Despite the success of autoregressive large language models in text generation, it remains a major challenge to generate text that satisfies complex constraints: sampling from the conditional distribution ${\Pr}(\text{text} | \alpha)$ is intractable for even the simplest lexical constraints $\alpha$. To overcome this challenge, we propose to use tractable probabilistic models (TPMs) to impose lexical constraints in autoregressive text generation models, which we refer to as GeLaTo (Generating Language with Tractable Constraints). To demonstrate the effectiveness of this framework, we use distilled hidden Markov models, where we can efficiently compute ${\Pr}(\text{text} | \alpha)$, to guide autoregressive generation from GPT2. GeLaTo achieves state-of-the-art performance on challenging benchmarks for constrained text generation (e.g., CommonGen), beating various strong baselines by a large margin. Our work not only opens up new avenues for controlling large language models but also motivates the development of more expressive TPMs.

Relying on crowdsourced workers, data crowdsourcing platforms are able to efficiently provide vast amounts of labeled data. Due to the variability in the annotation quality of crowd workers, modern techniques resort to redundant annotations and subsequent label aggregation to infer true labels. However, these methods require model updating during the inference, posing challenges in real-world implementation. Meanwhile, in recent years, many data labeling tasks have begun to require skilled and experienced annotators, leading to an increasing demand for long-term annotators. These annotators could leave substantial historical annotation records on the crowdsourcing platforms, which can benefit label aggregation, but are ignored by previous works. Hereby, in this paper, we propose a novel label aggregation technique, which does not need any model updating during inference and can extensively explore the historical annotation records. We call it SuperLA, a Supervised Label Aggregation method. Inside this model, we design three types of input features and a straightforward neural network structure to merge all the information together and subsequently produce aggregated labels. Based on comparison experiments conducted on 22 public datasets and 11 baseline methods, we find that SuperLA not only outperforms all those baselines in inference performance but also offers significant advantages in terms of efficiency.

Unexpectedness is a central concept in Simplicity Theory, a theory of cognition relating various inferential processes to the computation of Kolmogorov complexities, rather than probabilities. Its predictive power has been confirmed by several experiments with human subjects, yet its theoretical basis remains largely unexplored: why does it work? This paper lays the groundwork for three theoretical conjectures. First, unexpectedness can be seen as a generalization of Bayes' rule. Second, the frequentist core of unexpectedness can be connected to the function of tracking ergodic properties of the world. Third, unexpectedness can be seen as constituent of various measures of divergence between the entropy of the world (environment) and the variety of the observer (system). The resulting framework hints to research directions that go beyond the division between probabilistic and logical approaches, potentially bringing new insights into the extraction of causal relations, and into the role of descriptive mechanisms in learning.

Recovering masked feedback with neural models is a popular paradigm in recommender systems. Seeing the success of diffusion models in solving ill-posed inverse problems, we introduce a conditional diffusion framework for collaborative filtering that iteratively reconstructs a user's hidden preferences guided by its historical interactions. To better align with the intrinsic characteristics of implicit feedback data, we implement forward diffusion by applying synthetic smoothing filters to interaction signals on an item-item graph. The resulting reverse diffusion can be interpreted as a personalized process that gradually refines preference scores. Through graph Fourier transform, we equivalently characterize this model as an anisotropic Gaussian diffusion in the graph spectral domain, establishing both forward and reverse formulations. Our model outperforms state-of-the-art methods by a large margin on one dataset and yields competitive results on the others.

We consider the problem of variational Bayesian inference in a latent variable model where a (possibly complex) observed stochastic process is governed by the solution of a latent stochastic differential equation (SDE). Motivated by the challenges that arise when trying to learn an (almost arbitrary) latent neural SDE from large-scale data, such as efficient gradient computation, we take a step back and study a specific subclass instead. In our case, the SDE evolves on a homogeneous latent space and is induced by stochastic dynamics of the corresponding (matrix) Lie group. In learning problems, SDEs on the unit $n$-sphere are arguably the most relevant incarnation of this setup. Notably, for variational inference, the sphere not only facilitates using a truly uninformative prior SDE, but we also obtain a particularly simple and intuitive expression for the Kullback-Leibler divergence between the approximate posterior and prior process in the evidence lower bound. Experiments demonstrate that a latent SDE of the proposed type can be learned efficiently by means of an existing one-step geometric Euler-Maruyama scheme. Despite restricting ourselves to a less diverse class of SDEs, we achieve competitive or even state-of-the-art performance on various time series interpolation and classification benchmarks.

