Search

Society's capacity for algorithmic problem-solving has never been greater. Artificial Intelligence is now applied across more domains than ever, a consequence of powerful abstractions, abundant data, and accessible software. As capabilities have expanded, so have risks, with models often deployed without fully understanding their potential impacts. Interpretable and interactive machine learning aims to make complex models more transparent and controllable, enhancing user agency. This review synthesizes key principles from the growing literature in this field. We first introduce precise vocabulary for discussing interpretability, like the distinction between glass box and explainable models. We then explore connections to classical statistical and design principles, like parsimony and the gulfs of interaction. Basic explainability techniques -- including learned embeddings, integrated gradients, and concept bottlenecks -- are illustrated with a simple case study. We also review criteria for objectively evaluating interpretability approaches. Throughout, we underscore the importance of considering audience goals when designing interactive data-driven systems. Finally, we outline open challenges and discuss the potential role of data science in addressing them. Code to reproduce all examples can be found at https://go.wisc.edu/3k1ewe.

Microbiome interventions provide valuable data about microbial ecosystem structure and dynamics. Despite their ubiquity in microbiome research, few rigorous data analysis approaches are available. In this study, we extend transfer function-based intervention analysis to the microbiome setting, drawing from advances in statistical learning and selective inference. Our proposal supports the simulation of hypothetical intervention trajectories and False Discovery Rate-guaranteed selection of significantly perturbed taxa. We explore the properties of our approach through simulation and re-analyze three contrasting microbiome studies. An R package, mbtransfer, is available at https://go.wisc.edu/crj6k6. Notebooks to reproduce the simulation and case studies can be found at https://go.wisc.edu/dxuibh and https://go.wisc.edu/emxv33.

By linking conceptual theories with observed data, generative models can support reasoning in complex situations. They have come to play a central role both within and beyond statistics, providing the basis for power analysis in molecular biology, theory building in particle physics, and resource allocation in epidemiology, for example. We introduce the probabilistic and computational concepts underlying modern generative models and then analyze how they can be used to inform experimental design, iterative model refinement, goodness-of-fit evaluation, and agent-based simulation. We emphasize a modular view of generative mechanisms and discuss how they can be flexibly recombined in new problem contexts. We provide practical illustrations throughout, and code for reproducing all examples is available at https://github.com/krisrs1128/generative_review. Finally, we observe how research in generative models is currently split across several islands of activity, and we highlight opportunities lying at disciplinary intersections.

Spatially resolved transcriptomics (ST) measures gene expression along with the spatial coordinates of the measurements. The analysis of ST data involves significant computation complexity. In this work, we propose gene expression dimensionality reduction algorithm that retains spatial structure. We combine the wavelet transformation with matrix factorization to select spatially-varying genes. We extract a low-dimensional representation of these genes. We consider Empirical Bayes setting, imposing regularization through the prior distribution of factor genes. Additionally, We provide visualization of extracted representation genes capturing the global spatial pattern. We illustrate the performance of our methods by spatial structure recovery and gene expression reconstruction in simulation. In real data experiments, our method identifies spatial structure of gene factors and outperforms regular decomposition regarding reconstruction error. We found the connection between the fluctuation of gene patterns and wavelet technique, providing smoother visualization. We develop the package and share the workflow generating reproducible quantitative results and gene visualization. The package is available at https://github.com/OliverXUZY/waveST.

The use of remote sensing in humanitarian crisis response missions is well-established and has proven relevant repeatedly. One of the problems is obtaining gold annotations as it is costly and time consuming which makes it almost impossible to fine-tune models to new regions affected by the crisis. Where time is critical, resources are limited and environment is constantly changing, models has to evolve and provide flexible ways to adapt to a new situation. The question that we want to answer is if prioritization of samples provide better results in fine-tuning vs other classical sampling methods under annotated data scarcity? We propose a method to guide data collection during fine-tuning, based on estimated model and sample properties, like predicted IOU score. We propose two formulas for calculating sample priority. Our approach blends techniques from interpretability, representation learning and active learning. We have applied our method to a deep learning model for semantic segmentation, U-Net, in a remote sensing application of building detection - one of the core use cases of remote sensing in humanitarian applications. Preliminary results shows utility in prioritization of samples for tuning semantic segmentation models under scarcity of data condition.

