In this paper, we investigate the multi-variate sequence classification problem from a multi-instance learning perspective. Real-world sequential data commonly show discriminative patterns only at specific time periods. For instance, we can identify a cropland during its growing season, but it looks similar to a barren land after harvest or before planting. Besides, even within the same class, the discriminative patterns can appear in different periods of sequential data. Due to such property, these discriminative patterns are also referred to as shifting patterns. The shifting patterns in sequential data severely degrade the performance of traditional classification methods without sufficient training data. We propose a novel sequence classification method by automatically mining shifting patterns from multi-variate sequence. The method employs a multi-instance learning approach to detect shifting patterns while also modeling temporal relationships within each multi-instance bag by an LSTM model to further improve the classification performance. We extensively evaluate our method on two real-world applications - cropland mapping and affective state recognition. The experiments demonstrate the superiority of our proposed method in sequence classification performance and in detecting discriminative shifting patterns.

Dynamic Boltzmann Machine (DyBM) has been shown highly efficient to predict time-series data. Gaussian DyBM is a DyBM that assumes the predicted data is generated by a Gaussian distribution whose first-order moment (mean) dynamically changes over time but its second-order moment (variance) is fixed. However, in many financial applications, the assumption is quite limiting in two aspects. First, even when the data follows a Gaussian distribution, its variance may change over time. Such variance is also related to important temporal economic indicators such as the market volatility. Second, financial time-series data often requires learning datasets generated by the generalized Gaussian distribution with an additional shape parameter that is important to approximate heavy-tailed distributions. Addressing those aspects, we show how to extend DyBM that results in significant performance improvement in predicting financial time-series data.

Sample-based tree search (SBTS) is an approach to solving Markov decision problems based on constructing a lookahead search tree using random samples from a generative model of the MDP. It encompasses Monte Carlo tree search (MCTS) algorithms like UCT as well as algorithms such as sparse sampling. SBTS is well-suited to solving MDPs with large state spaces due to the relative insensitivity of SBTS algorithms to the size of the state space. The limiting factor in the performance of SBTS tends to be the exponential dependence of sample complexity on the depth of the search tree. The number of samples required to build a search tree is O((|A|B)^d), where |A| is the number of available actions, B is the number of possible random outcomes of taking an action, and d is the depth of the tree. State abstraction can be used to reduce B by aggregating random outcomes together into abstract states. Recent work has shown that abstract tree search often performs substantially better than tree search conducted in the ground state space. This paper presents a theoretical and empirical evaluation of tree search with both fixed and adaptive state abstractions. We derive a bound on regret due to state abstraction in tree search that decomposes abstraction error into three components arising from properties of the abstraction and the search algorithm. We describe versions of popular SBTS algorithms that use fixed state abstractions, and we introduce the Progressive Abstraction Refinement in Sparse Sampling (PARSS) algorithm, which adapts its abstraction during search. We evaluate PARSS as well as sparse sampling with fixed abstractions on 12 experimental problems, and find that PARSS outperforms search with a fixed abstraction and that search with even highly inaccurate fixed abstractions outperforms search without abstraction. These results establish progressive abstraction refinement as a promising basis for new tree search algorithms, and we propose directions for future work within the progressive refinement framework.

This paper introduces the R package sgmcmc; which can be used for Bayesian inference on problems with large datasets using stochastic gradient Markov chain Monte Carlo (SGMCMC). Traditional Markov chain Monte Carlo (MCMC) methods, such as Metropolis-Hastings, are known to run prohibitively slowly as the dataset size increases. SGMCMC solves this issue by only using a subset of data at each iteration. SGMCMC requires calculating gradients of the log likelihood and log priors, which can be time consuming and error prone to perform by hand. The sgmcmc package calculates these gradients itself using automatic differentiation, making the implementation of these methods much easier. To do this, the package uses the software library TensorFlow, which has a variety of statistical distributions and mathematical operations as standard, meaning a wide class of models can be built using this framework. SGMCMC has become widely adopted in the machine learning literature, but less so in the statistics community. We believe this may be partly due to lack of software; this package aims to bridge this gap.

We seek to infer the parameters of an ergodic Markov process from samples taken independently from the steady state. Our focus is on non-equilibrium processes, where the steady state is not described by the Boltzmann measure, but is generally unknown and hard to compute, which prevents the application of established equilibrium inference methods. We propose a quantity we call propagator likelihood, which takes on the role of the likelihood in equilibrium processes. This propagator likelihood is based on fictitious transitions between those configurations of the system which occur in the samples. The propagator likelihood can be derived by minimising the relative entropy between the empirical distribution and a distribution generated by propagating the empirical distribution forward in time. Maximising the propagator likelihood leads to an efficient reconstruction of the parameters of the underlying model in different systems, both with discrete configurations and with continuous configurations. We apply the method to non-equilibrium models from statistical physics and theoretical biology, including the asymmetric simple exclusion process (ASEP), the kinetic Ising model, and replicator dynamics.

