Faced with distribution shift between training and test set, we wish to detect and quantify the shift, and to correct our classifiers without test set labels. Motivated by medical diagnosis, where diseases (targets), cause symptoms (observations), we focus on label shift, where the label marginal $p(y)$ changes but the conditional $p(x|y)$ does not. We propose Black Box Shift Estimation (BBSE) to estimate the test distribution $p(y)$. BBSE exploits arbitrary black box predictors to reduce dimensionality prior to shift correction. While better predictors give tighter estimates, BBSE works even when predictors are biased, inaccurate, or uncalibrated, so long as their confusion matrices are invertible. We prove BBSE's consistency, bound its error, and introduce a statistical test that uses BBSE to detect shift. We also leverage BBSE to correct classifiers. Experiments demonstrate accurate estimates and improved prediction, even on high-dimensional datasets of natural images.

Vector quantization aims to form new vectors/matrices with shared values close to the original. It could compress data with acceptable information loss, and could be of great usefulness in areas like Image Processing, Pattern Recognition and Machine Learning. In recent years, the importance of quantization has been soaring as it has been discovered huge potentials in deploying practical neural networks, which is among one of the most popular research topics. Conventional vector quantization methods usually suffer from their own flaws: hand-coding domain rules quantization could produce poor results when encountering complex data, and clustering-based algorithms have the problem of inexact solution and high time consumption. In this paper, we explored vector quantization problem from a new perspective of sparse least square optimization and designed multiple algorithms with their program implementations. Specifically, deriving from a sparse form of coefficient matrix, three types of sparse least squares, with $l_0$, $l_1$, and generalized $l_1 + l_2$ penalizations, are designed and implemented respectively. In addition, to produce quantization results with given amount of quantized values(instead of penalization coefficient $\lambda$), this paper proposed a cluster-based least square quantization method, which could also be regarded as an improvement of information preservation of conventional clustering algorithm. The algorithms were tested on various data and tasks and their computational properties were analyzed. The paper offers a new perspective to probe the area of vector quantization, while the algorithms proposed could provide more appropriate options for quantization tasks under different circumstances.

Semidefinite programs (SDP) are important in learning and combinatorial optimization with numerous applications. In pursuit of low-rank solutions and low complexity algorithms, we consider the Burer--Monteiro factorization approach for solving SDPs. We show that all approximate local optima are global optima for the penalty formulation of appropriately rank-constrained SDPs as long as the number of constraints scales sub-quadratically with the desired rank of the optimal solution. Our result is based on a simple penalty function formulation of the rank-constrained SDP along with a smoothed analysis to avoid worst-case cost matrices. We particularize our results to two applications, namely, Max-Cut and matrix completion.

Human inertial thinking schemes can be formed through learning, which are then applied to quickly solve similar problems later. However, when problems are significantly different, inertial thinking generally presents the solutions that are definitely imperfect. In such cases, people will apply creative thinking, such as reverse thinking, to solve problems. Similarly, machine learning methods also form inertial thinking schemes through learning the knowledge from a large amount of data. However, when the testing data are vastly difference, the formed inertial thinking schemes will inevitably generate errors. This kind of inertial thinking is called illusion inertial thinking. Because all machine learning methods do not consider illusion inertial thinking, in this paper we propose a new method that uses reverse thinking to correct illusion inertial thinking, which increases the generalization ability of machine learning methods. Experimental results on benchmark datasets are used to validate the proposed method.

In this paper, we propose deep learning techniques for econometrics, specifically for causal inference and for estimating individual as well as average treatment effects. The contribution of this paper is twofold: 1. For generalized neighbor matching to estimate individual and average treatment effects, we analyze the use of autoencoders for dimensionality reduction while maintaining the local neighborhood structure among the data points in the embedding space. This deep learning based technique is shown to perform better than simple k nearest neighbor matching for estimating treatment effects, especially when the data points have several features/covariates but reside in a low dimensional manifold in high dimensional space. We also observe better performance than manifold learning methods for neighbor matching. 2. Propensity score matching is one specific and popular way to perform matching in order to estimate average and individual treatment effects. We propose the use of deep neural networks (DNNs) for propensity score matching, and present a network called PropensityNet for this. This is a generalization of the logistic regression technique traditionally used to estimate propensity scores and we show empirically that DNNs perform better than logistic regression at propensity score matching. Code for both methods will be made available shortly on Github at: https://github.com/vikas84bf