Concept discovery is one of the open problems in the interpretability literature that is important for bridging the gap between non-deep learning experts and model end-users. Among current formulations, concepts defines them by as a direction in a learned representation space. This definition makes it possible to evaluate whether a particular concept significantly influences classification decisions for classes of interest. However, finding relevant concepts is tedious, as representation spaces are high-dimensional and hard to navigate. Current approaches include hand-crafting concept datasets and then converting them to latent space directions; alternatively, the process can be automated by clustering the latent space. In this study, we offer another two approaches to guide user discovery of meaningful concepts, one based on multiple hypothesis testing, and another on interactive visualization. We explore the potential value and limitations of these approaches through simulation experiments and an demo visual interface to real data. Overall, we find that these techniques offer a promising strategy for discovering relevant concepts in settings where users do not have predefined descriptions of them, but without completely automating the process., Comment: KDD'19, Workshop Explainable AI/ML (XAI) for Accountability, Fairness, and Transparency, August 04-08, 2019, Anchorage, AK, USA

Models that perform out-of-domain generalization borrow knowledge from heterogeneous source data and apply it to a related but distinct target task. Transfer learning has proven effective for accomplishing this generalization in many applications. However, poor selection of a source dataset can lead to poor performance on the target, a phenomenon called negative transfer. In order to take full advantage of available source data, this work studies source data selection with respect to a target task. We propose two source selection methods that are based on the multi-bandit theory and random search, respectively. We conduct a thorough empirical evaluation on both simulated and real data. Our proposals can be also viewed as diagnostics for the existence of a reweighted source subsamples that perform better than the random selection of available samples., Comment: 18 pages, 16 figures

Algorithmic feature learners provide high-dimensional vector representations for non-matrix structured signals, like images, audio, text, and graphs. Low-dimensional projections derived from these representations can be used to explore variation across collections of these data. However, it is not clear how to assess the uncertainty associated with these projections. We adapt methods developed for bootstrapping principal components analysis to the setting where features are learned from non-matrix data. We empirically compare the derived confidence regions in simulations, varying factors that influence both feature learning and the bootstrap. Approaches are illustrated on spatial proteomic data. Code, data, and trained models are released as an R compendium., Comment: arXiv admin note: text overlap with arXiv:2102.10388

Interpretability has attracted increasing attention in earth observation problems. We apply interactive visualization and representation analysis to guide interpretation of glacier segmentation models. We visualize the activations from a U-Net to understand and evaluate the model performance. We build an online interface using the Shiny R package to provide comprehensive error analysis of the predictions. Users can interact with the panels and discover model failure modes. Further, we discuss how visualization can provide sanity checks during data preprocessing and model training., Comment: 10 pages, 10 figures

Dimensionality reduction of spatial omic data can reveal shared, spatially structured patterns of expression across a collection of genomic features. We study strategies for discovering and interactively visualizing low-dimensional structure in spatial omic data based on the construction of neighborhood features. We design quantile and network-based spatial features that result in spatially consistent embeddings. A simulation compares embeddings made with and without neighborhood-based featurization, and a re-analysis of [Keren et al., 2019] illustrates the overall workflow. We provide an R package, NBFvis, to support computation and interactive visualization for the proposed dimensionality reduction approach. Code and data for reproducing experiments and analysis is available at https://github.com/XTH1114/NBFvis., Comment: 26 pages, 22 figures

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.

