Search

As the size of datasets used in statistical learning continues to grow, distributed training of models has attracted increasing attention. These methods partition the data and exploit parallelism to reduce memory and runtime, but suffer increasingly from communication costs as the data size or the number of iterations grows. Recent work on linear models has shown that a surrogate likelihood can be optimized locally to iteratively improve on an initial solution in a communication-efficient manner. However, existing versions of these methods experience multiple shortcomings as the data size becomes massive, including diverging updates and efficiently handling sparsity. In this work we develop solutions to these problems which enable us to learn a communication-efficient distributed logistic regression model even beyond millions of features. In our experiments we demonstrate a large improvement in accuracy over distributed algorithms with only a few distributed update steps needed, and similar or faster runtimes. Our code is available at \url{https://github.com/FutureComputing4AI/ProxCSL}., Comment: KDD 2024, Research Track

While distributed training is often viewed as a solution to optimizing linear models on increasingly large datasets, inter-machine communication costs of popular distributed approaches can dominate as data dimensionality increases. Recent work on non-interactive algorithms shows that approximate solutions for linear models can be obtained efficiently with only a single round of communication among machines. However, this approximation often degenerates as the number of machines increases. In this paper, building on the recent optimal weighted average method, we introduce a new technique, ACOWA, that allows an extra round of communication to achieve noticeably better approximation quality with minor runtime increases. Results show that for sparse distributed logistic regression, ACOWA obtains solutions that are more faithful to the empirical risk minimizer and attain substantially higher accuracy than other distributed algorithms., Comment: Under review

Industry practitioners care about small improvements in malware detection accuracy because their models are deployed to hundreds of millions of machines, meaning a 0.1\% change can cause an overwhelming number of false positives. However, academic research is often restrained to public datasets on the order of ten thousand samples and is too small to detect improvements that may be relevant to industry. Working within these constraints, we devise an approach to generate a benchmark of configurable difficulty from a pool of available samples. This is done by leveraging malware family information from tools like AVClass to construct training/test splits that have different generalization rates, as measured by a secondary model. Our experiments will demonstrate that using a less accurate secondary model with disparate features is effective at producing benchmarks for a more sophisticated target model that is under evaluation. We also ablate against alternative designs to show the need for our approach., Comment: To appear in Conference on Applied Machine Learning for Information Security 2023

To the best of our knowledge, there are no methods today for training differentially private regression models on sparse input data. To remedy this, we adapt the Frank-Wolfe algorithm for $L_1$ penalized linear regression to be aware of sparse inputs and to use them effectively. In doing so, we reduce the training time of the algorithm from $\mathcal{O}( T D S + T N S)$ to $\mathcal{O}(N S + T \sqrt{D} \log{D} + T S^2)$, where $T$ is the number of iterations and a sparsity rate $S$ of a dataset with $N$ rows and $D$ features. Our results demonstrate that this procedure can reduce runtime by a factor of up to $2,200\times$, depending on the value of the privacy parameter $\epsilon$ and the sparsity of the dataset., Comment: To appear in the 37th Conference on Neural Information Processing Systems (NeurIPS 2023)

Convolutional layers have long served as the primary workhorse for image classification. Recently, an alternative to convolution was proposed using the Sharpened Cosine Similarity (SCS), which in theory may serve as a better feature detector. While multiple sources report promising results, there has not been to date a full-scale empirical analysis of neural network performance using these new layers. In our work, we explore SCS's parameter behavior and potential as a drop-in replacement for convolutions in multiple CNN architectures benchmarked on CIFAR-10. We find that while SCS may not yield significant increases in accuracy, it may learn more interpretable representations. We also find that, in some circumstances, SCS may confer a slight increase in adversarial robustness., Comment: Accepted to I Can't Believe It's Not Better Workshop (ICBINB) at NeurIPS 2022

Differential privacy (DP) is the prevailing technique for protecting user data in machine learning models. However, deficits to this framework include a lack of clarity for selecting the privacy budget $\epsilon$ and a lack of quantification for the privacy leakage for a particular data row by a particular trained model. We make progress toward these limitations and a new perspective by which to visualize DP results by studying a privacy metric that quantifies the extent to which a model trained on a dataset using a DP mechanism is ``covered" by each of the distributions resulting from training on neighboring datasets. We connect this coverage metric to what has been established in the literature and use it to rank the privacy of individual samples from the training set in what we call a privacy profile. We additionally show that the privacy profile can be used to probe an observed transition to indistinguishability that takes place in the neighboring distributions as $\epsilon$ decreases, which we suggest is a tool that can enable the selection of $\epsilon$ by the ML practitioner wishing to make use of DP., Comment: Approved for Public Release; Distribution Unlimited. PA #:AFRL-2022-3639

