Search

Deep Learning (DL) methods have dramatically increased in popularity in recent years, with significant growth in their application to supervised learning problems in the biomedical sciences. However, the greater prevalence and complexity of missing data in modern biomedical datasets present significant challenges for DL methods. Here, we provide a formal treatment of missing data in the context of deeply learned generalized linear models, a supervised DL architecture for regression and classification problems. We propose a new architecture, \textit{dlglm}, that is one of the first to be able to flexibly account for both ignorable and non-ignorable patterns of missingness in input features and response at training time. We demonstrate through statistical simulation that our method outperforms existing approaches for supervised learning tasks in the presence of missing not at random (MNAR) missingness. We conclude with a case study of a Bank Marketing dataset from the UCI Machine Learning Repository, in which we predict whether clients subscribed to a product based on phone survey data. Supplementary materials for this article are available online.

Arbitrary conditioning is an important problem in unsupervised learning, where we seek to model the conditional densities $p(\mathbf{x}_u \mid \mathbf{x}_o)$ that underly some data, for all possible non-intersecting subsets $o, u \subset \{1, \dots , d\}$. However, the vast majority of density estimation only focuses on modeling the joint distribution $p(\mathbf{x})$, in which important conditional dependencies between features are opaque. We propose a simple and general framework, coined Posterior Matching, that enables Variational Autoencoders (VAEs) to perform arbitrary conditioning, without modification to the VAE itself. Posterior Matching applies to the numerous existing VAE-based approaches to joint density estimation, thereby circumventing the specialized models required by previous approaches to arbitrary conditioning. We find that Posterior Matching is comparable or superior to current state-of-the-art methods for a variety of tasks with an assortment of VAEs (e.g.~discrete, hierarchical, VaDE)., Comment: Accepted at NeurIPS 2022

Modern high-throughput single-cell immune profiling technologies, such as flow and mass cytometry and single-cell RNA sequencing can readily measure the expression of a large number of protein or gene features across the millions of cells in a multi-patient cohort. While bioinformatics approaches can be used to link immune cell heterogeneity to external variables of interest, such as, clinical outcome or experimental label, they often struggle to accommodate such a large number of profiled cells. To ease this computational burden, a limited number of cells are typically \emph{sketched} or subsampled from each patient. However, existing sketching approaches fail to adequately subsample rare cells from rare cell-populations, or fail to preserve the true frequencies of particular immune cell-types. Here, we propose a novel sketching approach based on Kernel Herding that selects a limited subsample of all cells while preserving the underlying frequencies of immune cell-types. We tested our approach on three flow and mass cytometry datasets and on one single-cell RNA sequencing dataset and demonstrate that the sketched cells (1) more accurately represent the overall cellular landscape and (2) facilitate increased performance in downstream analysis tasks, such as classifying patients according to their clinical outcome. An implementation of sketching with Kernel Herding is publicly available at \url{https://github.com/vishalathreya/Set-Summarization}., Comment: Accepted by ACM-BCB 2022

Modeling distributions of covariates, or density estimation, is a core challenge in unsupervised learning. However, the majority of work only considers the joint distribution, which has limited utility in practical situations. A more general and useful problem is arbitrary conditional density estimation, which aims to model any possible conditional distribution over a set of covariates, reflecting the more realistic setting of inference based on prior knowledge. We propose a novel method, Arbitrary Conditioning with Energy (ACE), that can simultaneously estimate the distribution $p(\mathbf{x}_u \mid \mathbf{x}_o)$ for all possible subsets of unobserved features $\mathbf{x}_u$ and observed features $\mathbf{x}_o$. ACE is designed to avoid unnecessary bias and complexity -- we specify densities with a highly expressive energy function and reduce the problem to only learning one-dimensional conditionals (from which more complex distributions can be recovered during inference). This results in an approach that is both simpler and higher-performing than prior methods. We show that ACE achieves state-of-the-art for arbitrary conditional likelihood estimation and data imputation on standard benchmarks., Accepted at NeurIPS 2021

Truly intelligent systems are expected to make critical decisions with incomplete and uncertain data. Active feature acquisition (AFA), where features are sequentially acquired to improve the prediction, is a step towards this goal. However, current AFA models all deal with a small set of candidate features and have difficulty scaling to a large feature space. Moreover, they are ignorant about the valid domains where they can predict confidently, thus they can be vulnerable to out-of-distribution (OOD) inputs. In order to remedy these deficiencies and bring AFA models closer to practical use, we propose several techniques to advance the current AFA approaches. Our framework can easily handle a large number of features using a hierarchical acquisition policy and is more robust to OOD inputs with the help of an OOD detector for partially observed data. Extensive experiments demonstrate the efficacy of our framework over strong baselines.

