Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

169,341 papers · 148 categories

Trend · papers per month

2805618411,121 · Jun 202019922001200920182026
48 results for small generators

Generative Latent Implicit Conditional Optimization (GLICO) learns from small samples.

problem Learning from small labeled datasets.
method Generative Latent Implicit Conditional Optimization (GLICO) learns a latent space and generator from small labeled data.
result GLICO synthesizes new samples for every class using as few as 10 examples per class.

Connectedness of small clusters in Riemannian and Finsler manifolds proven.

problem Understanding connectedness of small clusters in Riemannian and Finsler manifolds.
method Proved connectedness and small diameter properties for clusters of small volume in both manifolds.
result Clusters in Riemannian manifolds are connected and have small diameter; in Finsler manifolds, they are at most m connected components of small diameter.

The paper studies spheres with small diameter in 3D manifolds concentrating at scalar curvature critical points.

problem Understanding the behavior of Willmore spheres with small diameter in 3D manifolds.
method Analyzes spheres under bounded Willmore energy and small diameter constraints, focusing on scalar curvature critical points.
result Embedded Willmore spheres concentrate at critical points of scalar curvature under small diameter and bounded energy conditions.

Enhances classification performance with small, additive perturbations.

problem Improving classification performance using small, additive perturbations.
method Proposes a perturbation generation network (PGN) based on adversarial learning to enhance classifier performance.
result Demonstrates that PGN can enhance overall classification performance without altering the target classifier network.

Large initial learning rate helps neural nets generalize better.

problem Understanding why large initial learning rates lead to better neural net generalization.
method Developed a proof for a two-layer network and demonstrated with experiments on CIFAR-10.
result Proved that a two-layer network trained with a large initial learning rate and annealing generalizes better than one trained with a small learning rate.

Deep residual networks trained with gradient descent have small generalization gap.

problem Limited theoretical understanding of why residual networks generalize well.
method Analyzing overparameterized deep residual networks trained by gradient descent.
result Demonstrates that residual networks have a small generalization gap between training and test error.

Generative model initializes 2-layer network weights for small datasets.

problem Approximating functions with 2-layer networks using small datasets and gradient-based training.
method Initialize hidden weights with a learned proposal distribution parameterized as a deep generative model. Refine with gradient-based post-processing and regularization.
result Demonstrates effectiveness of the approach with numerical examples.

Small sub-Riemannian balls have diameter close to twice their radius.

problem Understanding the diameter of small sub-Riemannian balls.
method Analyzing C1,1C^{1,1} and C0C^0 sub-Riemannian manifolds.
result The diameter of small sub-Riemannian balls equals twice the radius in C1,1C^{1,1} manifolds, and is close to twice the radius in C0C^0 manifolds.

Study evaluates synthetic data augmentation for small datasets, highlighting inconsistencies in traditional metrics.

problem Inconsistent validation of synthetic data generated for small sample sizes.
method Proposes a normalized Bottleneck distance metric to evaluate synthetic tabular data.
result Common metrics like propensity scoring and MMD fail for small datasets, showing instability and high variability.

In the present paper we find a bijection between the set of small covers over an nn-cube and the set of acyclic digraphs with nn labeled nodes. Using this, we give a formula of the number of small covers over an nn-cube (generally, a product of simplices) up to Davis-Januszkiewicz equivalence classes and $\mathbf{Z}…

2008-02-14abs ↗pdf ↗

In these notes we give a shortened and more direct proof of Goto's generalized Kaehler stability theorem stating that if (J_1,J_2) is a generalized kaehler structure for which J_2 is determined by a nowhere vanishing closed form, then small deformations of J_1 can be coupled with small deformations of J_2 so that the p…

2012-06-10abs ↗pdf ↗

We examine geometric properties of a knot J that are unchanged by taking a (p,q)-cable K of J. Specifically, we relate w(K) to w(J), where w(K) is the width of K in the sense of Gabai. We use this information to demonstrate that thin position is a minimal bridge position of J if and only if the same is true for K, and …

2010-10-15abs ↗pdf ↗

This paper uses synthetic data to improve machine learning performance on small, imbalanced datasets.

problem Improving machine learning performance on small and imbalanced datasets.
method Generates synthetic data through convex combination and uses it in a semi-supervised learning framework with support vector machines.
result Synthetic data over-sampling supports the cluster assumption in semi-supervised learning, leading to outstanding results for small high-dimensional datasets and imbalanced learning problems.

