Finite-width neural networks are approximated by Gaussian processes with finite size corrections.
problem Understanding the behavior of finite-width neural networks as they approach infinite width.
method Analyzing the distribution of outputs at initialization for large, finite neural networks with a single hidden layer.
result The distribution of outputs at initialization is well described by a Gaussian perturbed by the fourth Hermite polynomial, with the perturbation scale inversely proportional to the number of network units.
Estimates neural representation dimensionality from small sample sizes.
problem Estimating neural representation dimensionality from limited data.
method Proposed a bias-corrected estimator for participation ratio of eigenvalues.
result The estimator is more accurate with finite samples and noise.
Non-intrusive algorithm corrects corrupted data in classification models.
problem Classification problems with corrupted training data.
method Post-processes a trained classification model with a correction procedure.
result Corrected models deliver correct classification results even with corrupted data.
In the information-based paradigm of inference, model selection is performed by selecting the candidate model with the best estimated predictive performance. The success of this approach depends on the accuracy of the estimate of the predictive complexity. In the large-sample-size limit of a regular model, the predicti…
A bridge between continuous signals and discrete Ising spins for associative memory.
problem Associative memory in continuous-signal-driven Ising spin systems.
method Multilayer Ising framework with PCA whitening and SimHash projection, coupled to pseudo-inverse memory couplings.
result Finite-size scaling of operational storage capacity with αc(N)=αc(∞)−cN−1/2, approaching αc(∞)≈0.50. We provide finite-sample analysis of a general framework for using k-nearest neighbor statistics to estimate functionals of a nonparametric continuous probability density, including entropies and divergences. Rather than plugging a consistent density estimate (which requires k→∞ as the sample size $n \to \in…
Corrects GCV for inconsistent risk estimation in finite ensembles of penalized estimators.
problem Inconsistent risk estimation of GCV for finite ensembles of penalized estimators.
method Identifies a correction involving an additional scalar correction based on degrees of freedom adjusted training errors from each ensemble component.
result CGCV maintains computational advantages of GCV and is model-free uniformly consistent for ridge regression.
The Fisher information approximation (FIA) is an implementation of the minimum description length principle for model selection. Unlike information criteria such as AIC or BIC, it has the advantage of taking the functional form of a model into account. Unfortunately, FIA can be misleading in finite samples, resulting i…
Optimal δ-correct algorithm for heavy-tailed distributions with minimal samples.
problem Identifying the best arm from a set of unknown distributions with heavy tails.
method Developed a δ-correct algorithm that matches lower bounds under mild restrictions.
result The proposed algorithm achieves optimal sample complexity for heavy-tailed distributions.
Improved solver maintains positivity and accuracy across all time steps.
problem Linear second-order schemes for Fokker-Planck equation cannot preserve positivity.
method Flux-Corrected Diagonal Frog (FCDF) framework using nonlinear extension and iterative limiter.
result FCDF schemes are unconditionally positive across all time steps and maintain second-order accuracy.
We propose a new method of estimation in topic models, that is not a variation on the existing simplex finding algorithms, and that estimates the number of topics K from the observed data. We derive new finite sample minimax lower bounds for the estimation of A, as well as new upper bounds for our proposed estimator. W…
This work proposes using Conformal Prediction to improve OOD detection scores and vice versa.
problem Improper evaluation of OOD detection scores due to finite sample size.
method Defining new conformal AUROC and FRP@TPR95 metrics and using OOD scores as non-conformity scores.
result Improved evaluation metrics and better interpretation of OOD scores.
Analyzes DNNs trained with noisy gradients, finding FWCs negligible for large n.
problem Analyzing DNNs trained with noisy gradients.
method Introduced analytical framework to analyze non-Gaussian stochastic process.
result FWCs negligible for large n, improving CNN performance.
We study the asymptotic properties of the adaptive Lasso in cointegration regressions in the case where all covariates are weakly exogenous. We assume the number of candidate I(1) variables is sub-linear with respect to the sample size (but possibly larger) and the number of candidate I(0) variables is polynomial with …
AMAGOLD improves stochastic gradient MCMC by infrequent Metropolis-Hastings corrections.
problem Bias in stochastic gradient Hamiltonian Monte Carlo (SGHMC).
method AMAGOLD infrequently uses Metropolis-Hastings corrections to remove bias, with a fixed step size schedule.
result AMAGOLD converges to the target distribution with a fixed, rather than a diminishing, step size, and at most a constant factor slower convergence rate.
