arXiv subjects

Julia Fukuyama

Publications and source records attributed to Julia Fukuyama.

8 recordsLinked to original sources

arXiv2026

A comparative analysis of machine learning models in SHAP analysis

In this growing age of data and technology, large black-box models are becoming the norm due to their ability to handle vast amounts of data and learn incredibly complex data patterns. The deficiency of these methods, however, is their inability to explain the prediction process, making them untrustworthy and their use precarious in high-stakes situations. SHapley Additive exPlanations (SHAP) analysis is an explainable AI method growing in popularity for its ability to explain model predictions in terms of the original features. For each sample and feature in the data set, an associated SHAP value quantifies the contribution of that feature to the prediction of that sample. Analysis of these SHAP values provides valuable insight into the model's decision-making process, which can be leveraged to create data-driven solutions. The interpretation of these SHAP values, however, is model-dependent, so there does not exist a universal analysis procedure. To aid in these efforts, we present a detailed investigation of SHAP analysis across various machine learning models and data sets. In uncovering the details and nuance behind SHAP analysis, we hope to empower analysts in this less-explored territory. We also present a novel generalization of the waterfall plot to the multi-classification problem.

cs.LG↗

arXiv2026

DRtool: An Interactive Tool for Analyzing High-Dimensional Clusterings

When faced with new data, we often conduct a cluster analysis to obtain a better understanding of the data's structure and the archetypical samples present in the data. This process often includes visualization of the data, either as a way to discover or verify clusters. However, the increases in data complexity and dimensionality has made this step very tricky. To visualize data, nonlinear dimension reduction methods are the de facto standard for their ability to non-uniformly stretch and shrink space in order to preserve local clusters. Because this process requires a drastic manipulation of space, however, nonlinear dimension reduction methods are known to produce false structures, especially when mishandled. A common consequence that often goes undetected by the untrained eye is over-clustering of the data. In efforts to deal with this phenomenon, we developed an interactive tool that empowers analysts to distinguish false clusters and better interpret their high-dimensional clustering results. The tool uses various analytical plots to provide a multi-faceted perspective on the data's global structure as well as local inter-cluster relationships, helping users determine the legitimacy of their high-dimensional clustering results. The tool is available via an R package named DRtool.

stat.AP↗

arXiv2025

SHAP-Based Supervised Clustering for Sample Classification and the Generalized Waterfall Plot

In this growing age of data and technology, large black-box models are becoming the norm due to their ability to handle vast amounts of data and learn incredibly complex input-output relationships. The deficiency of these methods, however, is their inability to explain the prediction process, making them untrustworthy and their use precarious in high-stakes situations. SHapley Additive exPlanations (SHAP) analysis is an explainable AI method growing in popularity for its ability to explain model predictions in terms of the original features. For each sample and feature in the data set, we associate a SHAP value that quantifies the contribution of that feature to the prediction of that sample. Clustering these SHAP values can provide insight into the data by grouping samples that not only received the same prediction, but received the same prediction for similar reasons. In doing so, we map the various pathways through which distinct samples arrive at the same prediction. To showcase this methodology, we present a simulated experiment in addition to a case study in Alzheimer's disease using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. We also present a novel generalization of the waterfall plot for multi-classification.

cs.LG↗

arXiv2024

Calibrating dimension reduction hyperparameters in the presence of noise

The goal of dimension reduction tools is to construct a low-dimensional representation of high-dimensional data. These tools are employed for a variety of reasons such as noise reduction, visualization, and to lower computational costs. However, there is a fundamental issue that is discussed in other modeling problems that is often overlooked in dimension reduction -- overfitting. In the context of other modeling problems, techniques such as feature-selection, cross-validation, and regularization are employed to combat overfitting, but rarely are such precautions taken when applying dimension reduction. Prior applications of the two most popular non-linear dimension reduction methods, t-SNE and UMAP, fail to acknowledge data as a combination of signal and noise when assessing performance. These methods are typically calibrated to capture the entirety of the data, not just the signal. In this paper, we demonstrate the importance of acknowledging noise when calibrating hyperparameters and present a framework that enables users to do so. We use this framework to explore the role hyperparameter calibration plays in overfitting the data when applying t-SNE and UMAP. More specifically, we show previously recommended values for perplexity and n_neighbors are too small and overfit the noise. We also provide a workflow others may use to calibrate hyperparameters in the presence of noise.

