arXiv research

New system tracks musical performances in raw sheet images without preprocessing.

Paper tackles score following in full-page sheet music images.

Worldsheet wraps a 3D mesh sheet onto a single image to synthesize novel views.

Reduces data leakage in distributed deep learning models.

We introduce Embed to Control (E2C), a method for model learning and control of non-linear dynamical systems from raw pixel images. E2C consists of a deep generative model, belonging to the family of variational autoencoders, that learns to generate image trajectories from a latent space in which the dynamics is constr…

This paper explores deep learning for improving X-ray CT image reconstruction from undersampled data.

A world sheet in anti-de Sitter space is a timelike submanifold consisting of a one-parameter family of spacelike submanifolds. We consider the family of lightlike hypersurfaces along spacelike submanifolds in the world sheet. The locus of the singularities of lightlike hypersurfaces along spacelike submanifolds forms …

Generative Neuro-Symbolic model learns from raw data with rich conceptual representations.

In the Minkowski space-time, a world hyper-sheet is a timelike hypersurface consisting of a one-parameter family of spacelike submanifolds. Recently, Bousso and Randall introduced the notion of caustics of world hyper-sheets in order to define the notion of holographic domains in space-time. Here, we give a mathematica…

We consider a disk-shaped thin elastic sheet bonded to a compliant sphere. (Our sheet can slip along the sphere; the bonding controls only its normal displacement.) If the bonding is stiff (but not too stiff), the geometry of the sphere makes the sheet wrinkle to avoid azimuthal compression. The total energy of this sy…

A world sheet in Lorentz-Minkowski space is a timelike submanifold consisting of a one-parameter family of spacelike submanifolds in Lorentz-Minkowski space. In this paper we investigate differential geometry of world sheets in Lorentz-Minkowski space as an application of the theory of big wave fronts.

Model-free reinforcement learning has recently been shown to successfully learn navigation policies from raw sensor data. In this work, we address the problem of learning driving policies for an autonomous agent in a high-fidelity simulator. Building upon recent research that applies deep reinforcement learning to navi…

Computer aided diagnostic (CAD) system is crucial for modern med-ical imaging. But almost all CAD systems operate on reconstructed images, which were optimized for radiologists. Computer vision can capture features that is subtle to human observers, so it is desirable to design a CAD system op-erating on the raw data. …

Improved RL for grasping in cluttered scenes using state representation learning.

For an autonomous agent to fulfill a wide range of user-specified goals at test time, it must be able to learn broadly applicable and general-purpose skill repertoires. Furthermore, to provide the requisite level of generality, these skills must handle raw sensory input such as images. In this paper, we propose an algo…

Study compares empirical systemic risk with balance sheet risk in interbank networks.

A framework visualizes embedding spaces of neural survival analysis models using anchor directions.

Automated generation of medical reports from chest x-rays using expert annotations.

Hempel has shown that the fundamental groups of knot complements are residually finite. This implies that every nontrivial knot must have a finite-sheeted, noncyclic cover. We give an explicit bound, $Φ(c)$, such that if $K$ is a nontrivial knot in the three-sphere with a diagram with $c$ crossings and a particularly s…

In the United States, heart disease is the leading cause of death for both men and women, accounting for 610,000 deaths each year [1]. Physicians use Magnetic Resonance Imaging (MRI) scans to take images of the heart in order to non-invasively estimate its structural and functional parameters for cardiovascular diagnos…

A new XVA strategy rooted in balance sheet perspective improves equity process for bank shareholders.

Deep learning improves 3D microscopy resolution without matched target images.

