May 2025
·
10 Reads
·
1 Citation
Computer Methods in Applied Mechanics and Engineering
This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.
May 2025
·
10 Reads
·
1 Citation
Computer Methods in Applied Mechanics and Engineering
April 2025
·
1 Read
Proceedings of the AAAI Conference on Artificial Intelligence
Dimension reduction (DR) algorithms have proven to be extremely useful for gaining insight into large-scale high-dimensional datasets, particularly finding clusters in transcriptomic data. The initial phase of these DR methods often involves converting the original high-dimensional data into a graph. In this graph, each edge represents the similarity or dissimilarity between pairs of data points. However, this graph is frequently suboptimal due to unreliable high-dimensional distances and the limited information extracted from the high-dimensional data. This problem is exacerbated as the dataset size increases. If we reduce the size of the dataset by selecting points for a specific sections of the embeddings, the clusters observed through DR are more separable since the extracted subgraphs are more reliable. In this paper, we introduce LocalMAP, a new dimensionality reduction algorithm that dynamically and locally adjusts the graph to address this challenge. By dynamically extracting subgraphs and updating the graph on-the-fly, LocalMAP is capable of identifying and separating real clusters within the data that other DR methods may overlook or combine. We demonstrate the benefits of LocalMAP through a case study on biological datasets, highlighting its utility in helping users more accurately identify clusters for real-world problems.
April 2025
·
6 Reads
·
1 Citation
Proceedings of the AAAI Conference on Artificial Intelligence
Health outcomes depend on complex environmental and sociodemographic factors whose effects change over location and time. Only recently has fine-grained spatial and temporal data become available to study these effects, namely the MEDSAT dataset of English health, environmental, and sociodemographic information. Leveraging this new resource, we use a variety of variable importance techniques to robustly identify the most informative predictors across multiple health outcomes. We then develop an interpretable machine learning framework based on Generalized Additive Models (GAMs) and Multiscale Geographically Weighted Regression (MGWR) to analyze both local and global spatial dependencies of each variable on various health outcomes. Our findings identify NO2 as a global predictor for asthma, hypertension, and anxiety, alongside other outcome-specific predictors related to occupation, marriage, and vegetation. Regional analyses reveal local variations with air pollution and solar radiation, with notable shifts during COVID. This comprehensive approach provides actionable insights for addressing health disparities, and advocates for the integration of interpretable machine learning in public health.
April 2025
·
34 Reads
International Journal of Mechanical Sciences
March 2025
·
7 Reads
Interpretability is critical for machine learning models in high-stakes settings because it allows users to verify the model's reasoning. In computer vision, prototypical part models (ProtoPNets) have become the dominant model type to meet this need. Users can easily identify flaws in ProtoPNets, but fixing problems in a ProtoPNet requires slow, difficult retraining that is not guaranteed to resolve the issue. This problem is called the "interaction bottleneck." We solve the interaction bottleneck for ProtoPNets by simultaneously finding many equally good ProtoPNets (i.e., a draw from a "Rashomon set"). We show that our framework - called Proto-RSet - quickly produces many accurate, diverse ProtoPNets, allowing users to correct problems in real time while maintaining performance guarantees with respect to the training set. We demonstrate the utility of this method in two settings: 1) removing synthetic bias introduced to a bird identification model and 2) debugging a skin cancer identification model. This tool empowers non-machine-learning experts, such as clinicians or domain experts, to quickly refine and correct machine learning models without repeated retraining by machine learning experts.
February 2025
·
3 Reads
Faithful explanations are essential for machine learning models in high-stakes applications. Inherently interpretable models are well-suited for these applications because they naturally provide faithful explanations by revealing their decision logic. However, model designers often need to keep these models proprietary to maintain their value. This creates a tension: we need models that are interpretable--allowing human decision-makers to understand and justify predictions, but not transparent, so that the model's decision boundary is not easily replicated by attackers. Shielding the model's decision boundary is particularly challenging alongside the requirement of completely faithful explanations, since such explanations reveal the true logic of the model for an entire subspace around each query point. This work provides an approach, FaithfulDefense, that creates model explanations for logical models that are completely faithful, yet reveal as little as possible about the decision boundary. FaithfulDefense is based on a maximum set cover formulation, and we provide multiple formulations for it, taking advantage of submodularity.