Markov models, or probabilistic graphical models, explicitly establish a correspondence between statistical independence in a probability distribution and certain separation criteria holding in a graph. They were originated at the interface between statistics, where Markov random fields were predominant [Darroch et al., 1980], and artificial intelligence, with a focus on Bayesian networks [Pearl, 1985, 1986]. These two models are now considered the traditional ones, but still are widely applied and nowadays there is a significant amount of research devoted to them [Daly et al., 2011, Uhler, 2012]. They both share the modelling of conditional independences: Bayesian networks relate them with acyclic directed graphs, whereas in Markov fields they are associated with undirected graphs. However, the models they represent are only equivalent under additional assumptions on the respective graphs.

Probabilistic (or Bayesian) modeling and learning offers interesting possibilities for systematic representation of uncertainty using probability theory. However, probabilistic learning often leads to computationally challenging problems. Some problems of this type that were previously intractable can now be solved on standard personal computers thanks to recent advances in Monte Carlo methods. In particular, for learning of unknown parameters in nonlinear state-space models, methods based on the particle filter (a Monte Carlo method) have proven very useful. A notoriously challenging problem, however, still occurs when the observations in the state-space model are highly informative, i.e. when there is very little or no measurement noise present, relative to the amount of process noise. The particle filter will then struggle in estimating one of the basic components for probabilistic learning, namely the likelihood $p($data$|$parameters$)$. To this end we suggest an algorithm which initially assumes that there is substantial amount of artificial measurement noise present. The variance of this noise is sequentially decreased in an adaptive fashion such that we, in the end, recover the original problem or possibly a very close approximation of it. The main component in our algorithm is a sequential Monte Carlo (SMC) sampler, which gives our proposed method a clear resemblance to the SMC^2 method. Another natural link is also made to the ideas underlying the approximate Bayesian computation (ABC). We illustrate it with numerical examples, and in particular show promising results for a challenging Wiener-Hammerstein benchmark problem.

Abstract--Manufacturers of safety-critical systems must make the case that their product is sufficiently safe for public deployment. Much of this case often relies upon critical event outcomes from real-world testing, requiring manufacturers to be strategic about how they allocate testing resources in order to maximize their chances of demonstrating system safety. This work frames the partially observable and belief-dependent problem of test scheduling as a Markov decision process, which can be solved efficiently to yield closed-loop manufacturer testing policies. By solving for policies over a wide range of problem formulations, we are able to provide high-level guidance for manufacturers and regulators on issues relating to the testing of safety-critical systems. This guidance spans an array of topics, including circumstances under which manufacturers should continue testing despite observed incidents, when manufacturers should test aggressively, and when regulators should increase or reduce the real-world testing requirements for an autonomous vehicle. I. INTRODUCTION Confidence must be established in safety-critical systems such as autonomous vehicles prior to their widespread release. Establishing confidence is difficult because the space of driving scenarios is vast and accidents are rare. Automotive manufacturers can build confidence by conducting test drives on public roadways and make the safety case based on the frequency of observed hazardous events like disengagements and traffic accidents. Each manufacturer must devise a testing strategy capable of providing sufficient evidence that their system is safe enough for widespread adoption. Real-world testing that is too aggressive may yield hazardous events that diminish confidence in system safety. However, a manufacturer that is reluctant to test their product may forfeit opportunities to identify and address shortcomings, and may ultimately not be able to compete in the market. The fundamental tension between the desire to thoroughly test a product and the urgency to forego further testing in favor of bringing the product to market is not unique to the automotive industry.

Sensor-based activity recognition seeks the profound high-level knowledge about human activities from multitudes of low-level sensor readings. Conventional pattern recognition approaches have made tremendous progress in the past years. However, those methods often heavily rely on heuristic hand-crafted feature extraction, which could hinder their generalization performance. Additionally, existing methods are undermined for unsupervised and incremental learning tasks. Recently, the recent advancement of deep learning makes it possible to perform automatic high-level feature extraction thus achieves promising performance in many areas. Since then, deep learning based methods have been widely adopted for the sensor-based activity recognition tasks. This paper surveys the recent advance of deep learning based sensor-based activity recognition. We summarize existing literature from three aspects: sensor modality, deep model, and application. We also present detailed insights on existing work and propose grand challenges for future research.

We compare and contrast the statistical physics and quantum physics inspired approaches for unsupervised generative modeling of classical data. The two approaches represent probabilities of observed data using energy-based models and quantum states respectively.Classical and quantum information patterns of the target datasets therefore provide principled guidelines for structural design and learning in these two approaches. Taking the restricted Boltzmann machines (RBM) as an example, we analyze the information theoretical bounds of the two approaches. We verify our reasonings by comparing the performance of RBMs of various architectures on the standard MNIST datasets.