Our goal is to extract meaningful transformations from raw images, such as varying the thickness of lines in handwriting or the lighting in a portrait. We propose an unsupervised approach to learn such transformations by attempting to reconstruct an image from a linear combination of transformations of its nearest neig…

The edge of torn elastic sheets and growing leaves often form a hierarchical buckling pattern. Within non-Euclidean plate theory this complex morphology can be understood as low bending energy isometric immersions of hyperbolic Riemannian metrics. With this motivation we study the isometric immersion problem in strip a…

Image post-processing is used in clinical-grade ultrasound scanners to improve image quality (e.g., reduce speckle noise and enhance contrast). These post-processing techniques vary across manufacturers and are generally kept proprietary, which presents a challenge for researchers looking to match current clinical-grad…

Weakly stable constant mean curvature (CMC) hypersurfaces are stable critical points of the area functional with respect to volume preserving deformations. We establish a pointwise curvature estimate (in the non-singular dimensions) and a sheeting theorem (in all dimensions) for weakly stable CMC hypersurfaces, giving …

In this fact sheet we give some preliminary research results on the Bayesian Decision Theory. This theory has been under construction for the past two years. But what started as an intuitive enough idea, now seems to have the makings of something more fundamental.

Accelerating Magnetic Resonance Imaging (MRI) by taking fewer measurements has the potential to reduce medical costs, minimize stress to patients and make MRI possible in applications where it is currently prohibitively slow or expensive. We introduce the fastMRI dataset, a large-scale collection of both raw MR measure…

Model shows how banks' hidden-to-maturity accounting can mask run risk and lead to financial instability.

Generates music with coherent rhythm, chords, and melody using LSTM models.

C2G-Net improves image classification of similar objects like cells.

Deep reinforcement learning (deep RL) has achieved superior performance in complex sequential tasks by using deep neural networks as function approximators to learn directly from raw input images. However, learning directly from raw images is data inefficient. The agent must learn feature representation of complex stat…

This paper shows how to create surface-links with many triple points.

We investigate compactness phenomena involving free boundary minimal hypersurfaces in Riemannian manifolds of dimension less than eight. We provide natural geometric conditions that ensure strong one-sheeted graphical subsequential convergence, discuss the limit behaviour when multi-sheeted convergence happens and deri…

In this work, we propose a simple yet effective solution to the problem of connectome inference in calcium imaging data. The proposed algorithm consists of two steps. First, processing the raw signals to detect neural peak activities. Second, inferring the degree of association between neurons from partial correlation …

Novel defects in hyperbolic sheets explain complex wrinkling patterns in nature.

New RL framework learns task completion without prior knowledge.

This paper shows how to construct anomaly free world sheet actions in string theory with $D$-branes. Our method is to use Deligne cohomology and bundle gerbe theory to define geometric objects which are naturally associated to $D$-branes and connections on them. The holonomy of these connections can be used to cancel g…

Graph Attention Networks improve image classification with superpixels.

A number of recent approaches to policy learning in 2D game domains have been successful going directly from raw input images to actions. However when employed in complex 3D environments, they typically suffer from challenges related to partial observability, combinatorial exploration spaces, path planning, and a scarc…

The paper studies quandles over a hyperboloid and computes a knot invariant.

The paper classifies ovals in 4D space for a specific flow.

Improved satellite image captions enhance descriptiveness without large models.

We prove that if $M \longrightarrow P$ is a small cover of a compact right-angled hyperbolic polyhedron $P$ then $M$ admits a cofinal tower of finite sheeted covers with positive rank gradient. As a corollary, if $π_1(M)$ is commensurable with the reflection group of $P$, then $M$ admits a cofinal tower of finite sheet…

New deep learning method improves financial stress testing accuracy.

In this work, we have proposed several enhancements to improve the performance of any facial emotion recognition (FER) system. We believe that the changes in the positions of the fiducial points and the intensities capture the crucial information regarding the emotion of a face image. We propose the use of the gradient…

Independent component analysis (ICA), as an approach to the blind source-separation (BSS) problem, has become the de-facto standard in many medical imaging settings. Despite successes and a large ongoing research effort, the limitation of ICA to square linear transformations have not been overcome, so that general INFO…

A combinatorial presentation of closed orientable 3-manifolds as bi-tricolored links is given together with two versions of a calculus via moves to manipulate bi-tricolored links without changing the represented manifold. That is, we provide a finite set of moves sufficient to relate any two manifestations of the same …