Modeling dependencies among features is fundamental for many machine learning tasks. Although there are often multiple related instances that may be leveraged to inform conditional dependencies, typical approaches only model conditional dependencies over individual instances. In this work, we propose a novel framework, partially observed exchangeable modeling (POEx) that takes in a set of related partially observed instances and infers the conditional distribution for the unobserved dimensions over multiple elements. Our approach jointly models the intra-instance (among features in a point) and inter-instance (among multiple points in a set) dependencies in data. POEx is a general framework that encompasses many existing tasks such as point cloud expansion and few-shot generation, as well as new tasks like few-shot imputation. Despite its generality, extensive empirical evaluations show that our model achieves state-of-the-art performance across a range of applications.

Deep Learning (DL) methods have dramatically increased in popularity in recent years. While its initial success was demonstrated in the classification and manipulation of image data, there has been significant growth in the application of DL methods to problems in the biomedical sciences. However, the greater prevalence and complexity of missing data in biomedical datasets present significant challenges for DL methods. Here, we provide a formal treatment of missing data in the context of Variational Autoencoders (VAEs), a popular unsupervised DL architecture commonly utilized for dimension reduction, imputation, and learning latent representations of complex data. We propose a new VAE architecture, NIMIWAE, that is one of the first to flexibly account for both ignorable and non-ignorable patterns of missingness in input features at training time. Following training, samples can be drawn from the approximate posterior distribution of the missing data can be used for multiple imputation, facilitating downstream analyses on high dimensional incomplete datasets. We demonstrate through statistical simulation that our method outperforms existing approaches for unsupervised learning tasks and imputation accuracy. We conclude with a case study of an EHR dataset pertaining to 12,000 ICU patients containing a large number of diagnostic measurements and clinical outcomes, where many features are only partially observed., Comment: 31 pages, 4 figures, 2 tables, under review (Biometrics Methodology)

Time series imputation is a fundamental task for understanding time series with missing data. Existing methods either do not directly handle irregularly-sampled data or degrade severely with sparsely observed data. In this work, we reformulate time series as permutation-equivariant sets and propose a novel imputation model NRTSI that does not impose any recurrent structures. Taking advantage of the permutation equivariant formulation, we design a principled and efficient hierarchical imputation procedure. In addition, NRTSI can directly handle irregularly-sampled time series, perform multiple-mode stochastic imputation, and handle data with partially observed dimensions. Empirically, we show that NRTSI achieves state-of-the-art performance across a wide range of time series imputation benchmarks., Comment: Codes available at https://github.com/lupalab/NRTSI

Many real-world situations allow for the acquisition of additional relevant information when making an assessment with limited or uncertain data. However, traditional ML approaches either require all features to be acquired beforehand or regard part of them as missing data that cannot be acquired. In this work, we consider models that perform active feature acquisition (AFA) and query the environment for unobserved features to improve the prediction assessments at evaluation time. Our work reformulates the Markov decision process (MDP) that underlies the AFA problem as a generative modeling task and optimizes a policy via a novel model-based approach. We propose learning a generative surrogate model (GSM) that captures the dependencies among input features to assess potential information gain from acquisitions. The GSM is leveraged to provide intermediate rewards and auxiliary information to aid the agent navigate a complicated high-dimensional action space and sparse rewards. Furthermore, we extend AFA in a task we coin active instance recognition (AIR) for the unsupervised case where the target variables are the unobserved features themselves and the goal is to collect information for a particular instance in a cost-efficient way. Empirical results demonstrate that our approach achieves considerably better performance than previous state of the art methods on both supervised and unsupervised tasks.