Probabilistic control design is founded on the principle that a rational agent attempts to match modelled with an arbitrary desired closed-loop system trajectory density. The framework was originally proposed as a tractable alternative to traditional optimal control design, parametrizing desired behaviour through fictitious transition and policy densities and using the information projection as a proximity measure. In this work we introduce an alternative parametrization of desired closed-loop behaviour and explore alternative proximity measures between densities. It is then illustrated how the associated probabilistic control problems solve into uncertain or probabilistic policies. Our main result is to show that the probabilistic control objectives majorize conventional, stochastic and risk sensitive, optimal control objectives. This observation allows us to identify two probabilistic fixed point iterations that converge to the deterministic optimal control policies establishing an explicit connection between either formulations. Further we demonstrate that the risk sensitive optimal control formulation is also technically equivalent to a Maximum Likelihood estimation problem on a probabilistic graph model where the notion of costs is directly encoded into the model. The associated treatment of the estimation problem is then shown to coincide with the moment projected probabilistic control formulation. That way optimal decision making can be reformulated as an iterative inference problem. Based on these insights we discuss directions for algorithmic development.

Electronic health records contain valuable information for monitoring patients' health trajectories over time. Disease progression models have been developed to understand the underlying patterns and dynamics of diseases using these data as sequences. However, analyzing temporal data from EHRs is challenging due to the variability and irregularities present in medical records. We propose a Markovian generative model of treatments developed to (i) model the irregular time intervals between medical events; (ii) classify treatments into subtypes based on the patient sequence of medical events and the time intervals between them; and (iii) segment treatments into subsequences of disease progression patterns. We assume that sequences have an associated structure of latent variables: a latent class representing the different subtypes of treatments; and a set of latent stages indicating the phase of progression of the treatments. We use the Expectation-Maximization algorithm to learn the model, which is efficiently solved with a dynamic programming-based method. Various parametric models have been employed to model the time intervals between medical events during the learning process, including the geometric, exponential, and Weibull distributions. The results demonstrate the effectiveness of our model in recovering the underlying model from data and accurately modeling the irregular time intervals between medical actions.

Regularization is one of the most fundamental topics in optimization, statistics and machine learning. To get sparsity in estimating a parameter $u\in\mathbb{R}^d$, an $\ell_q$ penalty term, $\Vert u\Vert_q$, is usually added to the objective function. What is the probabilistic distribution corresponding to such $\ell_q$ penalty? What is the correct stochastic process corresponding to $\Vert u\Vert_q$ when we model functions $u\in L^q$? This is important for statistically modeling large dimensional objects, e.g. images, with penalty to preserve certainty properties, e.g. edges in the image. In this work, we generalize the $q$-exponential distribution (with density proportional to) $\exp{(- \frac{1}{2}|u|^q)}$ to a stochastic process named $Q$-exponential (Q-EP) process that corresponds to the $L_q$ regularization of functions. The key step is to specify consistent multivariate $q$-exponential distributions by choosing from a large family of elliptic contour distributions. The work is closely related to Besov process which is usually defined by the expanded series. Q-EP can be regarded as a definition of Besov process with explicit probabilistic formulation and direct control on the correlation length. From the Bayesian perspective, Q-EP provides a flexible prior on functions with sharper penalty ($q<2$) than the commonly used Gaussian process (GP). We compare GP, Besov and Q-EP in modeling functional data, reconstructing images, and solving inverse problems and demonstrate the advantage of our proposed methodology.

We present sparse tree-based and list-based density estimation methods for binary/categorical data. Our density estimation models are higher dimensional analogies to variable bin width histograms. In each leaf of the tree (or list), the density is constant, similar to the flat density within the bin of a histogram. Histograms, however, cannot easily be visualized in more than two dimensions, whereas our models can. The accuracy of histograms fades as dimensions increase, whereas our models have priors that help with generalization. Our models are sparse, unlike high-dimensional fixed-bin histograms. We present three generative modeling methods, where the first one allows the user to specify the preferred number of leaves in the tree within a Bayesian prior. The second method allows the user to specify the preferred number of branches within the prior. The third method returns density lists (rather than trees) and allows the user to specify the preferred number of rules and the length of rules within the prior. The new approaches often yield a better balance between sparsity and accuracy of density estimates than other methods for this task. We present an application to crime analysis, where we estimate how unusual each type of modus operandi is for a house break-in.

This paper investigates the asymptotic distribution of the K-fold cross validation error in an i.i.d. setting. As the number of observations n goes to infinity while keeping the number of folds K fixed, the K-fold cross validation error is √ n-consistent for the expected out-of-sample error and has an asymptotically normal distribution. A consistent estimate of the asymptotic variance is derived and used to construct asymptotically valid confidence intervals for the expected out-of-sample error. A hypothesis test is developed for comparing two estimators’ expected out-of-sample errors and a subsampling procedure is used to obtain critical values. Monte Carlo simulations demonstrate the asymptotic validity of our confidence intervals for the expected out-of-sample error and investigate the size and power properties of our test. In our empirical application, we use our estimator selection test to compare the out-of-sample predictive performance of OLS, Neural Networks, and Random Forests for predicting the sale price of a domain name in a GoDaddy expiry auction.