arXiv research

Serverless cloud computing speeds up double machine learning model estimation.

Characterizes a specific type of Courant algebroid with a Calabi-Yau structure.

Recently, Rips produced an example of a double of two free groups which has unsolvable generalized word problem. In this paper, we show that Rips's example fits into a large class of doubles of groups, each member of which contains F_2 x F_2 and therefore has unsolvable generalized word problem and is incoherent.

Deep learning models can generalize well even when they fit training data perfectly.

We introduce a stochastic model to explain a double power-law distribution which exhibits two different Paretian behaviors in the upper and the lower tail and widely exists in social and economic systems. The model incorporates fitness consideration and noise fluctuation. We find that if the number of variables (e.g. t…

The study of quotient structures in multi-graded bundles, including double vector bundles.

Improved estimators for causal inference using cross-fitting and undersmoothing.

We study the linear subspace fitting problem in the overparameterized setting, where the estimated subspace can perfectly interpolate the training examples. Our scope includes the least-squares solutions to subspace fitting tasks with varying levels of supervision in the training data (i.e., the proportion of input-out…

Paper combines machine learning and model averaging for robust parameter estimation.

Double Lie algebroids were discovered by Kirill Mackenzie from the study of double Lie groupoids and were defined in terms of rather complicated conditions making use of duality theory for Lie algebroids and double vector bundles. In this paper we establish a simple alternative characterization of double Lie algebroids…

Develops a test for conditional local independence of counting processes.

Chernozhukov, Chetverikov, Demirer, Duflo, Hansen, and Newey (2016) provide a generic double/de-biased machine learning (DML) approach for obtaining valid inferential statements about focal parameters, using Neyman-orthogonal scores and cross-fitting, in settings where nuisance parameters are estimated using a new gene…

Study shows double descent curve in high-dimensional linear regression with random projections.

Paper proposes FPG algorithm for unbiased off-policy PG estimation.

Breakthroughs in machine learning are rapidly changing science and society, yet our fundamental understanding of this technology has lagged far behind. Indeed, one of the central tenets of the field, the bias-variance trade-off, appears to be at odds with the observed behavior of methods used in the modern machine lear…

Non-negative matrix factorization (NMF) is a technique for finding latent representations of data. The method has been applied to corpora to construct topic models. However, NMF has likelihood assumptions which are often violated by real document corpora. We present a double parametric bootstrap test for evaluating the…

We price weather-contingent options by use of Monte Carlo simulations. After calibrating the models to fit quoted prices, we analyze bid-ask spreads in terms of correlations across markets. Results are presented for a double-trigger Weather vs. Natural Gas call option.

High-dimensional models can outperform simpler ones in causal inference.

Deep networks can interpolate noisy data without losing generalization.

New insights into double descent phenomenon in neural networks.

Regression models can interpolate noisy data and still perform well, contrary to the bias-variance tradeoff.

Research provides explicit NPV expressions for double barrier strategies.

We give a concise summary of the para-Hermitian geometry that describes a doubled target space fit for a covariant description of T-duality in string theory. This provides a generalized differentiable structure on the doubled space and leads to a kinematical setup which allows for the recovery of the physical spacetime…

Improves treatment effect estimates using coordinated deep learning.

Paper adapts DML for panel data, addressing unobserved heterogeneity.

Deep learning models can overfit noisy data without losing generalization.

SHIFT improves robustness in estimating dose-response functions with heavy-tailed contamination.

The paper develops methods to estimate treatment effects in sample selection models.

New method bypasses global fit for LISA's Galactic binaries, extracting population parameters directly.

Overparameterized models generalize well despite fitting noisy data.

Developed scalable ABM for complex financial markets.

The paper uses Gaussian mixture models for Bayesian networks and proposes an optimization algorithm.

Two new algorithms improve Q* approximation in batch RL with linear error propagation.

New geometric interpretation explains over-parameterized models and adversarial perturbations.

The paper explores how complex models can improve system identification beyond traditional limits.

The model outperforms other models in option pricing, especially for short-term implied volatility.

A new method improves estimation of COVID-19 vaccine effectiveness.

Improves DRL for long-term causal inference with semiparametric methods.

Deep networks generalize well even when they fit training data perfectly, thanks to overparametrization.

Efficient exploration improves large language model performance with fewer queries.

New method identifies uncertainty shocks in financial markets using revised VIX.

PipeDream-2BW accelerates large model training by 20x with minimal memory usage.

Study on determining metrics from minimal surface areas, extending earlier work.

Proposes a new method using GANs for testing conditional independence.

The study analyzes a model for aggregate losses with dependent and overdispersed inter-losses times.

A new decomposition explains over-parameterized models' counterintuitive behaviors.

Model shows loss curve with two distinct exponents due to sparse activations.

New model tackles PU data with better accuracy.