We propose a hierarchical tensor-network approach for approximating high-dimensional probability density via empirical distribution. This leverages randomized singular value decomposition (SVD) techniques and involves solving linear equations for tensor cores in this tensor network. The complexity of the resulting algorithm scales linearly in the dimension of the high-dimensional density. An analysis of estimation error demonstrates the effectiveness of this method through several numerical experiments.

Learning causal relationships solely from observational data provides insufficient information about the underlying causal mechanism and the search space of possible causal graphs. As a result, often the search space can grow exponentially for approaches such as Greedy Equivalence Search (GES) that uses a score-based approach to search the space of equivalence classes of graphs. Prior causal information such as the presence or absence of a causal edge can be leveraged to guide the discovery process towards a more restricted and accurate search space. In this study, we present KGS, a knowledge-guided greedy score-based causal discovery approach that uses observational data and structural priors (causal edges) as constraints to learn the causal graph. KGS is a novel application of knowledge constraints that can leverage any of the following prior edge information between any two variables: the presence of a directed edge, the absence of an edge, and the presence of an undirected edge. We extensively evaluate KGS across multiple settings in both synthetic and benchmark real-world datasets. Our experimental results demonstrate that structural priors of any type and amount are helpful and guide the search process towards an improved performance and early convergence.

The stochastic generalised linear bandit is a well-understood model for sequential decision-making problems, with many algorithms achieving near-optimal regret guarantees under immediate feedback. However, the stringent requirement for immediate rewards is unmet in many real-world applications where the reward is almost always delayed. We study the phenomenon of delayed rewards in generalised linear bandits in a theoretical manner. We show that a natural adaptation of an optimistic algorithm to the delayed feedback achieves a regret bound where the penalty for the delays is independent of the horizon. This result significantly improves upon existing work, where the best known regret bound has the delay penalty increasing with the horizon. We verify our theoretical results through experiments on simulated data.

Conditional randomization tests (CRTs) assess whether a variable $x$ is predictive of another variable $y$, having observed covariates $z$. CRTs require fitting a large number of predictive models, which is often computationally intractable. Existing solutions to reduce the cost of CRTs typically split the dataset into a train and test portion, or rely on heuristics for interactions, both of which lead to a loss in power. We propose the decoupled independence test (DIET), an algorithm that avoids both of these issues by leveraging marginal independence statistics to test conditional independence relationships. DIET tests the marginal independence of two random variables: $F(x \mid z)$ and $F(y \mid z)$ where $F(\cdot \mid z)$ is a conditional cumulative distribution function (CDF). These variables are termed "information residuals." We give sufficient conditions for DIET to achieve finite sample type-1 error control and power greater than the type-1 error rate. We then prove that when using the mutual information between the information residuals as a test statistic, DIET yields the most powerful conditionally valid test. Finally, we show DIET achieves higher power than other tractable CRTs on several synthetic and real benchmarks.

No free lunch theorems for supervised learning state that no learner can solve all problems or that all learners achieve exactly the same accuracy on average over a uniform distribution on learning problems. Accordingly, these theorems are often referenced in support of the notion that individual problems require specially tailored inductive biases. While virtually all uniformly sampled datasets have high complexity, real-world problems disproportionately generate low-complexity data, and we argue that neural network models share this same preference, formalized using Kolmogorov complexity. Notably, we show that architectures designed for a particular domain, such as computer vision, can compress datasets on a variety of seemingly unrelated domains. Our experiments show that pre-trained and even randomly initialized language models prefer to generate low-complexity sequences. Whereas no free lunch theorems seemingly indicate that individual problems require specialized learners, we explain how tasks that often require human intervention such as picking an appropriately sized model when labeled data is scarce or plentiful can be automated into a single learning algorithm. These observations justify the trend in deep learning of unifying seemingly disparate problems with an increasingly small set of machine learning models.

How to behave efficiently and flexibly is a central problem for understanding biological agents and creating intelligent embodied AI. It has been well known that behavior can be classified as two types: reward-maximizing habitual behavior, which is fast while inflexible; and goal-directed behavior, which is flexible while slow. Conventionally, habitual and goal-directed behaviors are considered handled by two distinct systems in the brain. Here, we propose to bridge the gap between the two behaviors, drawing on the principles of variational Bayesian theory. We incorporate both behaviors in one framework by introducing a Bayesian latent variable called "intention". The habitual behavior is generated by using prior distribution of intention, which is goal-less; and the goal-directed behavior is generated by the posterior distribution of intention, which is conditioned on the goal. Building on this idea, we present a novel Bayesian framework for modeling behaviors. Our proposed framework enables skill sharing between the two kinds of behaviors, and by leveraging the idea of predictive coding, it enables an agent to seamlessly generalize from habitual to goal-directed behavior without requiring additional training. The proposed framework suggests a fresh perspective for cognitive science and embodied AI, highlighting the potential for greater integration between habitual and goal-directed behaviors.

News can convey bearish or bullish views on financial assets. Institutional investors need to evaluate automatically the implied news sentiment based on textual data. Given the huge amount of news articles published each day, most of which are neutral, we present a systematic news screening method to identify the ``true'' impactful ones, aiming for more effective development of news sentiment learning methods. Based on several liquidity-driven variables, including volatility, turnover, bid-ask spread, and book size, we associate each 5-min time bin to one of two specific liquidity modes. One represents the ``calm'' state at which the market stays for most of the time and the other, featured with relatively higher levels of volatility and trading volume, describes the regime driven by some exogenous events. Then we focus on the moments where the liquidity mode switches from the former to the latter and consider the news articles published nearby impactful. We apply naive Bayes on these filtered samples for news sentiment classification as an illustrative example. We show that the screened dataset leads to more effective feature capturing and thus superior performance on short-term asset return prediction compared to the original dataset.

We consider the problem of Bayesian parameter estimation for deep neural networks, which is important in problem settings where we may have little data, and/ or where we need accurate posterior predictive densities p(y|x, D), e.g., for applications involving bandits or active learning. One simple approach to this is to use online Monte Carlo methods, such as SGLD (stochastic gradient Langevin dynamics). Unfortunately, such a method needs to store many copies of the parameters (which wastes memory), and needs to make predictions using many versions of the model (which wastes time). We describe a method for "distilling" a Monte Carlo approximation to the posterior predictive density into a more compact form, namely a single deep neural network. We compare to two very recent approaches to Bayesian neural networks, namely an approach based on expectation propagation [HLA15] and an approach based on variational Bayes [BCKW15]. Our method performs better than both of these, is much simpler to implement, and uses less computation at test time.

Mean field variational Bayes (MFVB) is a popular posterior approximation method due to its fast runtime on large-scale data sets. However, a well known major failing of MFVB is that it underestimates the uncertainty of model variables (sometimes severely) and provides no information about model variable covariance. We generalize linear response methods from statistical physics to deliver accurate uncertainty estimates for model variables--both for individual variables and coherently across variables. We call our method linear response variational Bayes (LRVB). When the MFVB posterior approximation is in the exponential family, LRVB has a simple, analytic form, even for non-conjugate models. Indeed, we make no assumptions about the form of the true posterior. We demonstrate the accuracy and scalability of our method on a range of models for both simulated and real data.

Variational inference is a scalable technique for approximate Bayesian inference. Deriving variational inference algorithms requires tedious model-specific calculations; this makes it difficult for non-experts to use.