A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
Extended elliptical slice sampling for infinite-dimensional spaces, proving reversibility.
Elliptical slice sampling converges geometrically, providing reliable sampling for Bayesian learning.
A new algorithm speeds up elliptical slice sampling for truncated multivariate normals.
Efficiently implements polar slice sampling for high-dimensional distributions.
In this work, it is shown that a simply-connected, rationally-elliptic torus orbifold is equivariantly rationally homotopy equivalent to the quotient of a product of spheres by an almost-free, linear torus action, where this torus has rank equal to the number of odd-dimensional spherical factors in the product. As an a…
This work optimizes MCMC algorithms for modern accelerators without synchronization overheads.
Probabilistic models are conceptually powerful tools for finding structure in data, but their practical effectiveness is often limited by our ability to perform inference in them. Exact inference is frequently intractable, so approximate inference is often performed using Markov chain Monte Carlo (MCMC). To achieve the…
Identifies half-space neighborhoods of pleating rays in the Riley slice of Schottky groups.
Many probabilistic models introduce strong dependencies between variables using a latent multivariate Gaussian distribution or a Gaussian process. We present a new Markov chain Monte Carlo algorithm for performing inference in models with multivariate Gaussian priors. Its key properties are: 1) it has simple, generic c…
Nested Slice Sampling accelerates Nested Sampling for GPU acceleration.
New algorithm speeds up SVAR inference for large datasets.
The paper provides a new inequality for 4-manifolds and uses it to study knot sliceness and symplectic embeddings.
A new slicing method speeds up sliced Wasserstein estimation.
We consider several geometric inequalities in general relativity involving mass, area, charge, and angular momentum for asymptotically hyperboloidal initial data. We show how to reduce each one to the known maximal (or time symmetric) case in the asymptotically flat setting, whenever a geometrically motivated system of…
ESS improves MCMC efficiency for correlated & multimodal distributions.
Bayesian inference was once a gold standard for learning with neural networks, providing accurate full predictive distributions and well calibrated uncertainty. However, scaling Bayesian inference techniques to deep neural networks is challenging due to the high dimensionality of the parameter space. In this paper, we …
A new slicing method reduces computational cost for cross-domain alignment.
Based on ideas of L. Alías, D. Impera and M. Rigoli developed in "Hypersurfaces of constant higher order mean curvature in warped products", we develope a fairly general weak/Omori-Yau maximum principle for trace operators. We apply this version of maximum principle to generalize several higher order mean curvature est…
This work investigates the properties of Gaussian-smoothed sliced divergences for comparing distributions.
ESS-Flow guides flow models without retraining, using Bayesian inference in source space.
We unify slice sampling and Hamiltonian Monte Carlo (HMC) sampling, demonstrating their connection via the Hamiltonian-Jacobi equation from Hamiltonian mechanics. This insight enables extension of HMC and slice sampling to a broader family of samplers, called Monomial Gamma Samplers (MGS). We provide a theoretical anal…
We propose an exact slice sampler for Hierarchical Dirichlet process (HDP) and its associated mixture models (Teh et al., 2006). Although there are existing MCMC algorithms for sampling from the HDP, a slice sampler has been missing from the literature. Slice sampling is well-known for its desirable properties includin…
A new method for comparing image probability measures using convolution operators.
The paper explores statistical and topological properties of sliced probability divergences.
New method constructs translationally equivariant hyperbolic affine spheres.
Local well-posedness proved for Bartnik static extension near Schwarzschild spheres.
Proposes a new distance metric for multi-marginal optimal transport.
Generative adversarial nets (GANs) and variational auto-encoders have significantly improved our distribution modeling capabilities, showing promise for dataset augmentation, image-to-image translation and feature learning. However, to model high-dimensional distributions, sequential training and stacked architectures …
An irreducible representation of the free group on two generators X,Y into SL(2,C) is determined up to conjugation by the traces of X,Y and XY. We study the diagonal slice of representations for which X,Y and XY have equal trace. Using the three-fold symmetry and Keen-Series pleating rays we locate those groups which a…
Study efficient iterative method for distribution matching using sliced optimal transport.
Conformally compact and complete smooth solutions to the Strominger system with non vanishing flux, non-trivial instanton and non-constant dilaton using the first Pontrjagin form of the (-)-connection} on 6-dimensional non-Kaehler nilmanifold are presented. In the conformally compact case the dilaton is determined by t…
In this article we construct a family of knot surgery -manifolds admitting arbitrarily many nonisomorphic Lefschetz fibration structures with the same genus fiber. We obtain such families by performing knot surgery on an elliptic surface using connected sums of fibered knots obtained by Stallings twist from a…
Proposes an energy-based sliced Wasserstein distance for improved probability measure comparison.
CB-SLICE identifies concept-based error slices in deep learning models.
A new approach simplifies Sliced-Wasserstein distances to improve learning performance.
A new method for estimating SW from streaming data.
The application of standard sufficient dimension reduction methods for reducing the dimension space of predictors without losing regression information requires inverting the covariance matrix of the predictors. This has posed a number of challenges especially when analyzing high-dimensional data sets in which the numb…
Integrals of linearly constrained multivariate Gaussian densities are a frequent problem in machine learning and statistics, arising in tasks like generalized linear models and Bayesian optimization. Yet they are notoriously hard to compute, and to further complicate matters, the numerical values of such integrals may …
Gaussian processes are flexible function approximators, with inductive biases controlled by a covariance kernel. Learning the kernel is the key to representation learning and strong predictive performance. In this paper, we develop functional kernel learning (FKL) to directly infer functional posteriors over kernels. I…
Gradient-guided nested sampling improves posterior inference efficiency.
The Gaussian process (GP) is a popular way to specify dependencies between random variables in a probabilistic model. In the Bayesian framework the covariance structure can be specified using unknown hyperparameters. Integrating over these hyperparameters considers different possible explanations for the data when maki…
Empower efficient representation of distributions through moment-preserving methods.
Exploratory cancer drug studies test multiple tumor cell lines against multiple candidate drugs. The goal in each paired (cell line, drug) experiment is to map out the dose-response curve of the cell line as the dose level of the drug increases. We propose Bayesian Tensor Filtering (BTF), a hierarchical Bayesian model …
A permutation-based SW test achieves minimax-optimal power for two-sample testing.
Proposes an online method for high-dimensional streaming data.
Sharp bounds for max-sliced Wasserstein distances derived for empirical distributions.
A new distance measure balances projection exploration and informativeness.
In this paper we discuss a class of AutoEncoder based generative models based on one dimensional sliced approach. The idea is based on the reduction of the discrimination between samples to one-dimensional case. Our experiments show that methods can be divided into two groups. First consists of methods which are a modi…