Clustering and prediction are two primary tasks in the fields of unsupervised and supervised learning, respectively. Although much of the recent advances in machine learning have been centered around those two tasks, the interdependent, mutually beneficial relationship between them is rarely explored. One could reasonably expect appropriately clustering the data would aid the downstream prediction task and, conversely, a better prediction performance for the downstream task could potentially inform a more appropriate clustering strategy. In this work, we focus on the latter part of this mutually beneficial relationship. To this end, we introduce Deep Goal-Oriented Clustering (DGC), a probabilistic framework that clusters the data by jointly using supervision via side-information and unsupervised modeling of the inherent data structure in an end-to-end fashion. We show the effectiveness of our model on a range of datasets by achieving prediction accuracies comparable to the state-of-the-art, while, more importantly in our setting, simultaneously learning congruent clustering strategies., 15 pages

In this work we develop a novel Bayesian neural network methodology to achieve strong adversarial robustness without the need for online adversarial training. Unlike previous efforts in this direction, we do not rely solely on the stochasticity of network weights by minimizing the divergence between the learned parameter distribution and a prior. Instead, we additionally require that the model maintain some expected uncertainty with respect to all input covariates. We demonstrate that by encouraging the network to distribute evenly across inputs, the network becomes less susceptible to localized, brittle features which imparts a natural robustness to targeted perturbations. We show empirical robustness on several benchmark datasets.

Many real-world situations allow for the acquisition of additional relevant information when making an assessment with limited or uncertain data. However, traditional ML approaches either require all features to be acquired beforehand or regard part of them as missing data that cannot be acquired. In this work, we propose models that dynamically acquire new features to further improve the prediction assessment. To trade off the improvement with the cost of acquisition, we leverage an information theoretic metric, conditional mutual information, to select the most informative feature to acquire. We leverage a generative model, arbitrary conditional flow (ACFlow), to learn the arbitrary conditional distributions required for estimating the information metric. We also learn a Bayesian network to accelerate the acquisition process. Our model demonstrates superior performance over baselines evaluated in multiple settings.

Reasoning over an instance composed of a set of vectors, like a point cloud, requires that one accounts for intra-set dependent features among elements. However, since such instances are unordered, the elements' features should remain unchanged when the input's order is permuted. This property, permutation equivariance, is a challenging constraint for most neural architectures. While recent work has proposed global pooling and attention-based solutions, these may be limited in the way that intradependencies are captured in practice. In this work we propose a more general formulation to achieve permutation equivariance through ordinary differential equations (ODE). Our proposed module, Exchangeable Neural ODE (ExNODE), can be seamlessly applied for both discriminative and generative tasks. We also extend set modeling in the temporal dimension and propose a VAE based model for temporal set modeling. Extensive experiments demonstrate the efficacy of our method over strong baselines.

Understanding the dependencies among features of a dataset is at the core of most unsupervised learning tasks. However, a majority of generative modeling approaches are focused solely on the joint distribution $p(x)$ and utilize models where it is intractable to obtain the conditional distribution of some arbitrary subset of features $x_u$ given the rest of the observed covariates $x_o$: $p(x_u \mid x_o)$. Traditional conditional approaches provide a model for a fixed set of covariates conditioned on another fixed set of observed covariates. Instead, in this work we develop a model that is capable of yielding all conditional distributions $p(x_u \mid x_o)$ (for arbitrary $x_u$) via tractable conditional likelihoods. We propose a novel extension of (change of variables based) flow generative models, arbitrary conditioning flow models (AC-Flow), that can be conditioned on arbitrary subsets of observed covariates, which was previously infeasible. We apply AC-Flow to the imputation of features, and also develop a unified platform for both multiple and single imputation by introducing an auxiliary objective that provides a principled single "best guess" for flow models. Extensive empirical evaluations show that our models achieve state-of-the-art performance in both single and multiple imputation across image inpainting and feature imputation in synthetic and real-world datasets. Code is available at https://github.com/lupalab/ACFlow.

