arXiv research

Study subjective perception of low light restored images and develop an unsupervised QA model.

Recurrent CNNs improve image classification in low light conditions.

Physen-Noise2Noise tackles defocus deblurring in low-light conditions with physics-guided self-supervised learning.

A reconstruction theorem in terms of the topology and geometrical structures on the spaces of light rays and skies of a given space-time is discussed. This result can be seen as part of Penrose and Low's programme intending to describe the causal structure of a space-time $M$ in terms of the topological and geometrical…

Let $(X^{m+1}, g)$ be a globally hyperbolic spacetime with Cauchy surface diffeomorphic to an open subset of $\mathbb R^m$. The Legendrian Low conjecture formulated by Natário and Tod says that two events $x,y\inß$ are causally related if and only if the Legendrian link of spheres $\mathfrak S_x, \mathfrak S_y$ whose p…

Adds a precortical module to CNNs for improved robustness to light variations.

Noise is an inherent issue of low-light image capture, one which is exacerbated on mobile devices due to their narrow apertures and small sensors. One strategy for mitigating noise in a low-light situation is to increase the shutter time of the camera, thus allowing each photosite to integrate more light and decrease n…

We study the low-regularity (in-)extendibility of spacetimes within the synthetic-geometric framework of Lorentzian length spaces developed in [KS:17]. To this end, we introduce appropriate notions of geodesics and timelike geodesic completeness and prove a general inextendibility result. Our results shed new light on …

We propose an automatic method to infer high dynamic range illumination from a single, limited field-of-view, low dynamic range photograph of an indoor scene. In contrast to previous work that relies on specialized image capture, user input, and/or simple scene models, we train an end-to-end deep neural network that di…

Proves conditions for Cauchy horizons in low-regularity spacetimes.

We formulate the generalization of the Legendrian Low conjecture of Natario and Tod (proved by Nemirovski and myself before) to the case of causally simple spacetimes. We prove a weakened version of the corresponding statement. In all known examples, a causally simple spacetime $(X, g)$ can be conformally embedded as a…

The moduli space of Riemann surfaces with at least two punctures can be decomposed into a cell complex by using a particular family of ribbon graphs called Nakamura graphs. We distinguish the moduli space with all punctures labelled from that with a single labelled puncture. In both cases, we describe a cell decomposit…

Enhances 2D face recognition with 3D features using active illumination.

ROTS improves sentence similarity by incorporating structural information.

This paper investigates the theoretical foundations of metric learning, focused on three key questions that are not fully addressed in prior work: 1) we consider learning general low-dimensional (low-rank) metrics as well as sparse metrics; 2) we develop upper and lower (minimax)bounds on the generalization error; 3) w…

New methods rank players using covariates and comparisons, outperforming existing algorithms.

In this paper, we propose a general framework for sparse and low-rank tensor estimation from cubic sketchings. A two-stage non-convex implementation is developed based on sparse tensor decomposition and thresholded gradient descent, which ensures exact recovery in the noiseless case and stable recovery in the noisy cas…

New model enhances SPIM for solving low-rank combinatorial optimization and statistical learning problems.

Novel approach embeds loss tunnels in neural networks, revealing insights into their structure.

A new method for active learning works well across all label budgets.

Two methods factor out prior knowledge from low-dimensional embeddings.

This paper proposes a fast and accurate method for sparse regression in the presence of missing data. The underlying statistical model encapsulates the low-dimensional structure of the incomplete data matrix and the sparsity of the regression coefficients, and the proposed algorithm jointly learns the low-dimensional s…

FlashIV solves Black-Scholes implied volatility efficiently and accurately.

Enhances nighttime vehicle detection using style transfer and augmentation.

We simplify SSL by approximating redundant structural components with low-rank factorization.

Robust tensor ring completion improves tensor recovery accuracy and efficiency.

Improved convergence for overparameterized low-rank matrix sensing.

We investigate the initial value problem for the Einstein-Euler equations of general relativity under the assumption of Gowdy symmetry on T3, and we construct matter spacetimes with low regularity. These spacetimes admit, both, impulsive gravitational waves in the metric (for instance, Dirac mass curvature singularitie…

Proposes a new model for image restoration combining deep learning and total variation.

Using open source data, we observe the fascinating dynamics of nighttime light. Following a global economic regime shift, the planetary center of light can be seen moving eastwards at a pace of about 60 km per year. Introducing spatial light Gini coefficients, we find a universal pattern of human settlements across dif…

Deep learning and prior maps improve traffic light recognition for autonomous cars.

This paper proposes and implements an intuitive and pervasive solution for neonatal EEG monitoring assisted by sonification and deep learning AI that provides information about neonatal brain health to all neonatal healthcare professionals, particularly those without EEG interpretation expertise. The system aims to inc…

The problem of Poisson denoising appears in various imaging applications, such as low-light photography, medical imaging and microscopy. In cases of high SNR, several transformations exist so as to convert the Poisson noise into an additive i.i.d. Gaussian noise, for which many effective algorithms are available. Howev…

RES-PCA efficiently recovers low-rank matrices without precise rank knowledge.

Deep neural networks often fit low-frequency functions, contrary to conventional numerical schemes.

Classifies surfaces with zero mean curvature in a light cone.

Solves surface problem in 3D light cone.

Study light ray transform on Lorentzian manifolds without conjugate points.

Paper extends previous result on hypersurfaces with degenerate light-like points.

New algorithm reduces RL complexity with low switching costs.

The natural topological, differentiable and geometrical structures on the space of light rays of a given spacetime are discussed. The relation between the causality properties of the original spacetime and the natural structures on the space of light rays are stressed. Finally, a symplectic geometrical approach to the …

Study light ray transform in pseudo-Euclidean space, derive inversion formula, and prove stability.

Consider a constant mean curvature immersion $F:U(\subset \boldsymbol{R}^n)\to M$ into an arbitrary Lorentzian $(n+1)$-manifold $M$. A point $o\in U$ is called a light-like point if the first fundamental form $ds^2$ of $F$ degenerates at $o$. We denote by $B_F$ the determinant function of the symmetric matrix associate…

Develops efficient method for updating models with small data changes.

The paper explores the injectivity of the light ray transform on Lorentzian manifolds.

In this paper, we are concerned with light-like extremal surfaces in curved spacetimes. It is interesting to find that under a diffeomorphic transformation of variables, the light-like extremal surfaces can be described by a system of nonlinear geodesic equations. Particularly, we investigate the light-like extremal su…

Designing models that are robust to small adversarial perturbations of their inputs has proven remarkably difficult. In this work we show that the reverse problem---making models more vulnerable---is surprisingly easy. After presenting some proofs of concept on MNIST, we introduce a generic tilting attack that injects …

In this paper we prove several multiplicity results of $t$-periodic light rays in conformally stationary spacetimes using the Fermat metric and the extensions of the classical theorems of Gromoll-Meyer and Bangert-Hingston to Finsler manifolds. Moreover, we exhibit some stationary spacetimes with a finite number of $t$…