Like all sub-fields of machine learning Bayesian Deep Learning is driven by empirical validation of its theoretical proposals. Given the many aspects of an experiment it is always possible that minor or even major experimental flaws can slip by both authors and reviewers. One of the most popular experiments used to evaluate approximate inference techniques is the regression experiment on UCI datasets. However, in this experiment, models which have been trained to convergence have often been compared with baselines trained only for a fixed number of iterations. We find that a well-established baseline, Monte Carlo dropout, when evaluated under the same experimental settings shows significant improvements. In fact, the baseline outperforms or performs competitively with methods that claimed to be superior to the very same baseline method when they were introduced. Hence, by exposing this flaw in experimental procedure, we highlight the importance of using identical experimental setups to evaluate, compare, and benchmark methods in Bayesian Deep Learning.

Abstraction of Markov Decision Processes is a useful tool for solving complex problems, as it can ignore unimportant aspects of an environment, simplifying the process of learning an optimal policy. In this paper, we propose a new algorithm for finding abstract MDPs in environments with continuous state spaces. It is based on MDP homomorphisms, a structure-preserving mapping between MDPs. We demonstrate our algorithm's ability to learns abstractions from collected experience and show how to reuse the abstractions to guide exploration in new tasks the agent encounters. Our novel task transfer method beats a baseline based on a deep Q-network.

We present a formal measure-theoretical theory of neural networks (NN) built on probability coupling theory. Our main contributions are summarized as follows. * Built on the formalism of probability coupling theory, we derive an algorithm framework, named Hierarchical Measure Group and Approximate System (HMGAS), nicknamed S-System, that is designed to learn the complex hierarchical, statistical dependency in the physical world. * We show that NNs are special cases of S-System when the probability kernels assume certain exponential family distributions. Activation Functions are derived formally. We further endow geometry on NNs through information geometry, show that intermediate feature spaces of NNs are stochastic manifolds, and prove that "distance" between samples is contracted as layers stack up. * S-System shows NNs are inherently stochastic, and under a set of realistic boundedness and diversity conditions, it enables us to prove that for large size nonlinear deep NNs with a class of losses, including the hinge loss, all local minima are global minima with zero loss errors, and regions around the minima are flat basins where all eigenvalues of Hessians are concentrated around zero, using tools and ideas from mean field theory, random matrix theory, and nonlinear operator equations. * S-System, the information-geometry structure and the optimization behaviors combined completes the analog between Renormalization Group (RG) and NNs. It shows that a NN is a complex adaptive system that estimates the statistic dependency of microscopic object, e.g., pixels, in multiple scales. Unlike clear-cut physical quantity produced by RG in physics, e.g., temperature, NNs renormalize/recompose manifolds emerging through learning/optimization that divide the sample space into highly semantically meaningful groups that are dictated by supervised labels (in supervised NNs).

Goal-oriented dialogue systems typically rely on components specifically developed for a single task or domain. This limits such systems in two different ways: If there is an update in the task domain, the dialogue system usually needs to be updated or completely re-trained. It is also harder to extend such dialogue systems to different and multiple domains. The dialogue state tracker in conventional dialogue systems is one such component - it is usually designed to fit a well-defined application domain. For example, it is common for a state variable to be a categorical distribution over a manually-predefined set of entities (Henderson et al., 2013), resulting in an inflexible and hard-to-extend dialogue system. In this paper, we propose a new approach for dialogue state tracking that can generalize well over multiple domains without incorporating any domain-specific knowledge. Under this framework, discrete dialogue state variables are learned independently and the information of a predefined set of possible values for dialogue state variables is not required. Furthermore, it enables adding arbitrary dialogue context as features and allows for multiple values to be associated with a single state variable. These characteristics make it much easier to expand the dialogue state space. We evaluate our framework using the widely used dialogue state tracking challenge data set (DSTC2) and show that our framework yields competitive results with other state-of-the-art results despite incorporating little domain knowledge. We also show that this framework can benefit from widely available external resources such as pre-trained word embeddings.

A key task in Bayesian machine learning is sampling from distributions that are only specified up to a partition function (i.e., constant of proportionality). One prevalent example of this is sampling posteriors in parametric distributions, such as latent-variable generative models. However sampling (even very approximately) can be #P-hard. Classical results going back to Bakry and \'Emery (1985) on sampling focus on log-concave distributions, and show a natural Markov chain called Langevin diffusion mixes in polynomial time. However, all log-concave distributions are uni-modal, while in practice it is very common for the distribution of interest to have multiple modes. In this case, Langevin diffusion suffers from torpid mixing. We address this problem by combining Langevin diffusion with simulated tempering. The result is a Markov chain that mixes more rapidly by transitioning between different temperatures of the distribution. We analyze this Markov chain for a mixture of (strongly) log-concave distributions of the same shape. In particular, our technique applies to the canonical multi-modal distribution: a mixture of gaussians (of equal variance). Our algorithm efficiently samples from these distributions given only access to the gradient of the log-pdf. For the analysis, we introduce novel techniques for proving spectral gaps based on decomposing the action of the generator of the diffusion. Previous approaches rely on decomposing the state space as a partition of sets, while our approach can be thought of as decomposing the stationary measure as a mixture of distributions (a "soft partition"). Additional materials for the paper can be found at http://tiny.cc/glr17. The proof and results have been improved and generalized from the precursor at www.arxiv.org/abs/1710.02736.