Cardiac Magnetic Resonance (CMR) is the most effective tool for the assessment and diagnosis of a heart condition, which malfunction is the world's leading cause of death. Software tools leveraging Artificial Intelligence already enhance radiologists and cardiologists in heart condition assessment but their lack of transparency is a problem. This project investigates if it is possible to discover concepts representative for different cardiac conditions from the deep network trained to segment crdiac structures: Left Ventricle (LV), Right Ventricle (RV) and Myocardium (MYO), using explainability methods that enhances classification system by providing the score-based values of qualitative concepts, along with the key performance metrics. With introduction of a need of explanations in GDPR explainability of AI systems is necessary. This study applies Discovering and Testing with Concept Activation Vectors (D-TCAV), an interpretaibilty method to extract underlying features important for cardiac disease diagnosis from MRI data. The method provides a quantitative notion of concept importance for disease classified. In previous studies, the base method is applied to the classification of cardiac disease and provides clinically meaningful explanations for the predictions of a black-box deep learning classifier. This study applies a method extending TCAV with a Discovering phase (D-TCAV) to cardiac MRI analysis. The advantage of the D-TCAV method over the base method is that it is user-independent. The contribution of this study is a novel application of the explainability method D-TCAV for cardiac MRI anlysis. D-TCAV provides a shorter pre-processing time for clinicians than the base method.

Many modern datasets don't fit neatly into $n \times p$ matrices, but most techniques for measuring statistical stability expect rectangular data. We study methods for stability assessment on non-rectangular data, using statistical learning algorithms to extract rectangular latent features. We design controlled simulations to characterize the power and practicality of competing approaches. This motivates new strategies for visualizing feature stability. Our stability curves supplement the direct analysis, providing information about the reliability of inferences based on learned features. Finally, we illustrate our approach using a spatial proteomics dataset, where machine learning tools can augment the scientist's workflow, but where guarantees of statistical reproducibility are still central. Our raw data, packaged code, and experimental outputs are publicly available., Comment: 39 pages, 25 figures

Glacier mapping is key to ecological monitoring in the hkh region. Climate change poses a risk to individuals whose livelihoods depend on the health of glacier ecosystems. In this work, we present a machine learning based approach to support ecological monitoring, with a focus on glaciers. Our approach is based on semi-automated mapping from satellite images. We utilize readily available remote sensing data to create a model to identify and outline both clean ice and debris-covered glaciers from satellite imagery. We also release data and develop a web tool that allows experts to visualize and correct model predictions, with the ultimate aim of accelerating the glacier mapping process., Comment: Accepted for a spotlight talk and a poster at the Tackling Climate Change with Machine Learning workshop at NeurIPS 2020

Generative deep learning has sparked a new wave of Super-Resolution (SR) algorithms that enhance single images with impressive aesthetic results, albeit with imaginary details. Multi-frame Super-Resolution (MFSR) offers a more grounded approach to the ill-posed problem, by conditioning on multiple low-resolution views. This is important for satellite monitoring of human impact on the planet -- from deforestation, to human rights violations -- that depend on reliable imagery. To this end, we present HighRes-net, the first deep learning approach to MFSR that learns its sub-tasks in an end-to-end fashion: (i) co-registration, (ii) fusion, (iii) up-sampling, and (iv) registration-at-the-loss. Co-registration of low-resolution views is learned implicitly through a reference-frame channel, with no explicit registration mechanism. We learn a global fusion operator that is applied recursively on an arbitrary number of low-resolution pairs. We introduce a registered loss, by learning to align the SR output to a ground-truth through ShiftNet. We show that by learning deep representations of multiple views, we can super-resolve low-resolution signals and enhance Earth Observation data at scale. Our approach recently topped the European Space Agency's MFSR competition on real-world satellite imagery., Comment: 15 pages, 5 figures

We introduce a conditional Generative Adversarial Network (cGAN) approach to generate cloud reflectance fields (CRFs) conditioned on large scale meteorological variables such as sea surface temperature and relative humidity. We show that our trained model can generate realistic CRFs from the corresponding meteorological observations, which represents a step towards a data-driven framework for stochastic cloud parameterization., Comment: Code is available on Github: https://github.com/krisrs1128/clouds_dist

