We extend the existing framework of semi-implicit variational inference (SIVI) and introduce doubly semi-implicit variational inference (DSIVI), a way to perform variational inference and learning when both the approximate posterior and the prior distribution are semi-implicit. In other words, DSIVI performs inference in models where the prior and the posterior can be expressed as an intractable infinite mixture of some analytic density with a highly flexible implicit mixing distribution. We provide a sandwich bound on the evidence lower bound (ELBO) objective that can be made arbitrarily tight. Unlike discriminator-based and kernel-based approaches to implicit variational inference, DSIVI optimizes a proper lower bound on ELBO that is asymptotically exact. We evaluate DSIVI on a set of problems that benefit from implicit priors. In particular, we show that DSIVI gives rise to a simple modification of VampPrior, the current state-of-the-art prior for variational autoencoders, which improves its performance.

Bayesian Optimisation (BO), refers to a suite of techniques for global optimisation of expensive black box functions, which use introspective Bayesian models of the function to efficiently find the optimum. While BO has been applied successfully in many applications, modern optimisation tasks usher in new challenges where conventional methods fail spectacularly. In this work, we present Dragonfly, an open source Python library for scalable and robust BO. Dragonfly incorporates multiple recently developed methods that allow BO to be applied in challenging real world settings; these include better methods for handling higher dimensional domains, methods for handling multi-fidelity evaluations when cheap approximations of an expensive function are available, methods for optimising over structured combinatorial spaces, such as the space of neural network architectures, and methods for handling parallel evaluations. Additionally, we develop new methodological improvements in BO for selecting the Bayesian model, selecting the acquisition function, and optimising over complex domains with different variable types and additional constraints. We compare Dragonfly to a suite of other packages and algorithms for global optimisation and demonstrate that when the above methods are integrated, they enable significant improvements in the performance of BO. The Dragonfly library is available at dragonfly.github.io.

Modeling item complementariness and user preferences from purchase data is essential for learning good representations of products and customers, which empowers the modern personalized recommender system for Walmart's e-commerce platform. The intrinsic complementary relationship among products captures the buy-also-buy patterns and provides great sources for recommendations. Product complementary patterns, though often reflected by population purchase behaviors, are not separable from customer-specific bias in purchase data. We propose a unified model with Bayesian network structure that takes account of both factors. In the meantime, we merge the contextual knowledge of both products and customers into their representations. We also use the dual product embeddings to capture the intrinsic properties of complementariness, such as asymmetry. The separating hyperplane theory sheds light on the geometric interpretation of using the additional embedding. We conduct extensive evaluations on our model before final production, and propose a novel ranking criterion based on product and customer embeddings. Our method compares favorably to existing approaches in various offline and online testings, and case studies demonstrate the advantage and usefulness of the dual product embeddings as well as the user embeddings.

We propose a novel ranking model that combines the Bradley-Terry-Luce probability model with a nonnegative matrix factorization framework to model and uncover the presence of latent variables that influence the performance of top tennis players. We derive an efficient, provably convergent, and numerically stable majorization-minimization-based algorithm to maximize the likelihood of datasets under the proposed statistical model. The model is tested on datasets involving the outcomes of matches between 20 top male and female tennis players over 14 major tournaments for men (including the Grand Slams and the ATP Masters 1000) and 16 major tournaments for women over the past 10 years. Our model automatically infers that the surface of the court (e.g., clay or hard court) is a key determinant of the performances of male players, but less so for females. Top players on various surfaces over this longitudinal period are also identified in an objective manner.

Multilayer switch networks are proposed as artificial generators of high-dimensional discrete data (e.g., binary vectors, categorical data, natural language, network log files, and discrete-valued time series). Unlike deconvolution networks which generate continuous-valued data and which consist of upsampling filters and reverse pooling layers, multilayer switch networks are composed of adaptive switches which model conditional distributions of discrete random variables. An interpretable, statistical framework is introduced for training these nonlinear networks based on a maximum-likelihood objective function. To learn network parameters, stochastic gradient descent is applied to the objective. This direct optimization is stable until convergence, and does not involve back-propagation over separate encoder and decoder networks, or adversarial training of dueling networks. While training remains tractable for moderately sized networks, Markov-chain Monte Carlo (MCMC) approximations of gradients are derived for deep networks which contain latent variables. The statistical framework is evaluated on synthetic data, high-dimensional binary data of handwritten digits, and web-crawled natural language data. Aspects of the model's framework such as interpretability, computational complexity, and generalization ability are discussed.