LASSO regularized logistic regression is particularly useful for its built-in feature selection, allowing coefficients to be removed from deployment and producing sparse solutions. Differentially private versions of LASSO logistic regression have been developed, but generally produce dense solutions, reducing the intrinsic utility of the LASSO penalty. In this paper, we present a differentially private method for sparse logistic regression that maintains hard zeros. Our key insight is to first train a non-private LASSO logistic regression model to determine an appropriate privatized number of non-zero coefficients to use in final model selection. To demonstrate our method's performance, we run experiments on synthetic and real-world datasets., Comment: 20 pages, 5 figures

Linear $L_1$-regularized models have remained one of the simplest and most effective tools in data analysis, especially in information retrieval problems where n-grams over text with TF-IDF or Okapi feature values are a strong and easy baseline. Over the past decade, screening rules have risen in popularity as a way to reduce the runtime for producing the sparse regression weights of $L_1$ models. However, despite the increasing need of privacy-preserving models in information retrieval, to the best of our knoweledge, no differentially private screening rule exists. In this paper, we develop the first differentially private screening rule for linear and logistic regression. In doing so, we discover difficulties in the task of making a useful private screening rule due to the amount of noise added to ensure privacy. We provide theoretical arguments and experimental evidence that this difficulty arises from the screening step itself and not the private optimizer. Based on our results, we highlight that developing an effective private $L_1$ screening method is an open problem in the differential privacy literature., Comment: 5 pages, 2 figures

Subsampling algorithms are a natural approach to reduce data size before fitting models on massive datasets. In recent years, several works have proposed methods for subsampling rows from a data matrix while maintaining relevant information for classification. While these works are supported by theory and limited experiments, to date there has not been a comprehensive evaluation of these methods. In our work, we directly compare multiple methods for logistic regression drawn from the coreset and optimal subsampling literature and discover inconsistencies in their effectiveness. In many cases, methods do not outperform simple uniform subsampling., Comment: To appear in the Thirty-Seventh AAAI Conference on Artificial Intelligence

We present a framework to statistically audit the privacy guarantee conferred by a differentially private machine learner in practice. While previous works have taken steps toward evaluating privacy loss through poisoning attacks or membership inference, they have been tailored to specific models or have demonstrated low statistical power. Our work develops a general methodology to empirically evaluate the privacy of differentially private machine learning implementations, combining improved privacy search and verification methods with a toolkit of influence-based poisoning attacks. We demonstrate significantly improved auditing power over previous approaches on a variety of models including logistic regression, Naive Bayes, and random forest. Our method can be used to detect privacy violations due to implementation errors or misuse. When violations are not present, it can aid in understanding the amount of information that can be leaked from a given dataset, algorithm, and privacy specification., Comment: NeurIPS 2022

Many metric learning tasks, such as triplet learning, nearest neighbor retrieval, and visualization, are treated primarily as embedding tasks where the ultimate metric is some variant of the Euclidean distance (e.g., cosine or Mahalanobis), and the algorithm must learn to embed points into the pre-chosen space. The study of non-Euclidean geometries is often not explored, which we believe is due to a lack of tools for learning non-Euclidean measures of distance. Recent work has shown that Bregman divergences can be learned from data, opening a promising approach to learning asymmetric distances. We propose a new approach to learning arbitrary Bergman divergences in a differentiable manner via input convex neural networks and show that it overcomes significant limitations of previous works. We also demonstrate that our method more faithfully learns divergences over a set of both new and previously studied tasks, including asymmetric regression, ranking, and clustering. Our tests further extend to known asymmetric, but non-Bregman tasks, where our method still performs competitively despite misspecification, showing the general utility of our approach for asymmetric learning., Comment: Published in ICLR 2023, more related works added

We explore the utility of information contained within a dropout based Bayesian neural network (BNN) for the task of detecting out of distribution (OOD) data. We first show how previous attempts to leverage the randomized embeddings induced by the intermediate layers of a dropout BNN can fail due to the distance metric used. We introduce an alternative approach to measuring embedding uncertainty, justify its use theoretically, and demonstrate how incorporating embedding uncertainty improves OOD data identification across three tasks: image classification, language classification, and malware detection.

Ordinal regression is a classification task where classes have an order and prediction error increases the further the predicted class is from the true class. The standard approach for modeling ordinal data involves fitting parallel separating hyperplanes that optimize a certain loss function. This assumption offers sample efficient learning via inductive bias, but is often too restrictive in real-world datasets where features may have varying effects across different categories. Allowing class-specific hyperplane slopes creates generalized logistic ordinal regression, increasing the flexibility of the model at a cost to sample efficiency. We explore an extension of the generalized model to the all-thresholds logistic loss and propose a regularization approach that interpolates between these two extremes. Our method, which we term continuously generalized ordinal logistic, significantly outperforms the standard ordinal logistic model over a thorough set of ordinal regression benchmark datasets. We further extend this method to deep learning and show that it achieves competitive or lower prediction error compared to previous models over a range of datasets and modalities. Furthermore, two primary alternative models for deep learning ordinal regression are shown to be special cases of our framework.