The fundamental task of general density estimation $p(x)$ has been of keen interest to machine learning. In this work, we attempt to systematically characterize methods for density estimation. Broadly speaking, most of the existing methods can be categorized into either using: \textit{a}) autoregressive models to estimate the conditional factors of the chain rule, $p(x_{i}\, |\, x_{i-1}, \ldots)$; or \textit{b}) non-linear transformations of variables of a simple base distribution. Based on the study of the characteristics of these categories, we propose multiple novel methods for each category. For example we proposed RNN based transformations to model non-Markovian dependencies. Further, through a comprehensive study over both real world and synthetic data, we show for that jointly leveraging transformations of variables and autoregressive conditional models, results in a considerable improvement in performance. We illustrate the use of our models in outlier detection and image modeling. Finally we introduce a novel data driven framework for learning a family of distributions.

Sophisticated gated recurrent neural network architectures like LSTMs and GRUs have been shown to be highly effective in a myriad of applications. We develop an un-gated unit, the statistical recurrent unit (SRU), that is able to learn long term dependencies in data by only keeping moving averages of statistics. The SRU's architecture is simple, un-gated, and contains a comparable number of parameters to LSTMs; yet, SRUs perform favorably to more sophisticated LSTM and GRU alternatives, often outperforming one or both in various tasks. We show the efficacy of SRUs as compared to LSTMs and GRUs in an unbiased manner by optimizing respective architectures' hyperparameters in a Bayesian optimization scheme for both synthetic and real-world tasks.

The use of distributions and high-level features from deep architecture has become commonplace in modern computer vision. Both of these methodologies have separately achieved a great deal of success in many computer vision tasks. However, there has been little work attempting to leverage the power of these to methodologies jointly. To this end, this paper presents the Deep Mean Maps (DMMs) framework, a novel family of methods to non-parametrically represent distributions of features in convolutional neural network models. DMMs are able to both classify images using the distribution of top-level features, and to tune the top-level features for performing this task. We show how to implement DMMs using a special mean map layer composed of typical CNN operations, making both forward and backward propagation simple. We illustrate the efficacy of DMMs at analyzing distributional patterns in image data in a synthetic data experiment. We also show that we extending existing deep architectures with DMMs improves the performance of existing CNNs on several challenging real-world datasets.

Many interesting machine learning problems are best posed by considering instances that are distributions, or sample sets drawn from distributions. Previous work devoted to machine learning tasks with distributional inputs has done so through pairwise kernel evaluations between pdfs (or sample sets). While such an approach is fine for smaller datasets, the computation of an $N \times N$ Gram matrix is prohibitive in large datasets. Recent scalable estimators that work over pdfs have done so only with kernels that use Euclidean metrics, like the $L_2$ distance. However, there are a myriad of other useful metrics available, such as total variation, Hellinger distance, and the Jensen-Shannon divergence. This work develops the first random features for pdfs whose dot product approximates kernels using these non-Euclidean metrics, allowing estimators using such kernels to scale to large datasets by working in a primal space, without computing large Gram matrices. We provide an analysis of the approximation error in using our proposed random features and show empirically the quality of our approximation both in estimating a Gram matrix and in solving learning tasks in real-world and synthetic data.

We study the problem of distribution to real-value regression, where one aims to regress a mapping $f$ that takes in a distribution input covariate $P\in \mathcal{I}$ (for a non-parametric family of distributions $\mathcal{I}$) and outputs a real-valued response $Y=f(P) + \epsilon$. This setting was recently studied, and a "Kernel-Kernel" estimator was introduced and shown to have a polynomial rate of convergence. However, evaluating a new prediction with the Kernel-Kernel estimator scales as $\Omega(N)$. This causes the difficult situation where a large amount of data may be necessary for a low estimation risk, but the computation cost of estimation becomes infeasible when the data-set is too large. To this end, we propose the Double-Basis estimator, which looks to alleviate this big data problem in two ways: first, the Double-Basis estimator is shown to have a computation complexity that is independent of the number of of instances $N$ when evaluating new predictions after training; secondly, the Double-Basis estimator is shown to have a fast rate of convergence for a general class of mappings $f\in\mathcal{F}$.