Microscopy images are powerful tools and widely used in the majority of research areas, such as biology, chemistry, physics and materials fields by various microscopies (scanning electron microscope (SEM), atomic force microscope (AFM) and the optical microscope, et al.). However, most of the microscopy images are colorless due to the unique imaging mechanism. Though investigating on some popular solutions proposed recently about colorizing images, we notice the process of those methods are usually tedious, complicated, and time-consuming. In this paper, inspired by the achievement of machine learning algorithms on different science fields, we introduce two artificial neural networks for gray microscopy image colorization: An end-to-end convolutional neural network (CNN) with a pre-trained model for feature extraction and a pixel-to-pixel neural style transfer convolutional neural network (NST-CNN), which can colorize gray microscopy images with semantic information learned from a user-provided colorful image at inference time. The results demonstrate that our algorithm not only can colorize the microscopy images under complex circumstances precisely but also make the color naturally according to the training of a massive number of nature images with proper hue and saturation., Comment: 27 pages, 6 figures

In this work, we present the application of convolutional neural networks for segmenting water bodies in satellite images. We first use a variant of the U-Net model to segment rivers and lakes from very high-resolution images from Peru. To circumvent the issue of scarce labelled data, we investigate the applicability of a knowledge transfer-based model that learns the mapping from high-resolution labelled images and combines it with the very high-resolution mapping so that better segmentation can be achieved. We train this model in a single process, end-to-end. Our preliminary results show that adding the information from the available high-resolution images does not help out-of-the-box, and in fact worsen results. This leads us to infer that the high-resolution data could be from a different distribution, and its addition leads to increased variance in our results., Comment: 5 pages, 3 figures, 1 table, NeurIPS 2019 Workshop on Machine Learning for the Developing World

Climate change is one of the greatest challenges facing humanity, and we, as machine learning experts, may wonder how we can help. Here we describe how machine learning can be a powerful tool in reducing greenhouse gas emissions and helping society adapt to a changing climate. From smart grids to disaster management, we identify high impact problems where existing gaps can be filled by machine learning, in collaboration with other fields. Our recommendations encompass exciting research questions as well as promising business opportunities. We call on the machine learning community to join the global effort against climate change., Comment: For additional resources, please visit the website that accompanies this paper: https://www.climatechange.ai/

Importance weighted variational inference (Burda et al., 2015) uses multiple i.i.d. samples to have a tighter variational lower bound. We believe a joint proposal has the potential of reducing the number of redundant samples, and introduce a hierarchical structure to induce correlation. The hope is that the proposals would coordinate to make up for the error made by one another to reduce the variance of the importance estimator. Theoretically, we analyze the condition under which convergence of the estimator variance can be connected to convergence of the lower bound. Empirically, we confirm that maximization of the lower bound does implicitly minimize variance. Further analysis shows that this is a result of negative correlation induced by the proposed hierarchical meta sampling scheme, and performance of inference also improves when the number of samples increases., Comment: Accepted by ICML 2019. 17 pages

We present a project that aims to generate images that depict accurate, vivid, and personalized outcomes of climate change using Cycle-Consistent Adversarial Networks (CycleGANs). By training our CycleGAN model on street-view images of houses before and after extreme weather events (e.g. floods, forest fires, etc.), we learn a mapping that can then be applied to images of locations that have not yet experienced these events. This visual transformation is paired with climate model predictions to assess likelihood and type of climate-related events in the long term (50 years) in order to bring the future closer in the viewers mind. The eventual goal of our project is to enable individuals to make more informed choices about their climate future by creating a more visceral understanding of the effects of climate change, while maintaining scientific credibility by drawing on climate model projections.