In this paper, we propose a listwise approach for constructing user-specific rankings in recommendation systems in a collaborative fashion. We contrast the listwise approach to previous pointwise and pairwise approaches, which are based on treating either each rating or each pairwise comparison as an independent instance respectively. By extending the work of (Cao et al., 2007), we cast listwise collaborative ranking as maximum likelihood under a permutation model which applies probability mass to permutations based on a low rank latent score matrix. We present a novel algorithm called SQL-Rank, which can accommodate ties and missing data and can run in linear time. We develop a theoretical framework for analyzing listwise ranking methods based on a novel representation theory for the permutation model. Applying this framework to collaborative ranking, we derive asymptotic statistical rates as the number of users and items grow together. We conclude by demonstrating that our SQL-Rank method often outperforms current state-of-the-art algorithms for implicit feedback such as Weighted-MF and BPR and achieve favorable results when compared to explicit feedback algorithms such as matrix factorization and collaborative ranking.

We present a new, fully generative model for constructing astronomical catalogs from optical telescope image sets. Each pixel intensity is treated as a Poisson random variable with a rate parameter that depends on the latent properties of stars and galaxies. These latent properties are themselves random, with scientific prior distributions constructed from large ancillary datasets. We compare two procedures for posterior inference: Markov chain Monte Carlo (MCMC) and variational inference (VI). MCMC excels at quantifying uncertainty while VI is 1000x faster. Both procedures outperform the current state-of-the-art method for measuring celestial bodies' colors, shapes, and morphologies. On a supercomputer, the VI procedure efficiently uses 665,000 CPU cores (1.3 million hardware threads) to construct an astronomical catalog from 50 terabytes of images.

In most machine learning applications, classification accuracy is not the primary metric of interest. Binary classifiers which face class imbalance are often evaluated by the $F_\beta$ score, area under the precision-recall curve, Precision at K, and more. The maximization of many of these metrics can be expressed as a constrained optimization problem, where the constraint is a function of the classifier's predictions. In this paper we propose a novel framework for learning with constraints that can be expressed as a predicted positive rate (or negative rate) on a subset of the training data. We explicitly model the threshold at which a classifier must operate to satisfy the constraint, yielding a surrogate loss function which avoids the complexity of constrained optimization. The method is model-agnostic and only marginally more expensive than minimization of the unconstrained loss. Experiments on a variety of benchmarks show competitive performance relative to existing baselines.

We propose a novel method called robust kernel principal component analysis (RKPCA) to decompose a partially corrupted matrix as a sparse matrix plus a high or full-rank matrix whose columns are drawn from a nonlinear low-dimensional latent variable model. RKPCA can be applied to many problems such as noise removal and subspace clustering and is so far the only unsupervised nonlinear method robust to sparse noises. We also provide theoretical guarantees for RKPCA. The optimization of RKPCA is challenging because it involves nonconvex and indifferentiable problems simultaneously. We propose two nonconvex optimization algorithms for RKPCA: alternating direction method of multipliers with backtracking line search and proximal linearized minimization with adaptive step size. Comparative studies on synthetic data and nature images corroborate the effectiveness and superiority of RKPCA in noise removal and robust subspace clustering.

Stability has been a recurrent issue in training generative adversarial networks (GANs). One common way to tackle this issue has been to propose new formulations of the GAN objective. Yet, surprisingly few studies have looked at optimization methods specifically designed for this adversarial training. In this work, we review the "variational inequality" framework which contains most formulations of the GAN objective introduced so far. Taping into the mathematical programming literature, we counter some common misconceptions about the difficulties of saddle point optimization and propose to extend standard methods designed for variational inequalities to GANs training, such as a stochastic version of the extragradient method, and empirically investigate their behavior on GANs.