhibited severe instability across successive runs due to its sensitivity to initial seed selection. This seed dependency caused convergence failures on unbalanced samples, produced arbitrary pseudo-labels, and risked degenerate single-cluster collapse.
Considered and rejected
Considered and rejected: Rejected simultaneous clustering and representation learning via naive k-means (DeepCluster style) due to susceptibility to degenerate single-cluster collapse; adopted PU-anchored k-means++ seeding instead
Robust and efficient learning in high dimensions from noisy data · UT Austin · PhD
Considered and rejected
Considered and rejected: Rejected K-means relocation clustering method in DAPC in favor of Ward's agglomerative hierarchical clustering method because K-means failed to converge across replicate runs for taxa with small/unbalanced sample sizes.
Considered and rejected
Considered and rejected: Rejected K-Means clustering because of sensitivity to initial seed points, opting for hierarchical clustering.
Glass on the Silk Roads: an SEM-EDS study of Islamic period artifacts from Rayy, Iran: their manufacture and trade connections · Nottingham · Master's
Considered and rejected
Considered and rejected: Rejected using global clustering (k-means) for pseudo-labeling in hidden unit discovery because of seed selection sensitivity and noisy/arbitrary clustering results.
Self-supervised learning for automatic speech recognition In low-resource environments · Nottingham · PhD
Tried and failed
K-means clustering without manual centroid presets applied to short text term clustering. Outcome: unstable. Reason: failed to generate clear, stable, and balanced topic clusters without manual centroid presets
Checking, Moderating, Adapting: Collective Engagement for Digital Integrity · Cornell · PhD
Considered and rejected
Considered and rejected: Rejected k-means clustering-based evaluation metrics (Normalized Mutual Information and F1 score) in metric learning due to seed sensitivity, failure to separate class cluster quality, and artificial inflation on datasets with few samples per class.
Correctly Evaluating Machine Learning Algorithms · Cornell · PhD
Unconstrained clustering creates severe cluster imbalance and empty or collapsed partitions
Unconstrained centroid optimization frequently concentrated the majority of data into one or two massive clusters while leaving other clusters sparse or empty. This behavior exacerbated class imbalances, isolated individual documents, and failed to represent minority categories.
Tried and failed
k-means and hierarchical clustering applied to pairwise co-occurrence and embedding distances. Reason: Produced highly unbalanced partitions dominated by one or two massive clusters
Deconstructing multiple long-term conditions: methods and applications of clustering diseases and people · Imperial · PhD
Considered and rejected
Considered and rejected: Rejected using k-means/k-modes cluster analysis as primary dimensionality reduction because clusters failed to capture variability in limitations subcodes and minority answer categories.
Variability in Measurement and Group Work Roles · Cornell · PhD
Considered and rejected
Considered and rejected: Rejected simple K-means clustering (with purely unsupervised centroids) for algospeak topic categorization due to inability to generate clear, stable, and balanced clusters.
Checking, Moderating, Adapting: Collective Engagement for Digital Integrity · Cornell · PhD
Considered and rejected
Considered and rejected: Rejected standard unconstrained K-means (SKM) because it creates severely imbalanced, non-contiguous spatial partitions.
Spatial Optimization Techniques for School Redistricting · Virginia Tech · Doctorate
Considered and rejected
Considered and rejected: Rejected k-means clustering with 3 to 5 clusters for semi-supervised labeling because it severely exacerbated class imbalance.
A COMPARISON OF ON-THE-EDGE MACHINE LEARNING CLASSIFICATION METHODS FOR FLIGHT PLANNING · Naval Postgraduate School · Master's
Considered and rejected
Considered and rejected: Rejected unconstrained k-means clustering for speech-to-manifesto authorship attribution, implementing minimum-size constrained k-means clustering instead to prevent the manifesto from forming an isolated single-document cluster.
Speech as data for the politics and partisanship of legislators · Oxford · PhD
High dimensionality and noisy latent representations degrade clustering quality
Applying k-means to high-dimensional embeddings or higher-order principal components degraded clustering performance due to the curse of dimensionality and added noise. In these settings, the algorithm produced arbitrary micro-clusters, struggled with density estimation, or suffered from centroid projection inaccuracies.
Tried and failed
k-means clustering on higher-order principal components applied to functional connectivity feature representations. Outcome: worse than baseline. Reason: higher-order principal components introduced noise that progressively degraded cluster separation quality
Sustained attention heterogeneity in post-9/11 Veterans: Functional connectivity profiles across large-scale cortical networks and subcortical regions during task-based fMRI · Harvard · Master's
Tried and failed
K-means clustering on latent vectors applied to degree-heterogeneous network nodes. Outcome: worse than baseline. Reason: Severe degree heterogeneity in the network biased clustering on weighted latent vectors.