stat.AP↗

arXiv2022

Multiscale Analysis of Count Data through Topic Alignment

Topic modeling is a popular method used to describe biological count data. With topic models, the user must specify the number of topics $K$. Since there is no definitive way to choose $K$ and since a true value might not exist, we develop techniques to study the relationships across models with different $K$. This can show how many topics are consistently present across different models, if a topic is only transiently present, or if a topic splits in two when $K$ increases. This strategy gives more insight into the process generating the data than choosing a single value of $K$ would. We design a visual representation of these cross-model relationships, which we call a topic alignment, and present three diagnostics based on it. We show the effectiveness of these tools for interpreting the topics on simulated and real data, and we release an accompanying R package, alto, available at https://lasy.github.io/alto.

stat.AP↗

arXiv2020

Local biplots for multi-dimensional scaling, with application to the microbiome

We present local biplots, a an extension of the classic principal components biplot to multi-dimensional scaling. Noticing that principal components biplots have an interpretation as the Jacobian of a map from data space to the principal subspace, we define local biplots as the Jacobian of the analogous map for multi-dimensional scaling. In the process, we show a close relationship between our local biplot axes, generalized Euclidean distances, and generalized principal components. In simulations and real data we show how local biplots can shed light on what variables or combinations of variables are important for the low-dimensional embedding provided by multi-dimensional scaling. They give particular insight into a class of phylogenetically-informed distances commonly used in the analysis of microbiome data, showing that different variants of these distances can be interpreted as implicitly smoothing the data along the phylogenetic tree and that the extent of this smoothing is variable.

stat.ME↗

arXiv2020

Lack of evidence for a substantial rate of templated mutagenesis in B cell diversification

B cell receptor sequences diversify through mutations introduced by purpose-built cellular machinery. A recent paper has concluded that a "templated mutagenesis" process is a major contributor to somatic hypermutation, and therefore immunoglobulin diversification, in mice and humans. In this proposed process, mutations in the immunoglobulin locus are introduced by copying short segments from other immunoglobulin genes. If true, this would overturn decades of research on B cell diversification, and would require a complete re-write of computational methods to analyze B cell data for these species. In this paper, we re-evaluate the templated mutagenesis hypothesis. By applying the original inferential method using potential donor templates absent from B cell genomes, we obtain estimates of the methods's false positive rates. We find false positive rates of templated mutagenesis in murine and human immunoglobulin loci that are similar to or even higher than the original rate inferences, and by considering the bases used in substitution we find evidence that if templated mutagenesis occurs, it is at a low rate. We also show that the statistically significant results in the original paper can easily result from a slight misspecification of the null model.

q-bio.PE↗

arXiv2017

Adaptive gPCA: A method for structured dimensionality reduction

When working with large biological data sets, exploratory analysis is an important first step for understanding the latent structure and for generating hypotheses to be tested in subsequent analyses. However, when the number of variables is large compared to the number of samples, standard methods such as principal components analysis give results which are unstable and difficult to interpret. To mitigate these problems, we have developed a method which allows the analyst to incorporate side information about the relationships between the variables in a way that encourages similar variables to have similar loadings on the principal axes. This leads to a low-dimensional representation of the samples which both describes the latent structure and which has axes which are interpretable in terms of groups of closely related variables. The method is derived by putting a prior encoding the relationships between the variables on the data and following through the analysis on the posterior distributions of the samples. We show that our method does well at reconstructing true latent structure in simulated data and we also demonstrate the method on a dataset investigating the effects of antibiotics on the composition of bacteria in the human gut.

stat.ME↗