The paper analyzes rates for a modified gradient descent method using Stein variational gradients.
problem Improving the accuracy of gradient descent methods for complex target distributions.
method Derives finite-particle rates for regularized Stein variational gradient descent (R-SVGD).
result Establishes explicit non-asymptotic bounds for time-averaged empirical measures.
ResNets and DenseNets converge to NTK with depth and width, offering advantages for kernel regression.
problem Understanding convergence of ResNets and DenseNets to Neural Tangent Kernel (NTK).
method Analysis of finite width and depth corrections for NTK of ResNets and DenseNets.
result ResNets and DenseNets can converge to NTK with depth and width, unlike vanilla networks.
New method corrects bias in density ratio estimation for missing data.
problem Missing data bias in density ratio estimation.
method Adapted KLIEP method (M-KLIEP) for MNAR data.
result M-KLIEP restores consistency and minimax optimality.
The paper develops methods for constructing confidence regions for regression functions in binary classification.
problem Building distribution-free confidence regions for regression functions in binary classification.
method Resampling test and empirical risk minimization approach for model classes with finite pseudo-dimensions and inverse Lipschitz parameterizations.
result Strong uniform consistency and exponential probably approximately correct bounds on the L2 sizes of the regions. Algorithm learns stochastic system dynamics from data.
problem Recovering interpretable symbolic expressions for stochastic systems.
method Data-driven, trajectory averaging, drift-informed correction.
result Recover coefficients and densities to within 5% and 0.01 in total variation, respectively.
We study how finite Bayesian neural networks adapt their hidden representations.
problem Understanding how finite Bayesian neural networks differ from infinite ones.
method We analyze the asymptotics of learned feature kernels for various network architectures.
result The leading finite-width corrections to feature kernels have a universal form.
Automates size normalization for fashion items.
problem Reduce merchandise returns in e-commerce.
method Uses sales data to automate size mapping.
result Automated size mappings comparable to human-generated ones.
Various bias-correction methods such as EXTRA, gradient tracking methods, and exact diffusion have been proposed recently to solve distributed {\em deterministic} optimization problems. These methods employ constant step-sizes and converge linearly to the {\em exact} solution under proper conditions. However, their per…
The paper improves A/B testing for non-Gaussian data, ensuring reliable results with large sample sizes.
problem Inaccurate A/B testing results due to non-normal data and unequal sample sizes.
method Derives explicit formulas for minimum sample size and introduces an Edgeworth-based correction.
result Corrected method improves reliability of A/B testing in real-world conditions.
New algorithms verify and search causal graphs with minimal interventions.
problem Recovering causal graphs from interventional data.
method Characterization of minimal intervention sets for verification, and adaptive graph separator algorithm for search.
result First provable algorithms for efficient verification and search of causal graphs.
Improved MALA method for neural networks uncertainty quantification.
problem Uncertainty quantification in Bayesian neural networks.
method Corrected Stochastic MALA (csMALA) with a simplified correction term.
result Improved surrogate posterior for quantifying uncertainties in neural networks.
New TD algorithms stabilize RL tasks by reformulating updates into fixed point equations.
problem TD learning's sensitivity to step size specification.
method Implicit TD algorithms reformulate TD updates into fixed point equations.
result Implicit TD algorithms are more stable and less sensitive to step size.
A neural framework corrects bias in estimating individual treatment effects.
problem Estimating individual treatment effects from observational data.
method An anchored neural architecture and precision-corrected intersection-bound inference.
result Corrected bias and maintained nominal coverage in high-dimensional settings.
SLOE speeds up logistic regression in high dimensions with accurate signal strength estimation.
problem Poor performance of logistic regression in high-dimensional settings.
method SLOE reparameterizes the signal strength for faster and more accurate estimation.
result SLOE provides a fast and accurate method for dimensionality correction in logistic regression.
Ensembles of neural networks improve training dynamics and performance.
problem Improving neural network performance through model size increase.
method Defining collegial ensembles (CE) as multiple independent models trained as a single model, and using theoretical results on NTK to optimize architecture search.
result CE dynamics simplify and scale favorably, resembling wide models, and can be efficiently implemented using group convolutions and block diagonal layers.
