arXiv research

We tackle linear bandits with partially observable features, achieving sublinear regret.

Efficiently extracts linear dynamics from complex observations.

Study efficient reinforcement learning for partially observed systems with linear structure.

New method controls linear systems with partial info and disturbances.

Algorithm reduces regret in partially observable systems by learning dynamics and using optimistic control.

Linear recurrent networks explain reinforcement learning performance in partially observable settings.

New algorithm for partially observable contexts in finance.

A new decision tree variant improves linear model performance.

New algorithm learns linear dynamical systems from measurements.

Modeling dynamical systems is important in many disciplines, e.g., control, robotics, or neurotechnology. Commonly the state of these systems is not directly observed, but only available through noisy and potentially high-dimensional observations. In these cases, system identification, i.e., finding the measurement map…

AdaptOn achieves logarithmic regret in adaptive control of unknown partially observable linear systems.

New framework for reinforcement learning with sporadic state observations.

New method disentangles perceptual uncertainty and behavioral costs in partially observable systems.

IDS optimizes regret in stochastic partial monitoring with linear rewards.

New estimator learns symmetric dynamics from few observations.

Study learns linear system dynamics from noisy bilinear data.

Paper studies IPG method's robustness in noisy distributed linear regression.

Graphical notation simplifies complex polynomial constraints in linear models.

Extends linear structural causal models to include deterministic relations and latent confounders for causal discovery.

We solve structure learning for cyclic linear causal models using observational data.

AM converges super-linearly for solving mixed linear regression problems.

Study evaluates machine learning for predicting treatment effects in observational studies.

A new classifier uses linear programming to classify sets based on their covariance.

A new approach predicts next observations without explicit decoding for better control.

This work leverages recent advances in probabilistic machine learning to discover conservation laws expressed by parametric linear equations. Such equations involve, but are not limited to, ordinary and partial differential, integro-differential, and fractional order operators. Here, Gaussian process priors are modifie…

Max-linear regression problem solved with convex programming.

We consider optimal investment problems for a diffusion market model with non-observable random drifts that evolve as an Ito's process. Admissible strategies do not use direct observations of the market parameters, but rather use historical stock prices. For a non-linear problem with a general performance criterion, th…

Study finds exact limits for sparse regression with fewer observations than usual.

We study two inverse problems on a globally hyperbolic Lorentzian manifold $(M,g)$. The problems are: 1. Passive observations in spacetime: Consider observations in a neighborhood $V\subset M$ of a time-like geodesic $μ$. Under natural causality conditions, we reconstruct the conformal type of the unknown open, relativ…

TOFU-POV tackles partially observed linear bandits, achieving sublinear regret with low-dimensional action vectors.

Study tackles causal structure learning in linear models with unobserved variables and measurement error.

Algorithm learns linear systems from partial observations with near-optimal rate.

New method identifies latent variables with causal dependencies from observed data.

We study the performance of the certainty equivalent controller on Linear Quadratic (LQ) control problems with unknown transition dynamics. We show that for both the fully and partially observed settings, the sub-optimality gap between the cost incurred by playing the certainty equivalent controller on the true system …

Observing a linear superposition principle, a family of new minimal hypersurfaces in Euclidean space is found, as well as that linear combinations of generalized helicoids induce new algebraic minimal cones of arbitrarily high degree.

New method identifies key genes affecting phenotypes in biological systems.

We introduce the Locally Linear Latent Variable Model (LL-LVM), a probabilistic model for non-linear manifold discovery that describes a joint distribution over observations, their manifold coordinates and locally linear maps conditioned on a set of neighbourhood relationships. The model allows straightforward variatio…

Estimates causal effects in Gaussian Linear SCMs with finite data.

Matrix completion aims to predict missing elements in a partially observed data matrix which in typical applications, such as collaborative filtering, is large and extremely sparsely observed. A standard solution is matrix factorization, which predicts unobserved entries as linear combinations of latent variables. We g…

Paper proposes a new method to identify causal graphs with latent variables using higher-order cumulants.

Causal inference concerns the identification of cause-effect relationships between variables, e.g. establishing whether a stimulus affects activity in a certain brain region. The observed variables themselves often do not constitute meaningful causal variables, however, and linear combinations need to be considered. In…

In a linear stochastic bandit model, each arm is a vector in an Euclidean space and the observed return at each time step is an unknown linear function of the chosen arm at that time step. In this paper, we investigate the problem of learning the best arm in a linear stochastic bandit model, where each arm's expected r…

The paper compares and optimizes estimators for treatment effects with observed confounders and mediators.

Estimates linear models from self-selected data, addressing econometric challenges.

This paper improves system identification by reducing sample complexity for high-dimensional linear dynamical systems.

Study learns state representations from observations for control, proving guarantees.

A basic problem in machine learning is to find a mapping $f$ from a low dimensional latent space $\mathcal{Y}$ to a high dimensional observation space $\mathcal{X}$. Modern tools such as deep neural networks are capable to represent general non-linear mappings. A learner can easily find a mapping which perfectly fits a…

The paper analyzes methods for estimating linear functionals from observational data, proving upper bounds and showing optimal procedures.