arXiv research

New concept of partial comonotonicity connects riskmetrics and dependence.

Partial dependence curves (FPD) introduced by Friedman, are an important model interpretation tool, but are often not accessible to business analysts and scientists who typically lack the skills to choose, tune, and assess machine learning models. It is also common for the same partial dependence algorithm on the same …

Study optimal policy regret in partially observable Markov games with adaptive opponents.

Study uses deep neural networks for inference in partially linear models with dependent data.

Time series models generalize ARMA and ARFIMA with non-Gaussian dependence.

Study expands multiclass classification models with new rates and partial concept classes.

New method learns frequency-dependent partial correlations.

Let $M$ be a compact hypersurface with boundary $\partial M=\partial D_1 \cup \partial D_2$, $\partial D_1 \subset Π_1$, $\partial D_2 \subset Π_2$, $Π_1$ and $Π_2$ two parallel hyperplanes in $\mathbb{R}^{n+1}$ ($n \geq 2$). Suppose that $M$ is contained in the slab determined by these hyperplanes and that the mean cu…

PDGM uses neural nets to solve complex financial equations.

Derives derivatives and geometric framework for functions with non-independent variables.

Researchers show how to manipulate Partial Dependence plots to deceive explanations of predictive models.

SHAP explains boosted trees with additively modeled features.

New algorithms improve privacy in bandit problems with partial information.

New model captures state-dependent variability in partially observed systems.

Using the adjoint action of the infinitesimal translations (with respect to some (in)dependant variables) on specific finite-dimensional subspaces of the space of generalized symmetries of some system of partial differential equations, we explicitly determine the dependance of coefficients of generalized symmetries fro…

We introduce the concept of numerical Gaussian processes, which we define as Gaussian processes with covariance functions resulting from temporal discretization of time-dependent partial differential equations. Numerical Gaussian processes, by construction, are designed to deal with cases where: (1) all we observe are …

GADGET framework decomposes global feature effects using recursive partitioning.

This paper studies the question of filtering and maximizing terminal wealth from expected utility in a partially information stochastic volatility models. The special features is that the only information available to the investor is the one generated by the asset prices, and the unobservable processes will be modeled …

New method identifies causal variables from partially observed data.

There has been rapid progress recently on the application of deep networks to the solution of partial differential equations, collectively labelled as Physics Informed Neural Networks (PINNs). In this paper, we develop Physics Informed Extreme Learning Machine (PIELM), a rapid version of PINNs which can be applied to s…

Probabilistic method combines space and time uncertainties in PDEs.

Let $(M, g(t))$, $t\in[0,T)$ be a closed Riemannian $n$-manifold whose Riemannian metric $g(t)$ evolves by the geometric flow $\frac{\partial }{\partial t} g_{ij}=-2S_{ij}$, where $S_{ij}(t)$ is a symmetric two-tensor on $(M,g(t))$. We discuss differential Harnack estimates for positive solution to the porous medium …

We describe various sets of conditional independence relationships, sufficient for qualitatively comparing non-vanishing squared partial correlations of a Gaussian random vector. These sufficient conditions are satisfied by several graphical Markov models. Rules for comparing degree of association among the vertices of…

In this paper I develop a new computational method for pricing path dependent options. Using the path integral representation of the option price, I show that in general it is possible to perform analytically a partial averaging over the underlying risk-neutral diffusion process. This result greatly eases the computati…

New method controls linear systems with partial info and disturbances.

New statistical models for predicting ranked preferences from partial orders.

New quantum algorithm simplifies complex financial derivatives pricing.

Enhances reinforcement learning with partial state information.

Researchers formalize PD and PFI to relate them to data generating process.

New approach interprets machine learning models through feature space transformations.

New method infers human sensorimotor costs from behavior.

We use PDPs with confidence bands to explain HPO results.

Study on $4$-dimensional almost-Hermitian manifolds, proving $\overline\partial$-harmonic forms invariant under certain metrics.

New findings on Malgrange-Galois groupoid for Painlevé VI equation parameters.

In this paper, we prove that there exists a dimensional constant $δ> 0$ such that given any background Kähler metric $ω$, the Calabi flow with initial data $u_0$ satisfying \begin{equation*} \partial \bar \partial u_0 \in L^\infty (M) \text{ and } (1- δ)ω< ω_{u_0} < (1+δ)ω, \end{equation*} admits a unique short time so…

A new method uses deep learning to efficiently solve complex physics equations in high dimensions.

A new approach uses partial likelihood to improve tree-based density estimation and inference.

Study on mappings in Carnot groups, proving rigidity results.

For the class of systems of PDEs, for which infinitesimal translations (with respect to some (in)dependent variables) possess specific finite-dimensional invariant subspaces of the space of generalized symmetries of the system considered. We establish when there exist generalized symmetries from these subspaces, which …

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

A new tree-based estimator, FastPD, efficiently estimates PD functions for machine learning models.

Derives FPDE for equity-linked insurance pricing.

The paper calculates MES bounds for systemic risk contributions under uncertain dependence.

Proposes a method to create predictive sets from partially labeled data.

New Thompson sampling algorithm for stochastic partial monitoring achieves logarithmic regret.

While there is currently a lot of enthusiasm about "big data", useful data is usually "small" and expensive to acquire. In this paper, we present a new paradigm of learning partial differential equations from {\em small} data. In particular, we introduce \emph{hidden physics models}, which are essentially data-efficien…

This paper detects Markov violations in RL with noise, improving policy development.

Homotopy equivalent boundaries of cube complexes are studied.