February 2025
·
17 Reads
Decision tree optimization is fundamental to interpretable machine learning. The most popular approach is to greedily search for the best feature at every decision point, which is fast but provably suboptimal. Recent approaches find the global optimum using branch and bound with dynamic programming, showing substantial improvements in accuracy and sparsity at great cost to scalability. An ideal solution would have the accuracy of an optimal method and the scalability of a greedy method. We introduce a family of algorithms called SPLIT (SParse Lookahead for Interpretable Trees) that moves us significantly forward in achieving this ideal balance. We demonstrate that not all sub-problems need to be solved to optimality to find high quality trees; greediness suffices near the leaves. Since each depth adds an exponential number of possible trees, this change makes our algorithms orders of magnitude faster than existing optimal methods, with negligible loss in performance. We extend this algorithm to allow scalable computation of sets of near-optimal trees (i.e., the Rashomon set).
February 2025
·
46 Reads
Despite antiretroviral therapy (ART), people with HIV (PWH) on ART experience higher rates of morbidity and mortality vs. age-matched HIV negative controls, which may be driven by chronic inflammation due to persistent virus. We performed bulk RNA sequencing (RNA-seq) on peripheral CD4+ T cells, as well as quantified plasma immune marker levels from 154 PWH on ART to identify host immune signatures associated with immune recovery (CD4:CD8) and HIV persistence (cell-associated HIV DNA and RNA). Using a novel dimension reduction tool - Pairwise Controlled Manifold Approximation (PaCMAP), we defined three distinct participant transcriptomic clusters. We found that these three clusters were largely defined by differential expression of genes regulated by the transcription factor NF-κB. While clustering was not associated with HIV reservoir size, we observed an association with CD4:CD8 ratio, a marker of immune recovery and prognostic factor for mortality in PWH on ART. Furthermore, distinct patterns of plasma IL-1β, TNF-α and GCSF were also strongly associated with the clusters, suggesting that these immune markers play a key role in CD4+ T cell transcriptomic diversity and immune recovery in PWH on ART. These findings reveal novel subgroups of PWH on ART with distinct immunological characteristics, and define a transcriptional signature associated with clinically significant immune parameters for PWH. A deeper understanding of these subgroups could advance clinical strategies to treat HIV-associated immune dysfunction.
January 2025
·
12 Reads
·
1 Citation
January 2025
·
13 Reads
·
1 Citation
Journal of the American Medical Informatics Association
Objective Prediction of mortality in intensive care unit (ICU) patients typically relies on black box models (that are unacceptable for use in hospitals) or hand-tuned interpretable models (that might lead to the loss in performance). We aim to bridge the gap between these 2 categories by building on modern interpretable machine learning (ML) techniques to design interpretable mortality risk scores that are as accurate as black boxes. Material and Methods We developed a new algorithm, GroupFasterRisk, which has several important benefits: it uses both hard and soft direct sparsity regularization, it incorporates group sparsity to allow more cohesive models, it allows for monotonicity constraint to include domain knowledge, and it produces many equally good models, which allows domain experts to choose among them. For evaluation, we leveraged the largest existing public ICU monitoring datasets (MIMIC III and eICU). Results Models produced by GroupFasterRisk outperformed OASIS and SAPS II scores and performed similarly to APACHE IV/IVa while using at most a third of the parameters. For patients with sepsis/septicemia, acute myocardial infarction, heart failure, and acute kidney failure, GroupFasterRisk models outperformed OASIS and SOFA. Finally, different mortality prediction ML approaches performed better based on variables selected by GroupFasterRisk as compared to OASIS variables. Discussion Group Faster Risk’s models performed better than risk scores currently used in hospitals, and on par with black box ML models, while being orders of magnitude sparser. Because GroupFasterRisk produces a variety of risk scores, it allows design flexibility—the key enabler of practical model creation. Conclusion Group Faster Risk is a fast, accessible, and flexible procedure that allows learning a diverse set of sparse risk scores for mortality prediction.
