arXiv research

Local probabilistic models simplify Bayesian classification for complex data.

We show how to control the generalization error of time series models wherein past values of the outcome are used to predict future values. The results are based on a generalization of standard i.i.d. concentration inequalities to dependent data without the mixing assumptions common in the time series setting. Our proo…

New algorithm tracks subspaces with missing and corrupted data, simpler and federated.

New criteria distinguish cause from effect in data, overcoming statistical limitations.

Paper proposes a new method to minimize submodular functions with fewer calls to simpler oracles.

Faster mean estimation with sub-Gaussian error bounds.

We extend the concept of genuine rigidity of submanifolds by allowing mild singularities, mainly to obtain new global rigidity results and unify the known ones. As one of the consequences, we simultaneously extend and unify Sacksteder and Dajczer-Gromoll theorems by showing that any compact $n$-dimensional submanifold …

Paper calculates Morse index of Y-singular minimal surfaces.

A simpler proof for apex graphs in McCarty and Thomas' conjecture.

Simpler algorithm learns shallow networks faster.

Simpler proof for Kielak's virtual fibering criterion.

The paper shows how to simplify complex optimization problems into simpler ones.

A new method solves complex constrained minimax problems.

We prove that, for every closed (not necessarily convex) hypersurface $Σ$ in $\mathbb{R}^{n+1}$ and every $p>n$, the $L^p$-norm of the trace-free part of the anisotropic second fundamental form controls from above the $W^{2,p}$-closeness of $Σ$ to the Wulff shape. In the isotropic setting, we provide a simpler proof. T…

New algorithm outperforms existing ones by focusing on mastering rate.

New framework improves restless bandit policies for large numbers of arms.

We find Weitzenböck formula for the Fueter-Dirac operator which controls the infinitesimal deformations of an associative submanifold in a $7$--manifold with a $G_2$--structure. We establish a vanishing theorem to conclude rigidity under some positivity assumptions on curvature, which are particularly mild in the nearl…

Test verifies if data meets SCAR assumption for PU learning.

Faster algorithm reduces contextual bandit regret with fewer offline regression calls.

We prove that every one-dimensional real Ambrosio-Kirchheim normal current in a Polish (i.e. complete separable metric) space can be naturally represented as an integral of simpler currents associated to Lipschitz curves. As a consequence a representation of every such current with zero boundary (i.e. a cycle) as an in…

This work closes the theory-practice gap for distributed optimization methods by introducing a new regularity condition.

We define a new Hurwitz problem which is essentially a small core of the simple Hurwitz problem. The corresponding Hurwitz numbers have simpler formulae, satisfy effective recursion relations and determine the simple Hurwitz numbers. We also apply this idea of finding a smaller simpler enumerative problem to orbifold H…

Simpler proof for quaternionic Calabi conjecture

We provide an alternative, simpler proof of the existence of thick triangulations for noncompact $\mathcal{C}^1$ manifolds. Moreover, this proof is simpler than the original one given in \cite{pe}, since it mainly uses tools of elementary differential topology. The role played by curvatures in this construction is also…

Langevin diffusion is a commonly used tool for sampling from a given distribution. In this work, we establish that when the target density $p^*$ is such that $\log p^*$ is $L$ smooth and $m$ strongly convex, discrete Langevin diffusion produces a distribution $p$ with $KL(p||p^*)\leq ε$ in $\tilde{O}(\frac{d}ε)$ steps,…

PAC-Bayesian learning bounds are of the utmost interest to the learning community. Their role is to connect the generalization ability of an aggregation distribution $ρ$ to its empirical risk and to its Kullback-Leibler divergence with respect to some prior distribution $π$. Unfortunately, most of the available bounds …

Reduces constructing multiplicative connections to simpler tasks.

SGD tends to favor simpler subnetworks, improving generalization.

Rejection Sampling is a fundamental Monte-Carlo method. It is used to sample from distributions admitting a probability density function which can be evaluated exactly at any given point, albeit at a high computational cost. However, without proper tuning, this technique implies a high rejection rate. Several methods h…

Simplified construction of p-adic transformation group action.

Neural networks learn more efficiently with hidden factorial structures.

A common assumption in causal modeling posits that the data is generated by a set of independent mechanisms, and algorithms should aim to recover this structure. Standard unsupervised learning, however, is often concerned with training a single model to capture the overall distribution or aspects thereof. Inspired by c…

A very simple $\mathbb R^3$ realization of the Möbius strip, significantly simpler than the common one, is given. For any, however large width/length ratio of the strip, it is shown that this realization, in contrast with the common one, is the union of a vertical segment and the graph of a simple rational function on …

Simpler proof for non-basic sets in 2D.

Simplified proof for Frank and Lieb's inequality on Heisenberg group.

Defines basic sections of LA-groupoids for simpler modeling.

New Gamma-Poisson model improves topic selection for short text.

Noise increases the Rashomon ratio, leading simpler models to perform similarly to complex ones.

Transformers prefer simpler explanations in hierarchical tasks.

Controller-Augmented Hidden Markov Models (CHMMs) are a framework for constrained sequential inference.

We introduce and analyze stochastic optimization methods where the input to each gradient update is perturbed by bounded noise. We show that this framework forms the basis of a unified approach to analyze asynchronous implementations of stochastic optimization algorithms.In this framework, asynchronous stochastic optim…

Tree ensembles, such as random forest and boosted trees, are renowned for their high prediction performance, whereas their interpretability is critically limited. In this paper, we propose a post processing method that improves the model interpretability of tree ensembles. After learning a complex tree ensembles in a s…

EQO uses a simple bonus term for efficient exploration in tabular RL.

Let $X$ be an asymptotically hyperbolic manifold and $M$ its conformal infinity. This paper is devoted to deduce several existence results of the fractional Yamabe problem on $M$ under various geometric assumptions on $X$ and $M$: Firstly, we handle when the boundary $M$ has a point at which the mean curvature is negat…

We address challenges in estimating parameters from adaptively collected data.

Develops a robust GMM estimator for outlier-tolerant inference.

EF21 improves convergence in distributed machine learning models.

In this article, we will prove that the subsectors of $α$-induced sectors for $M \rtimes \hat{G} \supset M$ forms a modular category, where $M \rtimes \hat{G}$ is the crossed product of $M$ by the group dual $\hat{G}$ of a finite group $G$. In fact, we will prove that it is equivalent to Müger's crossed product. By usi…