arXiv research

Let X be a geodesic metric space. Gromov proved that there exists k>0 such that if every sufficiently large triangle T satisfies the Rips condition with constant k times pr(T), where pr(T) is the perimeter T, then X is hyperbolic. We give an elementary proof of this fact, also giving an estimate for k. We also show tha…

The ROC curve is widely used to assess the quality of prediction/classification/ranking algorithms, and its properties have been extensively studied. The precision-recall (PR) curve has become the de facto replacement for the ROC curve in the presence of imbalance, namely where one class is far more likely than the oth…

We consider $d$-dimensional linear stochastic approximation algorithms (LSAs) with a constant step-size and the so called Polyak-Ruppert (PR) averaging of iterates. LSAs are widely applied in machine learning and reinforcement learning (RL), where the aim is to compute an appropriate $θ_{*} \in \mathbb{R}^d$ (that is a…

TD(0) with Polyak-Ruppert averaging achieves robust and fast convergence rates

We study the triple $(G,π,\prs)$ where $G$ is a connected and simply connected Lie group, $π$ and $\prs$ are, respectively, a multiplicative Poisson tensor and a left invariant Riemannian metric on $G$ such that the necessary conditions, introduced by Hawkins, to the existence of a non commutative deformation (in the d…

When sufficient labeled data are available, classical criteria based on Receiver Operating Characteristic (ROC) or Precision-Recall (PR) curves can be used to compare the performance of un-supervised anomaly detection algorithms. However , in many situations, few or no data are labeled. This calls for alternative crite…

Solves a challenging problem in imaging and communication.

In this note we prove that, for a vector bundle $E$ over a manifold $M$, a Dorfman bracket on $TM\oplus E^*$ anchored by $\operatorname{pr}_{TM}$ and with $E$ a vector bundle over $M$, is equivalent to a lift from $Γ(TM\oplus E^*)$ to linear sections of $TE\oplus T^*E\to E$, that intertwines the given Dorfman bracket w…

Classifies solutions to vacuum weighted Einstein equations on pr-waves.

Study finds differences in LTs across tasks and architectures, proposing a consensus-based method for generating refined lottery tickets.

Proposes MCC-F1 curve for better binary classification evaluation.

The broad set of deep generative models (DGMs) has achieved remarkable advances. However, it is often difficult to incorporate rich structured domain knowledge with the end-to-end DGMs. Posterior regularization (PR) offers a principled framework to impose structured constraints on probabilistic models, but has limited …

PRS improves rejection sampling by learning better proposals.

Geometric analysis of ROC and PR curves for binary classification.

A garland based on a manifold $P$ is a finite set of manifolds homeomorphic to $P$ with some of them glued together at marked points. Fix a manifold $M$ and consider a space $\NN$ of all smooth mappings of garlands based on $P$ into $M$. We construct operations $\bullet$ and $[-,-]$ on the bordism groups $\bor_*(\NN)$ …

Steinhaus conjectured that every closed oriented $C^1$-curve has a pair of anti-parallel tangents. Porter disproved the conjecture by showing that there exist curves with no anti-parallel tangents. Colin Adams rised the question of whether there exists a nontrivial knot in $\R^3$ which has no parallel or antiparallel t…

In this paper we define a Poincaré-Reidemeister scalar product on the determinant line of the cohomology of any flat vector bundle over a closed orientable odd-dimensional manifold. It is a combinatorial "torsion-type" invariant which refines the PR-metric, introduced earlier by the first author, and contains an additi…

New metrics fail adversarial tests, with some more robust than others.

In this article we revisit the definition of Precision-Recall (PR) curves for generative models proposed by Sajjadi et al. (arXiv:1806.00035). Rather than providing a scalar for generative quality, PR curves distinguish mode-collapse (poor recall) and bad quality (poor precision). We first generalize their formulation …

PR-GNN identifies salient brain regions for ASD biomarkers.

The paper explores CR structures and their leaf spaces in semi-Riemannian manifolds.

Correction for Error estimates for binomial approximations of game options [math.PR/0607123]

Unified four trade-off curves for assessing generative model proximity.

New method detects inconsistencies in AHP matrices using triadic preference reversals.

The paper tackles performative risk optimization under weak convexity assumptions.

New PSDMF algorithms derived from PR and ARM methods.

Develops new approach to recover CR structures from their Levi foliations.

We describe an effective method for simultaneously computing of $d$-invariants of infinite families of Brieskorn spheres $Σ(p,q,r)$ with $pq+pr-qr=1$.

DCC separates marginal estimation from dependence modeling for improved classification accuracy.

Optimal spectral initializers impact phase retrieval phase transitions.

Study symplectic embeddings of 4-manifolds using Lefschetz fibrations.

We propose a general technique for improving alternating optimization (AO) of nonconvex functions. Starting from the solution given by AO, we conduct another sequence of searches over subspaces that are both meaningful to the optimization problem at hand and different from those used by AO. To demonstrate the utility o…

We study two-layer belief networks of binary random variables in which the conditional probabilities Pr[childlparents] depend monotonically on weighted sums of the parents. In large networks where exact probabilistic inference is intractable, we show how to compute upper and lower bounds on many probabilities of intere…

New method detects global factors near BBP phase transition in high-dimensional data.

Assessing the performance of a learned model is a crucial part of machine learning. However, in some domains only positive and unlabeled examples are available, which prohibits the use of most standard evaluation metrics. We propose an approach to estimate any metric based on contingency tables, including ROC and PR cu…

We construct new knot polynomials. Let $V$ be the standard solid torus in 3-space and let $pr$ be its standard projection onto an annulus. Let $M$ be the space of all smooth oriented knots in $V$ such that the restriction of $pr$ is an immersion (e.g. regular diagrams of a classical knot in the complement of its meridi…

Study evaluates machine learning methods for large-scale network reliability, revealing ANN's and PR's performance.

This work improves understanding of projection robust optimal transport distances.

The paper analyzes the performance of constant step-size stochastic approximation algorithms.

Persistence diagrams from random matrices follow RMT universality, offering a new spectral diagnostic.

Paper introduces SCI to distinguish market signals from coordination.

Estimates roughness of financial volatility paths using horizontal visibility graphs.

Noisy PN learning is the problem of binary classification when training examples may be mislabeled (flipped) uniformly with noise rate rho1 for positive examples and rho0 for negative examples. We propose Rank Pruning (RP) to solve noisy PN learning and the open problem of estimating the noise rates, i.e. the fraction …

A new method avoids saddle points in Newton's method.

Paper analyzes LSA algorithm bias and error bounds with RR extrapolation.

A new method for nonparametric regression using mesh-based solutions.

Recently, deep learning approaches with various network architectures have achieved significant performance improvement over existing iterative reconstruction methods in various imaging problems. However, it is still unclear why these deep learning architectures work for specific inverse problems. To address these issu…

This paper develops an ensemble learning-based linearization approach for power flow, which differs from the network-parameter based direct current (DC) power flow or other extended versions of linearization. As a novel data-driven linearization through data mining, it firstly applies the polynomial regression (PR) as …