arXiv research

A new method for sparse linear bandits reduces exploration-exploitation tradeoff.

SPARKLE handles high-dimensional covariates for online decision-making.

Paper proposes Sinkformers for Transformers with doubly stochastic attention.

The paper analyzes how repeating epochs affects data scaling in linear regression.

New knot homology invariant grows exponentially with color.

InstanceFlow visualizes classifier confusion over training epochs.

New estimator improves ATT estimation efficiency with external controls.

Adam optimizer's bias is influenced by mini-batch size and momentum hyperparameters.

Recent neural network and language models rely on softmax distributions with an extremely large number of categories. Since calculating the softmax normalizing constant in this context is prohibitively expensive, there is a growing literature of efficiently computable but biased estimates of the softmax. In this paper …

Study optimal policy regret in partially observable Markov games with adaptive opponents.

Few-shot learning aims to train efficient predictive models with a few examples. The lack of training data leads to poor models that perform high-variance or low-confidence predictions. In this paper, we propose to meta-learn the ensemble of epoch-wise empirical Bayes models (E3BM) to achieve robust predictions. "Epoch…

Nowadays, deep learning can be employed to a wide ranges of fields including medicine, engineering, etc. In deep learning, Convolutional Neural Network (CNN) is extensively used in the pattern and sequence recognition, video analysis, natural language processing, spam detection, topic categorization, regression analysi…

New method shows hidden state can significantly improve differential privacy in SGD.

New algorithm tackles subgroup fairness in AI with multiple sensitive attributes.

We prove that the cardinality of the torsion subgroups in homology of a closed hyperbolic manifold of any dimension can be bounded by a doubly exponential function of its diameter. It would follow from a conjecture by Bergeron and Venkatesh that the order of growth in our bound is sharp. We also determine how the numbe…

A deep learning model, named IITNet, is proposed to learn intra- and inter-epoch temporal contexts from raw single-channel EEG for automatic sleep scoring. To classify the sleep stage from half-minute EEG, called an epoch, sleep experts investigate sleep-related events and consider the transition rules between the foun…

Epoch-GDA achieves optimal convergence rate for SCSC min-max problems.

This paper tackles overfitting in CTR models by introducing Multi-Epoch learning with Data Augmentation.

The paper examines Einstein doubly warped product manifolds with a semi-symmetric metric connection.

Characterizes spacetimes using doubly torqued vectors.

This paper extends ResNet theory to infinitely deep networks, linking them to diffusion processes.

In unsupervised learning, collecting more data is not always a costly process unlike the training. For example, it is not hard to enlarge the 40GB WebText used for training GPT-2 by modifying its sampling methodology considering how many webpages there are in the Internet. On the other hand, given that training on this…

New invariant measures doubly slice links, disproving previous bounds.

Characterizes a specific type of spacetime using vector fields.

Characterizes and examines gradient solitons on doubly warped product manifolds.

The article enumerates doubly symmetric diagrams for knots up to 18 crossings.

Proposes DR-ACI for causal effect intervals with temporal dependence.

This paper explains double descent in linear neural networks, identifying new factors.

Simplified tutorial on doubly robust learning for causal inference.

We show that if the connected sum of two knots with coprime Alexander polynomials is doubly slice, then the Ozsváth-Szabó correction terms as smooth double sliceness obstructions vanish for both knots. Recently, Jeffrey Meier gave smoothly slice knots that are topologically doubly slice, but not smoothly doubly slice. …

In this article, we present a complete study of two disjoint classes of conformal vector fields on doubly warped product manifolds as well as on doubly warped space-times. Then we study Ricci solitons on doubly warped product manifollds admitting these types of conformal vector fields.

Identifies doubly slice genera for 2909 prime knots with up to 12 crossings.

Paper develops a new estimator for dynamic treatment effects in high-dimensional settings.

Wasserstein Policy Learning for Distributional Outcomes

The paper shows some Montesinos links can't be doubly sliced strongly.

We define an obstruction for a knot to be Z[Z]-homology ribbon, and use this to provide restrictions on the integers that can occur as the triple linking numbers of derivative links of knots that are either homotopy ribbon or doubly slice. Our main application finds new non-doubly slice knots. In particular this gives …

Deep learning models can initially degrade in performance as they grow larger, then improve.

Hamiltonian cycles found in toroidal maps.

New lower bound for doubly slice genus using knot signatures.

In this paper we study fundamental geometric properties of doubly warped product immersion which is an extension of warped product immersion. Moreover, we study geometric inequality for doubly warped products isometrically immersed in arbitrary Riemannian manifolds.

One epoch training yields certifiably robust models.

We investigate several conjectures in geometric topology by assembling computer data obtained by studying weaving knots, a doubly infinite family $W(p,n)$ of examples of hyperbolic knots. In particular, we compute some important polynomial knot invariants, as well as knot homologies, for the subclass $W(3,n)$ of this f…

Paper defines new risk measures for elliptical distributions.

Some basic geometric properties of doubly twisted product immersions are established.

We give a formula for Alexander polynomials of doubly primitive knots.

We construct an infinite family of smoothly slice knots that we prove are topologically doubly slice. Using the correction terms coming from Heegaard Floer homology, we show that none of these knots is smoothly doubly slice. We use these knots to show that the subgroup of the double concordance group consisting of smoo…

The twisting number of a ribbon knot is at least as large as its doubly slice genus.

In this paper, we introduce horizontal and vertical warped product Finsler manifold. We prove that every C-reducible or proper Berwaldian doubly warped product Finsler manifold is Riemannian. Then, we find the relation between Riemmanian curvatures of doubly warped product Finsler manifold and its components, and consi…