arXiv research

Double descent risk in L2-regularized models explained and mitigated.

Optimal regularization can prevent the double descent phenomenon in learning models.

Classifies periodic points on regular and double n-gon surfaces.

Paper shows certain algebra types are not differentially smooth.

New proof for convex solutions of Monge-Ampère equation.

Hybrid regularization avoids double descent in random feature models.

Study on connection points on double regular polygons, providing coordinates and proving non-connection points.

In this paper we study 1/k-geodesics, those closed geodesics that minimize on any subinterval of length $L/k$, where $L$ is the length of the geodesic. We investigate the existence and behavior of these curves on doubled polygons and show that every doubled regular $n$-gon admits a $1/2n$-geodesic. For the doubled regu…

New theorem for generalized group sparsity improves consistency and convergence rates.

Improves neural network performance with double regularization.

The paper calculates gap distributions for translation surfaces, focusing on the double heptagon.

Dropout improves regularization in flexible models for rare features.

Research provides explicit NPV expressions for double barrier strategies.

New analysis reveals optimal regularization for ESNs, avoiding double descent.

A connected combinatorial 2-manifold is called degree-regular if each of its vertices have the same degree. A connected combinatorial 2-manifold is called weakly regular if it has a vertex-transitive automorphism group. Clearly, a weakly regular combinatorial 2-manifold is degree-regular and a degree-regular combinator…

The paper analyzes high-dimensional kernel regression, showing different risk curves based on data and regularization.

Efficient regularization mitigates catastrophic overfitting in single-step adversarial training.

We describe a natural $q$-deformation of Fock and Goncharov's canonical basis for the algebra of regular functions on a cluster variety associated to a quiver of type $A$. We then describe an extension of this construction involving a cluster variety called the symplectic double.

Double descent refers to the phase transition that is exhibited by the generalization error of unregularized learning models when varying the ratio between the number of parameters and the number of training samples. The recent success of highly over-parameterized machine learning models such as deep neural networks ha…

The aim of this paper is to establish two fundamental measure-metric properties of particular random geometric graphs. We consider $\varepsilon$-neighborhood graphs whose vertices are drawn independently and identically distributed from a common distribution defined on a regular submanifold of $\mathbb{R}^K$. We show t…

We consider double plumbings of two disk bundles over spheres. We calculate the Heegaard-Floer homology with its absolute grading of the boundary of such a plumbing. Given a closed smooth 4-manifold $X$ and a suitable pair of classes in $H_{2}(X)$, we investigate when this pair of classes may be represented by a config…

This paper classifies semi-equivelar gems on a double torus.

It is shown that for knots with a sufficiently regular character variety the Dubois' torsion detects the A-polynomial of the knot. A global formula for the integral of the Dubois torsion is given. The formula looks like the heat kernel regularization of the formula for the Witten-Reshetikhin-Turaev invariant of the dou…

On a Poisson manifold endowed with a Riemannian metric we will construct a vector field that generalizes the double bracket vector field defined on semi-simple Lie algebras. On a regular symplectic leaf we will construct a generalization of the normal metric such that the above vector field restricted to the symplectic…

New method for estimating parameters in inverse problems using double robustness.

The paper analyzes how over-parameterization affects reinforcement learning performance.

Kernel ridgeless regression with random features shows good generalization without explicit regularization.

One of the basic aims of this paper is to study the relationship between the geometry of ``hypersurface like'' subsets of Euclidean space and the properties of the measures they support. In this context we show that certain doubling properties of a measure determine the geometry of its support. A Radon measure is said …

Hexagonal norm double bubble problem solved with minimal configurations.

A new tradeoff between regularization and sharpness improves model performance in overparameterized settings.

Geodesic nets on flat spheres are studied using Gauss-Bonnet theorem.

Study normal operators of double fibration transforms with conjugate points.

In this work we revisit gradient regularization for adversarial robustness with some new ingredients. First, we derive new per-image theoretical robustness bounds based on local gradient information. These bounds strongly motivate input gradient regularization. Second, we implement a scaleable version of input gradient…

This paper explains why double descent sometimes occurs weakly or not at all from an optimization perspective.

The paper explores how overfitting can lead to better predictions in high-dimensional data.

Given a closed submanifold, or a compact regular domain, in euclidean space, we consider the Riesz energy defined as the double integral of some power of the distance between pairs of points. When this integral diverges, we compare two different regularization techniques (Hadamard's finite part and analytic continuatio…

Proposes fair and robust methods for estimating treatment effects.

A generic immersion of a planar graph into the 2-space is said to be knotted if there does not exist a trivial embedding of the graph into the 3-space obtained by lifting the immersion with respect to the natural projection from the 3-space to the 2-space. In this paper we show that if a generic immersion of a planar g…

Double descent found in DRL, improving generalization with model capacity.

We define an infinite series of translation coverings of Veech's double-n-gon for odd n greater or equal to 5 which share the same Veech group. Additionally we give an infinite series of translation coverings with constant Veech group of a regular n-gon for even n greater or equal to 8. These families give rise to expl…

Calculates the transient number of knots using homology groups.

This paper develops a novel analytically tractable Neumann series of Bessel functions representation for pricing (and hedging) European-style double barrier knock-out options, which can be applied to the whole class of one-dimensional time-homogeneous diffusions even for the cases where the corresponding transition den…

The paper uses double machine learning to estimate dynamic treatment effects robustly.

We introduce a general tensor model suitable for data analytic tasks for {\em heterogeneous} datasets, wherein there are joint low-rank structures within groups of observations, but also discriminative structures across different groups. To capture such complex structures, a double core tensor (DCOT) factorization mode…

The paper develops a theory for one-step Wasserstein-guided models for PDE-induced measures.

We define a computable topological invariant $μ(γ)$ for generic closed planar regular curves $γ$, which gives an effective lower bound for the number of inflection points on a given generic closed planar curve. Using it, we classify the topological types of locally convex curves (i.e. closed planar regular curves witho…

Improved portfolio optimization method reduces risk and improves performance.

We address the double bubble problem for the anisotropic Grushin perimeter $P_α$, $α\geq 0$, and the Lebesgue measure in $\mathbb R^2$, in the case of two equal volumes. We assume that the contact interface between the bubbles lays on either the vertical or the horizontal axis. Since no regularity theory is available i…