Statistical viewpoints on network model, PDE Identification, low-rank matrix estimation and deep learning · Georgia Tech · PhD
Considered and rejected
Considered and rejected: Rejected K-means and Gaussian Mixture Models / kernel-density clustering because cluster counts are unknown and high dimensionality (64D) impairs density estimation.
Quantifying the Effects of Knee Joint Biomechanics on Acoustical Emissions · Georgia Tech · PhD
Considered and rejected
Considered and rejected: Rejected direct K-means clustering on 768-dimensional BERT embeddings for topic modeling due to the formation of micro-clusters and the curse of dimensionality, adopting TopClus projection instead.
Considered and rejected
Considered and rejected: Rejected PCA dimensionality reduction prior to k-means because it requires loading the full dataset and projecting centroids back to original space yields inaccurate centroids and lower cluster quality.
Approaches to extract, characterize, and interpret dynamic functional network connectivity in fMRI data · Georgia Tech · PhD
Lack of outlier rejection and fixed cluster counts hinder noisy and exploratory analyses
Because k-means lacks mechanisms to discard noise points, outliers frequently corrupted cluster borders and produced overly broad groupings. Additionally, requiring the number of clusters to be pre-specified hindered analyses in unknown environments where natural cluster counts were uncertain.
Tried and failed
k-means clustering for anomaly detection applied to subsurface drilling sensor time series. Reason: produced overly broad clusters with low silhouette scores rather than isolating localized anomalies
Integrating Machine Learning Techniques with Measurement-While-Drilling Data for Subsurface Characterization in Open-Pit Mines · Virginia Tech · Master's
Considered and rejected
Considered and rejected: Rejected principal component analysis (PCA) + k-means clustering because specifying principal components and clusters does not scale and spectral space is non-Gaussian
Expanding plausible biosignature gas candidates for detection in habitable exoplanet atmospheres · MIT · PhD
Considered and rejected
Considered and rejected: Rejected k-means clustering for track identification because it requires pre-specifying k for unknown scenes and fails to discard noisy data points.
MARITIME DOMAIN AWARENESS THROUGH THE CHARACTERIZATION OF SHIP BEHAVIOR WITH AIS DATA · Naval Postgraduate School · Master's
Considered and rejected
Considered and rejected: Rejected K-Means clustering and Density-based Clustering for decompositional clustering of seasonal signals due to high sensitivity to outliers (K-Means) and suboptimal cluster bordering/distinction (Density-based).
Characterizing non-E. coli coliforms as indicators of groundwater susceptibility via “big data”, geostatistical analysis, and machine learning · Queen's University · PhD
Considered and rejected
Considered and rejected: Rejected K-means and hierarchical clustering for final spatio-temporal hotspot analysis because they cannot remove noise points and require pre-specifying cluster counts.
IDENTIFYING PROBABLE MARITIME PIRACY EVENTS USING MARITIME INCIDENT DATA · Naval Postgraduate School · Master's
Surface illumination artifacts and color variations distort image pixel clustering
In computer vision applications, k-means was easily corrupted by optical shadows, surface reflections, and heavy color jitter. Operating directly on color channels without spatial coordinates caused distinct semantic regions to be misclassified due to superficial spectral similarities.
Tried and failed
unsupervised k-means clustering applied to high-resolution aerial imagery segmentation. Reason: optical shadows were misclassified as burned ground due to spectral similarity
Multiscale spatiotemporal approaches to contemporary pyrogeography in savanna systems · UT Austin · PhD
Tried and failed
superpixel segmentation followed by color kmeans clustering applied to unlabeled image segmentation. Reason: brightness variation dominated the average RGB color features, causing incorrect cluster assignments
Information Extraction from Messy Data, Noisy Spectra, Incomplete Data, and Unlabeled Images · Georgia Tech · PhD
Tried and failed
K-means clustering on color channels applied to image segmentation. Outcome: no signal. Reason: strong surface reflections and discarding spatial pixel coordinates corrupted cluster separation
Information Extraction from Messy Data, Noisy Spectra, Incomplete Data, and Unlabeled Images · Georgia Tech · PhD
Tried and failed
k-means clustering on color-augmented image embeddings applied to histopathology morphology subtyping. Outcome: no signal. Reason: heavy color jitter prevented separating subtle morphological patterns
Integrated Molecular and Histopathological Studies of Testicular Germ Cell Tumor Biology · Harvard · PhD
Left open by the authors
Problems the authors named and did not get to.
