arXiv research

Tribuo integrates ML models with provenance in Java.

Vamsa tracks data usage in Python scripts for ML models.

The paper aims to improve reproducibility of machine learning experiments.

PrIU optimizes machine learning model updates after data cleaning.

Many methods for automated software test generation, including some that explicitly use machine learning (and some that use ML more broadly conceived) derive new tests from existing tests (often referred to as seeds). Often, the seed tests from which new tests are derived are manually constructed, or at least simpler t…

Study reveals heavy-tailed behavior in training ReLU gates.

Develops a technique to audit text-generation models trained on personal data.

A new algorithm splits Gaussian processes for efficient streaming data.

It is proven that a local Lie algebra in the sense of A. A. Kirillov determines the base manifold up to a diffeomorphism provided the anchor map is nowhere-vanishing. In particular, the Lie algebras of nowhere-vanishing Poisson or Jacobi brackets determine manifolds. This result has been proven for different types of d…

New approach identifies and explains errors in machine learning pipelines.

Exploratory data analysis is crucial for developing and understanding classification models from high-dimensional datasets. We explore the utility of a new unsupervised tree ensemble called uncharted forest for visualizing class associations, sample-sample associations, class heterogeneity, and uninformative classes fo…

Multiple conventions have been adopted for denoting Interval Exchange Transformations (IETs). The "non-labeled" convention was the original, while the "labeled" convention has proven convenient when investigating Flat Surfaces as described by IETs. We establish the relationship between Extended Rauzy Classes, an equiva…

The positive energy theorem is proven for certain spacetimes with irregular curvature.

We discuss several uses of blockchain (and, more generally, distributed ledger) technologies outside of cryptocurrencies with a pragmatic view. We mostly focus on three areas: the role of coin economies for what we refer to as data malls (specialized data marketplaces); data provenance (a historical record of data and …

New homomorphism proven using immersed curves on disks.

PROVEN extends neural network robustness verification to probabilistic settings.

Kyle's equilibrium model stability proven for 1-2 trading times, but not for 3 or more.

This paper is devoted to the specific class of pseudoconformal mappings of quaternion and octonion variables. Normal families of functions are defined and investigated. Four criteria of a family being normal are proven. Then groups of pseudoconformal diffeomorphisms of quaternion and octonion manifolds are investigated…

A general Lefschetz formula for the geodesic action on locally symmetric spaces is proven.

In the paper titled "Bockstein basis and resolution theorems in extension theory" (arXiv:0907.0491v2), we stated a theorem that we claimed to be a generalization of the Edwards-Walsh resolution theorem. The goal of this note is to show that the main theorem from (arXiv:0907.0491v2) is in fact equivalent to the Edwards-…

Mini-Hes improves LFA model performance on HDI tasks with missing data.

Closed hyperbolic manifolds are proven to minimize volume over all Alexandrov spaces with curvature bounded below by -1 in the same bilipschitz class. As a corollary compact convex cores with totally geodesic boundary are proven to minimize volume over all hyperbolic manifolds in the same bilipschitz class. Also, close…

Penrose conjecture proven for specific initial data sets.

BigNN classifier improves nearest neighbor classification for large datasets.

Extends SGM to functional spaces for multimodal data.

There are proven few analogues of the Theorem of Moser using The Approximation Theorem of Artin.

We propose a clustering-based iterative algorithm to solve certain optimization problems in machine learning, where we start the algorithm by aggregating the original data, solving the problem on aggregated data, and then in subsequent steps gradually disaggregate the aggregated data. We apply the algorithm to common m…

Method learns cell interaction rules from individual trajectories.

We study a fully nonlinear PDE involving a linear combination of symmetric polynomials of the Kähler form on a Kähler manifold. A $C^0$ \emph{a priori} estimate is proven in general and a gradient estimate is proven in certain cases. Independently, we also provide a method-of-continuity proof via a path of Kähler metri…

Unsupervised classification methods learn a discriminative classifier from unlabeled data, which has been proven to be an effective way of simultaneously clustering the data and training a classifier from the data. Various unsupervised classification methods obtain appealing results by the classifiers learned in an uns…

A universal inequality that bounds the charge of a body by its size is presented, and is proven as a consequence of the Einstein equations in the context of initial data sets which satisfy an appropriate energy condition. We also present a general sufficient condition for the formation of black holes due to concentrati…

Geometric zeta functions of Ihara and Hashimoto are generalized to higher rank. The $p$-adic version of the Patterson conjecture is proven.

We develop the Ercolani-Sinha construction of SU(2) monopoles and make this effective for (a five parameter family of centred) charge 3 monopoles. In particular we show how to solve the transcendental constraints arising on the spectral curve. For a class of symmetric curves the transcendental constraints become a numb…

It is proven a new analogue of the Theorem of Moser in a generalized context defined by Shilov Boundaries of Bounded and Symmetric Domains.

Equilibrium pricing has been proven to underlie the rational Insured expectancy of premia additivity for composition of policies fully covering independent risks.

Contractibility of tight contact structures on $S^3$ proven.

We present a general sufficient condition for the formation of black holes due to concentration of angular momentum. This is expressed in the form of a universal inequality, relating the size and angular momentum of bodies, and is proven in the context of axisymmetric initial data sets for the Einstein equations which …

Combines Fourier methods and RNNs for efficient time series prediction.

Universal Gaussian parity proven for 2D knots.

The stable Khovanov-Rozansky homology of torus knots has been conjecturally described as the Koszul homology of an explicit non-regular sequence of polynomials. We verify this conjecture against newly available computational data for sl(3)-homology. Special attention is paid to torsion. In addition, explicit conjectura…

A surgery classification theory is introduced for manifolds of bounded geometry up to quasi-isometry. The Borel conjecture for this theory is proven for flat Euclidean space.

We show how binary classification methods developed to work on i.i.d. data can be used for solving statistical problems that are seemingly unrelated to classification and concern highly-dependent time series. Specifically, the problems of time-series clustering, homogeneity testing and the three-sample problem are addr…

Stochastic gradient MCMC (SG-MCMC) algorithms have proven useful in scaling Bayesian inference to large datasets under an assumption of i.i.d data. We instead develop an SG-MCMC algorithm to learn the parameters of hidden Markov models (HMMs) for time-dependent data. There are two challenges to applying SG-MCMC in this…

SVRN accelerates Newton methods by reducing variance and improving performance.

Establishes a Penrose-type inequality for axisymmetric initial data with angular momentum and charge.

It is proven that the volume of an infinitesimally flexible polyhedron in $R^3$ is a multiple root of its volume polynomial.

Study on random knot diagrams and their probability of forming specific knots.

Researchers solved a model of an exhaustible resource with stochastic discoveries.