... Previous public health and urban studies have achieved varying degrees of success using traditional statistical methods [4,31,51], and machine learning models [33,82,116]. However, statistical approaches struggle with nonlinear effects, while machine learning models often fail to capture spatial relationships (e.g., how close neighborhoods are to each other, and geographic barriers like roads or rivers) and topological relationships (e.g., how connected neighborhoods are despite their physical distance), reducing predictive accuracy [56]. ...
April 2025
Proceedings of the AAAI Conference on Artificial Intelligence
... The Rashomon Effect is often present in datasets generated by processes that are nondeterministic, i.e., noisy or uncertain, including data used for bail and parole decisions, healthcare, and financial loan decisions -high stakes applications. In fact, as we will discuss, it has been proven in specific cases that datasets drawn from noisy processes tend to exhibit a large Rashomon Effect in that there are many approximately-equally accurate models (Semenova et al., 2023). Furthermore, a large Rashomon Effect correlates with the existence of simple-yet-accurate models (Semenova et al., 2022). ...
January 2025
Journal of the American Medical Informatics Association
... Based on their objections, we designed a set of examples that put into question the interpretability of prototypes. For example, Figure 1 shows an interpretation of the prototype-based network (specifically ProtoViT, Ma et al., 2024), where an image of a bird is classified based on features extracted from car images. Our proposed prototype manipulation highlights that visual confirmation bias (Klayman, 1995;Kim et al., 2022) is a threat potentially masking these models' inherent uninterpretability. ...
October 2024
... A wide array of extensions have followed the original ProtoPNet [7]. The majority of these works focus on improving components of the ProtoP-Net itself [10,30,32,33,37,38,47,48], improving the training regime [34,39,52], or applying ProtoPNets to high stakes tasks [1,2,8,51,54]. In principle, Proto-RSet can be combined with features from many of these extensions, particularly those that use a linear layer to map from prototype activations to class predictions. ...
June 2024
... Most contemporary methods are based on neural networks 3,[8][9][10][11][12][13] . These approaches outperform traditional methods by leveraging their large number of parameters to perform end-to-end artefact cleaning on time series data. ...
September 2024
Nature Machine Intelligence
... Our work could have substantial societal impact, for instance, if it is able to find clusters of patients that have different immune system properties (Semenova et al. 2024;Falcinelli et al. 2023). Our experiments indicate that LocalMAP has a higher chance of accomplishing this than past DR approaches. ...
September 2024
eLife
... Nonetheless, it is important to note that while comparing our method with others in the literature (Slapničar et al 2019, Aguet et al 2021, Ding et al 2024 could be insightful, significant variations in the employed datasets and evaluation setups make such comparisons challenging. These differences can lead to unfair assessments and misleading conclusions. ...
July 2024
Physiological Measurement
... This study aims to explore how the concave center and boundary visual principles in psychology, along with Gestalt principles, can be applied to optimize the design of new media marketing messages, thereby attracting more attention and enhancing recognition (Zhou et al., 2023). The concave center and boundary visual principle is one of the key psychological theories related to visual attention guidance. ...
March 2023
AIS Transactions on Human-Computer Interaction
... This shows there is a potential to nd more clinically optimal decision rules. Earlier work concluded that a large Rashomon set means an equally well-performing, interpretable model is likely to exist and correlates with the existence of models with good generalizability (21,47 (50)). ...
July 2024
... PPG signals are non-invasive, widely accessible, and easily collected through wearables and medical devices, making them suitable for foundational exploration [16]. Deep learning methods have already achieved notable success in single-task applications such as atrial fibrillation detection [17], heart rate estimation [18,19,20], and blood pressure prediction [21]. Techniques like convolutional recurrent regressors [18], cluster membership consistency methods [22], GAN-based augmentation for class imbalance [23], and signal quality-aware approaches like SQUWA [24] and [25] have further advanced the state-of-the-art in PPG signal analysis. ...
July 2024