Left open
Compare the satellite clustering performance of k-means against hierarchical clustering algorithms like CHAMELEON or HDBSCAN on the longitudinal tracking dataset. Blocker: None
Measuring Compliance to the International Telecommunication Union’s Geosynchronous Orbital Assignments · MIT · PhD
Left open
Analyze why Ward hierarchical clustering outperforms Neighbor-Joining on specific metric-learned cell embedding spaces. Blocker: None
Reconstructing Cell Lineage Trees from Phenotypic Features with Metric Learning · Penn · PhD
Left open
Subcluster regulatory T cells from the post-HSCT scRNA-seq dataset using iterative clustering algorithms like ARBOL to identify rare cell states. Blocker: Requires access to the patient single-cell sequencing dataset generated in the thesis.
Single-cell immune analysis of chronic GVHD after hematopoietic stem cell transplantation · Harvard · Master's
Left open
Develop hierarchical metrics to quantify clusterness and trajectoriness on subsets and individual clusters in single-cell datasets. Blocker: None
Developing Graph-based Computational Algorithms for Single-cell Data Science · Georgia Tech · PhD
Left open
Perform k-means clustering with k > 14 on mxbai-embed-large-v1 sentence embeddings to separate quantum from generic uncertainty themes in student physics responses. Blocker: Access to the private Cornell student physics text dataset.
Evaluating language models applied to student thinking about experiments · Cornell · PhD
Left open
Implement consensus clustering methods, such as voting or mixture models, using LSSM sparse similarity matrices on single-cell sequencing data. Blocker: None
Identifying cell types with single cell sequencing data · Saskatchewan · Master's
Left open
Evaluate deep clustering algorithms on a dedicated, expert-annotated ground truth single-cell dataset with identical marker panels. Blocker: Requires a dedicated, expert-annotated ground-truth single-cell multiplexed image dataset with identical marker panels.
Single-cell Methods and Spatial Analysis for Highly Multiplexed Tissue Images · Harvard · PhD
Left open
Apply k-means clustering with PCA or confidence filtering to separate experimental XSW or dF/dz spectra of C60 and H2O@C60. Blocker: Requires experimental XSW or dF/dz spectral measurement data of C60 and H2O@C60 from STM apparatus
Machine learning at the nanoscale · Nottingham · PhD
Left open
Benchmark alternative clustering algorithms beyond k-means within the iterative filtering and relabeling framework for neuroimaging data. Blocker: Vague task direction with no specific algorithms or validation targets specified
Data driven approaches to address inaccurate nosology in mental health from neuroimaging data · Georgia Tech · PhD
Who has worked on this
Authors of the public theses cited on this page, most-cited first.
- Andy Zhao · Cornell · PhD · 2025
Checking, Moderating, Adapting: Collective Engagement for Digital Integrity - Hongzhen Tian · Georgia Tech · PhD · 2021
Information Extraction from Messy Data, Noisy Spectra, Incomplete Data, and Unlabeled Images - Charis Li · Harvard · Master's · 2026
Learning and Benchmarking Temporal Representations in Claims Data: Telehealth Trajectories in Buprenorphine Treatment for Opioid Use Disorder - FATMA ALSUBHI · Harvard · Master's · 2026
Sustained attention heterogeneity in post-9/11 Veterans: Functional connectivity profiles across large-scale cortical networks and subcortical regions during task-based fMRI - Rebeckah Fussell · Cornell · PhD · 2025
Evaluating language models applied to student thinking about experiments - Dongjin Choi · Georgia Tech · PhD · 2025
Algorithms for Data Fusion, Representation Learning, and Scalable Clustering based on Constrained Low-Rank Approximation - Kamal M. Maher · MIT · PhD · 2025
Fundamental representations of regions and interactions in spatial transcriptomics - Md Abdur Rahaman · Georgia Tech · PhD · 2025
Granular factoring into neuroimaging dynamic across space, time, and modality - Justin Dawsey · Texas Tech · Master's · 2025
Methods for quantifying changes to Lesser Prairie-Chicken lek connectivity and habitat - Maximilian Heumann · Leibniz Universität Hannover · PhD · 2025
Information systems and decision support for sustainable urban transport, energy systems, and emerging methods in research and practice - Ilaria Ricchi · EPFL · PhD · 2025
Spinal Cord fMRI: Functional Connectivity, Network Modeling, and Brain-Spine Interactions - Aran P. McMurtray · Naval Postgraduate School · Master's · 2025
A COMPARISON OF ON-THE-EDGE MACHINE LEARNING CLASSIFICATION METHODS FOR FLIGHT PLANNING
Checking a claim in this area?
We can run the same search on any method or claim. If nothing turns up, we will say so, and that proves nothing on its own.