Many studies have been performed to characterize the dynamics and stability of the microbiome across a range of environmental contexts [Costello et al., 2012, Faust et al., 2015]. For example, it is often of interest to identify time intervals within which certain subsets of taxa have an interesting pattern of behavior. Viewed abstractly, these problems often have a flavor not just of time series modeling but also of regime detection, a problem with a rich history across a variety of applications, including speech recognition [Fox et al., 2011], finance [Lee, 2009], EEG analysis [Camilleri et al., 2014], and geophysics [Weatherley and Mora, 2002]. In spite of the parallels, regime detection methods are rarely used in microbiome analysis, most likely due to the fact that references for these methods are scattered across several literatures, descriptions are inaccessible outside limited research communities, and implementations are difficult to come across. We distill the core ideas of different regime detection methods, provide example applications, and share reproducible code, making these techniques more accessible to microbiome researchers. We re-analyze data of Dethlefsen and Relman [2011], a study of the effects of antibiotics on the microbiome, using Classification and Regression Trees (CART) [Breiman et al., 1984], Hidden Markov Models (HMMs) [Rabiner and Juang, 1986], Bayesian nonparametric HMMs [Teh and Jordan, 2010, Fox et al., 2008], mixtures of Gaussian Processes (GPs) [Rasmussen and Ghahramani, 2002], switching dynamical systems [Linderman et al., 2016], and multiple changepoint detection [Fan and Mackey, 2015]. Along the way, we summarize each method, their relevance to the microbiome, and tradeoffs associated with using them. Ultimately, our goal is to describe types of temporal or regime switching structure that can be incorporated into studies of microbiome dynamics.

The Government of Mexico's social development agency, SEDESOL, is responsible for the administration of social services and has the mission of lifting Mexican families out of poverty. One key challenge they face is matching people who have social service needs with the services SEDESOL can provide accurately and efficiently. In this work we describe two specific applications implemented in collaboration with SEDESOL to enhance their distribution of social services. The first problem relates to systematic underreporting on applications for social services, which makes it difficult to identify where to prioritize outreach. Responding that five people reside in a home when only three do is a type of underreporting that could occur while a social worker conducts a home survey with a family to determine their eligibility for services. The second involves approximating multidimensional poverty profiles across households. That is, can we characterize different types of vulnerabilities -- for example, food insecurity and lack of health services -- faced by those in poverty? We detail the problem context, available data, our machine learning formulation, experimental results, and effective feature sets. As far as we are aware this is the first time government data of this scale has been used to combat poverty within Mexico. We found that survey data alone can suggest potential underreporting. Further, we found geographic features useful for housing and service related indicators and transactional data informative for other dimensions of poverty. The results from our machine learning system for estimating poverty profiles will directly help better match 7.4 million individuals to social programs., Comment: This work was done as part of the 2016 Eric & Wendy Schmidt Data Science for Social Good Summer Fellowship at the University of Chicago

The human microbiome is a complex ecological system, and describing its structure and function under different environmental conditions is important from both basic scientific and medical perspectives. Viewed through a biostatistical lens, many microbiome analysis goals can be formulated as latent variable modeling problems. However, although probabilistic latent variable models are a cornerstone of modern unsupervised learning, they are rarely applied in the context of microbiome data analysis, in spite of the evolutionary, temporal, and count structure that could be directly incorporated through such models. We explore the application of probabilistic latent variable models to microbiome data, with a focus on Latent Dirichlet Allocation, Nonnegative Matrix Factorization, and Dynamic Unigram models. To develop guidelines for when different methods are appropriate, we perform a simulation study. We further illustrate and compare these techniques using the data of [10], a study on the effects of antibiotics on bacterial community composition. Code and data for all simulations and case studies are available publicly., Comment: 31 pages, 16 figures

Opiates are some of the most effective pain relief medications available for patients suffering from cancer and surgery-related pain. Despite the affordability and effectiveness of these medications, access to opiates is highly geographically variable. Pain researchers have attributed geographic variation to various factors including the fear of opioid addiction, diversion of legal opiods to the underground market and pharmaceutical industry influences. However, the extent to which there is inequity in untreated cancer and surgery-related pain is unknown. To help opioid investigators study these questions, we designed a tool, the Opioid Atlas, for exploring data on legal opioid consumption, by country and time, collected by the International Narcotics Control Board. Our design borrows ideas from the data visualization and multivariate statistics communities, especially the principles of linking and dimensionality reduction. Our work is relevant to policymakers and pain researchers who wish to systematically assess country-level factors that contribute to differences in opioid access for patients with cancer and surgery-related pain. The Opioid Atlas, and the code behind it, is freely available with an open source license.