Lifted inference scales to large probability models by exploiting symmetry. However, existing exact lifted inference techniques do not apply to general factor graphs, as they require a relational representation. In this work we provide a theoretical framework and algorithm for performing exact lifted inference on symmetric factor graphs by computing colored graph automorphisms, as is often done for approximate lifted inference. Our key insight is to represent variable assignments directly in the colored factor graph encoding. This allows us to generate representatives and compute the size of each orbit of the symmetric distribution. In addition to exact inference, we use this encoding to implement an MCMC algorithm that explores the space of orbits quickly by uniform orbit sampling.

Gaussian processes (GP) are a popular Bayesian approach for the optimization of black-box functions. Despite their effectiveness in simple problems, GP-based algorithms hardly scale to complex high-dimensional functions, as their per-iteration time and space cost is at least quadratic in the number of dimensions $d$ and iterations $t$. Given a set of $A$ alternative to choose from, the overall runtime $O(t^3A)$ quickly becomes prohibitive. In this paper, we introduce BKB (budgeted kernelized bandit), a novel approximate GP algorithm for optimization under bandit feedback that achieves near-optimal regret (and hence near-optimal convergence rate) with near-constant per-iteration complexity and no assumption on the input space or covariance of the GP. Combining a kernelized linear bandit algorithm (GP-UCB) with randomized matrix sketching technique (i.e., leverage score sampling), we prove that selecting inducing points based on their posterior variance gives an accurate low-rank approximation of the GP, preserving variance estimates and confidence intervals. As a consequence, BKB does not suffer from variance starvation, an important problem faced by many previous sparse GP approximations. Moreover, we show that our procedure selects at most $\tilde{O}(d_{eff})$ points, where $d_{eff}$ is the effective dimension of the explored space, which is typically much smaller than both $d$ and $t$. This greatly reduces the dimensionality of the problem, thus leading to a $O(TAd_{eff}^2)$ runtime and $O(A d_{eff})$ space complexity.

Markov chain Monte Carlo (MCMC) algorithms are widely used to sample from complicated distributions, especially to sample from the posterior distribution in Bayesian inference. However, MCMC is not directly applicable when facing the doubly intractable problem. In this paper, we discussed and compared two existing solutions -- Pseudo-marginal Monte Carlo and Exchange Algorithm. This paper also proposes a novel algorithm: Multi-armed Bandit MCMC (MABMC), which chooses between two (or more) randomized acceptance ratios in each step. MABMC could be applied directly to incorporate Pseudo-marginal Monte Carlo and Exchange algorithm, with higher average acceptance probability.

One of the biggest challenges in the field of biomedical imaging is the comprehension and the exploitation of the photon scattering through disordered media. Many studies have pursued the solution to this puzzle, achieving light-focusing control or reconstructing images in complex media. In the present work, we investigate how statistical inference helps the calculation of the transmission matrix in a complex scrambling environment, enabling its usage like a normal optical element. We convert a linear input-output transmission problem into a statistical formulation based on pseudolikelihood maximization, learning the coupling matrix via random sampling of intensity realizations. Our aim is to uncover insights from the scattering problem, encouraging the development of novel imaging techniques for better medical investigations, borrowing a number of statistical tools from spin-glass theory.

Phishing as one of the most well-known cybercrime activities is a deception of online users to steal their personal or confidential information by impersonating a legitimate website. Several machine learning-based strategies have been proposed to detect phishing websites. These techniques are dependent on the features extracted from the website samples. However, few studies have actually considered efficient feature selection for detecting phishing attacks. In this work, we investigate an agreement on the definitive features which should be used in phishing detection. We apply Fuzzy Rough Set (FRS) theory as a tool to select most effective features from three benchmarked data sets. The selected features are fed into three often used classifiers for phishing detection. To evaluate the FRS feature selection in developing a generalizable phishing detection, the classifiers are trained by a separate out-of-sample data set of 14,000 website samples. The maximum F-measure gained by FRS feature selection is 95% using Random Forest classification. Also, there are 9 universal features selected by FRS over all the three data sets. The F-measure value using this universal feature set is approximately 93% which is a comparable result in contrast to the FRS performance. Since the universal feature set contains no features from third-part services, this finding implies that with no inquiry from external sources, we can gain a faster phishing detection which is also robust toward zero-day attacks.