Ian Covert
YOU?
Author Swipe
View article: Locality Alignment Improves Vision-Language Models
Locality Alignment Improves Vision-Language Models Open
Vision language models (VLMs) have seen growing adoption in recent years, but many still struggle with basic spatial reasoning errors. We hypothesize that this is due to VLMs adopting pre-trained vision backbones, specifically vision trans…
View article: LIFT - XAI: Leveraging Important Features in Treatment Effects to Inform Clinical Decision-Making via Explainable AI
LIFT - XAI: Leveraging Important Features in Treatment Effects to Inform Clinical Decision-Making via Explainable AI Open
Clinicians rely on evidence from randomized controlled trials (RCTs) to decide on medical treatments for patients. However, RCTs often lack the granularity needed to inform decisions for individual patients or specific clinical scenarios. …
View article: Scaling Laws for the Value of Individual Data Points in Machine Learning
Scaling Laws for the Value of Individual Data Points in Machine Learning Open
Recent works have shown that machine learning models improve at a predictable rate with the total amount of training data, leading to scaling laws that describe the relationship between error and dataset size. These scaling laws can help d…
View article: Stochastic Amortization: A Unified Approach to Accelerate Feature and Data Attribution
Stochastic Amortization: A Unified Approach to Accelerate Feature and Data Attribution Open
Many tasks in explainable machine learning, such as data valuation and feature attribution, perform expensive computation for each data point and are intractable for large datasets. These methods require efficient approximations, and altho…
View article: Feature Selection in the Contrastive Analysis Setting
Feature Selection in the Contrastive Analysis Setting Open
Contrastive analysis (CA) refers to the exploration of variations uniquely enriched in a target dataset as compared to a corresponding background dataset generated from sources of variation that are irrelevant to a given task. For example,…
View article: On the Robustness of Removal-Based Feature Attributions
On the Robustness of Removal-Based Feature Attributions Open
To explain predictions made by complex machine learning models, many feature attribution methods have been developed that assign importance scores to input features. Some recent work challenges the robustness of these methods by showing th…
View article: Estimating Conditional Mutual Information for Dynamic Feature Selection
Estimating Conditional Mutual Information for Dynamic Feature Selection Open
Dynamic feature selection, where we sequentially query features to make accurate predictions with a minimal budget, is a promising paradigm to reduce feature acquisition costs and provide transparency into a model's predictions. The proble…
View article: Learning to Maximize Mutual Information for Dynamic Feature Selection
Learning to Maximize Mutual Information for Dynamic Feature Selection Open
Feature selection helps reduce data acquisition costs in ML, but the standard approach is to train models with static feature subsets. Here, we consider the dynamic feature selection (DFS) problem where a model sequentially queries feature…
View article: Algorithms to estimate Shapley value feature attributions
Algorithms to estimate Shapley value feature attributions Open
Feature attributions based on the Shapley value are popular for explaining machine learning models; however, their estimation is complex from both a theoretical and computational standpoint. We disentangle this complexity into two factors:…
View article: Learning to Estimate Shapley Values with Vision Transformers
Learning to Estimate Shapley Values with Vision Transformers Open
Transformers have become a default architecture in computer vision, but understanding what drives their predictions remains a challenging problem. Current explanation approaches rely on attention values or input gradients, but these provid…
View article: Predictive and robust gene selection for spatial transcriptomics
Predictive and robust gene selection for spatial transcriptomics Open
A prominent trend in single-cell transcriptomics is providing spatial context alongside a characterization of each cell’s molecular state. This typically requires targeting an a priori selection of genes, often covering less than 1% of the…
View article: FastSHAP: Real-Time Shapley Value Estimation
FastSHAP: Real-Time Shapley Value Estimation Open
Shapley values are widely used to explain black-box models, but they are costly to calculate because they require many model evaluations. We introduce FastSHAP, a method for estimating Shapley values in a single forward pass using a learne…
View article: Disrupting Model Training with Adversarial Shortcuts
Disrupting Model Training with Adversarial Shortcuts Open
When data is publicly released for human consumption, it is unclear how to prevent its unauthorized usage for machine learning purposes. Successful model training may be preventable with carefully designed dataset modifications, and we pre…
View article: Neural Granger Causality
Neural Granger Causality Open
While most classical approaches to Granger causality detection assume linear dynamics, many interactions in real-world applications, like neuroscience and genomics, are inherently nonlinear. In these cases, using linear models may lead to …
View article: Improving KernelSHAP: Practical Shapley Value Estimation via Linear\n Regression
Improving KernelSHAP: Practical Shapley Value Estimation via Linear\n Regression Open
The Shapley value concept from cooperative game theory has become a popular\ntechnique for interpreting ML models, but efficiently estimating these values\nremains challenging, particularly in the model-agnostic setting. Here, we\nrevisit …
View article: Improving KernelSHAP: Practical Shapley Value Estimation via Linear Regression
Improving KernelSHAP: Practical Shapley Value Estimation via Linear Regression Open
The Shapley value concept from cooperative game theory has become a popular technique for interpreting ML models, but efficiently estimating these values remains challenging, particularly in the model-agnostic setting. Here, we revisit the…
View article: Explaining by Removing: A Unified Framework for Model Explanation
Explaining by Removing: A Unified Framework for Model Explanation Open
Researchers have proposed a wide variety of model explanation approaches, but it remains unclear how most methods are related or when one method is preferable to another. We describe a new unified class of methods, removal-based explanatio…
View article: Feature Removal Is a Unifying Principle for Model Explanation Methods
Feature Removal Is a Unifying Principle for Model Explanation Methods Open
Researchers have proposed a wide variety of model explanation approaches, but it remains unclear how most methods are related or when one method is preferable to another. We examine the literature and find that many methods are based on a …
View article: Understanding Global Feature Contributions With Additive Importance Measures
Understanding Global Feature Contributions With Additive Importance Measures Open
Understanding the inner workings of complex machine learning models is a long-standing problem and most recent research has focused on local interpretability. To assess the role of individual input features in a global sense, we explore th…
View article: Understanding Global Feature Contributions With Additive Importance Measures
Understanding Global Feature Contributions With Additive Importance Measures Open
Understanding the inner workings of complex machine learning models is a long-standing problem and most recent research has focused on local interpretability. To assess the role of individual input features in a global sense, we explore th…
View article: Temporal Graph Convolutional Networks for Automatic Seizure Detection
Temporal Graph Convolutional Networks for Automatic Seizure Detection Open
Seizure detection from EEGs is a challenging and time consuming clinical problem that would benefit from the development of automated algorithms. EEGs can be viewed as structural time series, because they are multivariate time series where…
View article: Neural Granger Causality for Nonlinear Time Series
Neural Granger Causality for Nonlinear Time Series Open
While most classical approaches to Granger causality detection assume linear dynamics, many interactions in applied domains, like neuroscience and genomics, are inherently nonlinear. In these cases, using linear models may lead to inconsis…