Machine learning's predictive power is limited by sample size, as shown by the Limits-to-Learning Gap.
problem The limitations of machine learning in approximating true data-generating processes.
method Characterization of a universal lower bound (LLG) quantifying the discrepancy between empirical fit and population benchmark.
result Standard ML approaches can substantially understate true predictability in financial data.
Machine learning model predicts DFT total energy to complete basis set limit.
problem Finding a model to extrapolate DFT calculations to complete basis set limit.
method Quantile-random-forest model trained on binary solids data.
result Random-forest model achieves <25% symmetric MAPE for both DFT codes.
A new BN method corrects size mismatch in WLF for imbalanced data.
problem Learning from imbalanced datasets in neural networks.
method Proposes weighted batch normalization (WBN) to correct size mismatch between BN and WLF.
result WBN corrects size mismatch and improves classification performance in imbalanced datasets.
Bias correction improves language model training performance.
problem Stochastic update bias in preconditioned optimizers.
method Cross-fitted preconditioning and variance-corrected inversion.
result Reduces held-out pretraining loss by 0.15 nats.
We consider the Degree-Corrected Stochastic Block Model (DC-SBM): a random graph on n nodes, having i.i.d. weights (φu)u=1n (possibly heavy-tailed), partitioned into q≥2 asymptotically equal-sized clusters. The model parameters are two constants a,b>0 and the finite second moment of the weights $Φ^{…
In this paper, we consider asymptotic properties of the support vector machine (SVM) in high-dimension, low-sample-size (HDLSS) settings. We show that the hard-margin linear SVM holds a consistency property in which misclassification rates tend to zero as the dimension goes to infinity under certain severe conditions. …
Learn2Evaluate uses learning curves to estimate high-dimensional prediction performance.
problem Estimating test performance in high-dimensional data settings is challenging.
method Learn2Evaluate uses learning curves to estimate test performance at the total sample size.
result Learn2Evaluate provides a lower confidence bound for performance estimation.
Procedure confirms covariate balance anytime from unlabeled data.
problem Ensuring covariate balance in sequential data.
method Time-uniform confidence sequences for continuous monitoring.
result Probability of false confirmation controlled.
Community detection is a central problem of network data analysis. Given a network, the goal of community detection is to partition the network nodes into a small number of clusters, which could often help reveal interesting structures. The present paper studies community detection in Degree-Corrected Block Models (DCB…
This paper offers a distribution-free method for post-detection changepoint localization.
problem Locating the exact time of a change in distribution after a sequential detection procedure.
method A distribution-free framework using conformal test martingales for sequential change detection and post-detection inference.
result Valid post-detection coverage guarantees and non-asymptotic bounds on confidence set size.
This paper was withdrawn (temporarily?) by the author since an error needs to be corrected.
We make policy optimization algorithms batch size-invariant by decoupling proximal and behavior policies.
problem Some policy optimization algorithms do not have batch size-invariance, leading to inefficiencies.
method We decouple the proximal policy from the behavior policy to achieve batch size-invariance.
result Our approach makes policy optimization algorithms more efficient and allows them to use stale data more effectively.
This paper contains a correction of a mistake made in arXiv:1405.1324
Finite quotients of braid groups have large sizes.
problem Understanding the size of finite quotients of braid groups.
method Derived lower bounds on the size of quotients.
result Lower bounds are superexponential in the number of strands.
Corrected a mistake in a paper about minimal surfaces.
problem A mistake in a paper about complete minimal surfaces with finite total curvature.
method No new method introduced, just correcting an error.
result Corrected a mistake in a previously published paper.
We study products of random matrices in the regime where the number of terms and the size of the matrices simultaneously tend to infinity. Our main theorem is that the logarithm of the ℓ2 norm of such a product applied to any fixed vector is asymptotically Gaussian. The fluctuations we find can be thought of as a…
We analyse time series of CDS spreads for a set of major US and European institutions on a pe- riod overlapping the recent financial crisis. We extend the existing methodology of ε-drawdowns to the one of joint ε-drawups, in order to estimate the conditional probabilities of abrupt co-movements among spreads. We correc…
For a fixed p, there are only finitely many elliptic 3-manifolds given by p/q-surgery on a knot in S^3. We prove this result by using the Heegaard Floer correction terms (d-invariants) to obstruct elliptic manifolds from arising as knot surgery.