A small cover was introduced by Davis and Januszkiewicz as an nn-dimensional closed manifold with a locally standard Z2)nZ_2)^n-action such that its orbit space is a simple convex polytope. There exist a one-to-one correspondence between small covers and (Z2)n(Z_2)^n-colored polytopes. In this paper we study a construction…

2011-04-10abs ↗pdf ↗

New algorithm achieves small-loss bounds in online learning with improved rates.

problem Achieving strong stability in online learning algorithms.
method Introduces ρρ-separation to enforce strong stability, unifying previous approaches.
result Oracle-efficient algorithm achieves small-loss bounds with improved rates.

New bounds on diameters and generators for specific lattices and graphs.

problem Finding bounds on diameters and generators for arithmetic lattices and Ramanujan graphs.
method Analyzing arithmetic lattices from Eichler orders in quaternion algebras, applying techniques to definite quaternion algebras.
result Bounds on diameters and generators for arithmetic lattices and Ramanujan graphs.

The paper investigates why GNNs struggle to generalize from small to large graphs.

problem Challenges in graph neural networks' ability to generalize across different graph sizes.
method Identified and studied the effect of local structure on size generalization; proposed a novel SSL task.
result GNNs can converge to non-generalizing solutions when there is a discrepancy in local structure.

Improves probability estimates for small datasets in multi-class problems.

problem Inaccurate probability estimates in classification tasks, especially on small datasets.
method Introduced Data Generation and Grouping algorithm to improve calibration on small datasets, then applied to multi-class problems.
result Calibration error can be decreased using the proposed approach.

Improves generalization in learning problems with small parameter method.

problem Improving generalization in learning problems with high-dimensional nonlinear functions.
method Perturbation theory applied to a weakly-controlled gradient system.
result Approximate optimal solutions for improving generalization with small noise.

Random feature model shows slow self-correction of generalization gap.

problem Slow deterioration of generalization error in random feature model.
method Examined the dynamic behavior of gradient descent in the model's resonance regime.
result Gradient descent exhibits a self-correction mechanism, reducing generalization gap over time.

CoDistill-GRPO improves small models in GRPO by distilling knowledge from a larger model.

problem Small models in GRPO struggle with sparse rewards on difficult tasks.
method Simultaneously trains a large and small model using co-distillation and GRPO objectives.
result Significant improvement in small model performance over standard GRPO on mathematical benchmarks.

An investor with constant absolute risk aversion trades a risky asset with general Itô-dynamics, in the presence of small proportional transaction costs. In this setting, we formally derive a leading-order optimal trading policy and the associated welfare, expressed in terms of the local dynamics of the frictionless op…

2012-09-12abs ↗pdf ↗

Optimizes trading frequencies for multi-asset portfolios with small transaction costs.

problem Investment with multiple assets and small transaction costs.
method Optimizes trading frequencies explicitly for multidimensional diffusion setting, compares to alternatives.
result Explicit formulas for optimal trading frequencies and welfare losses.

The lifespan of Ricci flows is analyzed and generalized to noncompact manifolds.

problem Analyzing the lifespan and transfer rate of Ricci flows on manifolds with small Ricci curvature.
method Generalized lifespan estimate for local Ricci flow, proving short-time existence on noncompact manifolds with small curvature.
result Spatial transfer rate of Ricci flow resembles that of the heat equation under certain conditions.

We generalize the notion of a small sheaf of sets over a topological space or manifold to define the notion of a small stack of groupoids over an étale topological or differentiable stack. We then provide a construction analogous to the étalé space construction in this context, establishing an equivalence of 2-categori…

2010-11-28abs ↗pdf ↗

Method generates prototypes from small datasets for efficient learning.

problem Efficiently learning from small datasets with soft labels.
method Modular method for generating soft-label prototypical lines and Hierarchical Soft-Label Prototype k-Nearest Neighbor algorithm.
result High classification accuracy with significantly fewer prototypes than classes.

Small LLMs outperform large ones on simple tasks without extra labelling costs.

problem Performance of large commercial models in simple classification tasks.
method Logistic Regression on small LLM embeddings.
result Small LLMs equal or outperform large LLMs in 'tens-of-shot' classification tasks.

We prove that for every P there is a bound B depending only on P so that the mapping torus of every P--small irreducible train-track map can be obtained by surgery from one of B mapping tori. We show that given an integer P>0 there is a bound MM depending only on P, so that there exists a presentation of the fundament…

2012-09-25abs ↗pdf ↗