Deep neural networks (DNN) have been used successfully in many scientific problems for their high prediction accuracy, but their application to genetic studies remains challenging due to their poor interpretability. In this paper, we consider the problem of scalable, robust variable selection in DNN for the identification of putative causal genetic variants in genome sequencing studies. We identified a pronounced randomness in feature selection in DNN due to its stochastic nature, which may hinder interpretability and give rise to misleading results. We propose an interpretable neural network model, stabilized using ensembling, with controlled variable selection for genetic studies. The merit of the proposed method includes: (1) flexible modelling of the non-linear effect of genetic variants to improve statistical power; (2) multiple knockoffs in the input layer to rigorously control false discovery rate; (3) hierarchical layers to substantially reduce the number of weight parameters and activations to improve computational efficiency; (4) de-randomized feature selection to stabilize identified signals. We evaluated the proposed method in extensive simulation studies and applied it to the analysis of Alzheimer disease genetics. We showed that the proposed method, when compared to conventional linear and nonlinear methods, can lead to substantially more discoveries.

Non-pharmaceutical interventions (NPIs) have been crucial in curbing COVID-19 in the United States (US). Consequently, relaxing NPIs through a phased re-opening of the US amid still-high levels of COVID-19 susceptibility could lead to new epidemic waves. This calls for a COVID-19 early warning system. Here we evaluate multiple digital data streams as early warning indicators of increasing or decreasing state-level US COVID-19 activity between January and June 2020. We estimate the timing of sharp changes in each data stream using a simple Bayesian model that calculates in near real-time the probability of exponential growth or decay. Analysis of COVID-19-related activity on social network microblogs, Internet searches, point-of-care medical software, and a metapopulation mechanistic model, as well as fever anomalies captured by smart thermometer networks, shows exponential growth roughly 2-3 weeks prior to comparable growth in confirmed COVID-19 cases and 3-4 weeks prior to comparable growth in COVID-19 deaths across the US over the last 6 months. We further observe exponential decay in confirmed cases and deaths 5-6 weeks after implementation of NPIs, as measured by anonymized and aggregated human mobility data from mobile phones. Finally, we propose a combined indicator for exponential growth in multiple data streams that may aid in developing an early warning system for future COVID-19 outbreaks. These efforts represent an initial exploratory framework, and both continued study of the predictive power of digital indicators as well as further development of the statistical approach are needed.

Learning disentangled representations is regarded as a fundamental task for improving the generalization, robustness, and interpretability of generative models. However, measuring disentanglement has been challenging and inconsistent, often dependent on an ad-hoc external model or specific to a certain dataset. To address this, we present a method for quantifying disentanglement that only uses the generative model, by measuring the topological similarity of conditional submanifolds in the learned representation. This method showcases both unsupervised and supervised variants. To illustrate the effectiveness and applicability of our method, we empirically evaluate several state-of-the-art models across multiple datasets. We find that our method ranks models similarly to existing methods. We make ourcode publicly available at https://github.com/stanfordmlgroup/disentanglement., Comment: Published at ICLR 2021

Ambitious global goals have been established to provide universal access to affordable modern contraceptive methods. The UN's sustainable development goal 3.7.1 proposes satisfying the demand for family planning (FP) services by increasing the proportion of women of reproductive age using modern methods. To measure progress toward such goals in populous countries like Nigeria, it's essential to characterize the current levels and trends of FP indicators such as unmet need and modern contraceptive prevalence rates (mCPR). Moreover, the substantial heterogeneity across Nigeria and scale of programmatic implementation requires a sub-national resolution of these FP indicators. However, significant challenges face estimating FP indicators sub-nationally in Nigeria. In this article, we develop a robust, data-driven model to utilize all available surveys to estimate the levels and trends of FP indicators in Nigerian states for all women and by age-parity demographic subgroups. We estimate that overall rates and trends of mCPR and unmet need have remained low in Nigeria: the average annual rate of change for mCPR by state is 0.5% (0.4%,0.6%) from 2012-2017. Unmet need by age-parity demographic groups varied significantly across Nigeria; parous women express much higher rates of unmet need than nulliparous women. Our hierarchical Bayesian model incorporates data from a diverse set of survey instruments, accounts for survey uncertainty, leverages spatio-temporal smoothing, and produces probabilistic estimates with uncertainty intervals. Our flexible modeling framework directly informs programmatic decision-making by identifying age-parity-state subgroups with large rates of unmet need, highlights conflicting trends across survey instruments, and holistically interprets direct survey estimates., Comment: 27 pages, 16 figures

Dengue is a mosquito-borne disease that threatens more than half of the world's population. Despite being endemic to over 100 countries, government-led efforts and mechanisms to timely identify and track the emergence of new infections are still lacking in many affected areas. Multiple methodologies that leverage the use of Internet-based data sources have been proposed as a way to complement dengue surveillance efforts. Among these, the trends in dengue-related Google searches have been shown to correlate with dengue activity. We extend a methodological framework, initially proposed and validated for flu surveillance, to produce near real-time estimates of dengue cases in five countries/regions: Mexico, Brazil, Thailand, Singapore and Taiwan. Our result shows that our modeling framework can be used to improve the tracking of dengue activity in multiple locations around the world.