Stratifying patients at risk for postoperative complications may facilitate timely and accurate workups and reduce the burden of adverse events on patients and the health system. Currently, a widely-used surgical risk calculator created by the American College of Surgeons, NSQIP, uses 21 preoperative covariates to assess risk of postoperative complications, but lacks dynamic, real-time capabilities to accommodate postoperative information. We propose a new Hidden Markov Model sequence classifier for analyzing patients' postoperative temperature sequences that incorporates their time-invariant characteristics in both transition probability and initial state probability in order to develop a postoperative "real-time" complication detector. Data from elective Colectomy surgery indicate that our method has improved classification performance compared to 8 other machine learning classifiers when using the full temperature sequence associated with the patients' length of stay. Additionally, within 44 hours after surgery, the performance of the model is close to that of full-length temperature sequence.

Smoothed analysis is a powerful paradigm in overcoming worst-case intractability in unsupervised learning and high-dimensional data analysis. While polynomial time smoothed analysis guarantees have been obtained for worst-case intractable problems like tensor decompositions and learning mixtures of Gaussians, such guarantees have been hard to obtain for several other important problems in unsupervised learning. A core technical challenge is obtaining lower bounds on the least singular value for random matrix ensembles with dependent entries, that are given by low-degree polynomials of a few base underlying random variables. In this work, we address this challenge by obtaining high-confidence lower bounds on the least singular value of new classes of structured random matrix ensembles of the above kind. We then use these bounds to obtain polynomial time smoothed analysis guarantees for the following three important problems in unsupervised learning: 1. Robust subspace recovery, when the fraction $\alpha$ of inliers in the d-dimensional subspace $T \subset \mathbb{R}^n$ is at least $\alpha > (d/n)^\ell$ for any constant integer $\ell>0$. This contrasts with the known worst-case intractability when $\alpha< d/n$, and the previous smoothed analysis result which needed $\alpha > d/n$ (Hardt and Moitra, 2013). 2. Higher order tensor decompositions, where we generalize the so-called FOOBI algorithm of Cardoso to find order-$\ell$ rank-one tensors in a subspace. This allows us to obtain polynomially robust decomposition algorithms for $2\ell$'th order tensors with rank $O(n^{\ell})$. 3. Learning overcomplete hidden markov models, where the size of the state space is any polynomial in the dimension of the observations. This gives the first polynomial time guarantees for learning overcomplete HMMs in a smoothed analysis model.

One of the main challenges of deep learning tools is their inability to capture model uncertainty. While Bayesian deep learning can be used to tackle the problem, Bayesian neural networks often require more time and computational power to train than deterministic networks. Our work explores whether fully Bayesian networks are needed to successfully capture model uncertainty. We vary the number and position of Bayesian layers in a network and compare their performance on active learning with the MNIST dataset. We found that we can fully capture the model uncertainty by using only a few Bayesian layers near the output of the network, combining the advantages of deterministic and Bayesian networks.

Generalization is one of the most important issues in machine learning problems. In this paper, we consider the generalization in restricted Boltzmann machines. We propose a restricted Boltzmann machine with multivalued hidden variables, which is a simple extension of conventional restricted Boltzmann machines. We demonstrate that our model is better than the conventional one via numerical experiments: experiments for a contrastive divergence learning with artificial data and for a classification problem with MNIST.

Current state-of-the-art nonparametric Bayesian text clustering methods model documents through multinomial distribution on bags of words. Although these methods can effectively utilize the word burstiness representation of documents and achieve decent performance, they do not explore the sequential information of text and relationships among synonyms. In this paper, the documents are modeled as the joint of bags of words, sequential features and word embeddings. We proposed Sequential Embedding induced Dirichlet Process Mixture Model (SiDPMM) to effectively exploit this joint document representation in text clustering. The sequential features are extracted by the encoder-decoder component. Word embeddings produced by the continuous-bag-of-words (CBOW) model are introduced to handle synonyms. Experimental results demonstrate the benefits of our model in two major aspects: 1) improved performance across multiple diverse text datasets in terms of the normalized mutual information (NMI); 2) more accurate inference of ground truth cluster numbers with regularization effect on tiny outlier clusters.