We present practical Levenberg-Marquardt variants of Gauss-Newton and natural gradient methods for solving non-convex optimization problems that arise in training deep neural networks involving enormous numbers of variables and huge data sets. Our methods use subsampled Gauss-Newton or Fisher information matrices and e…
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.
168,657 papers · 148 categories
Trend · papers per month
48 results for “subsampled natural gradient”
New analysis shows SNG's effectiveness in small samples.
problem Limited understanding of SNG in small data settings.
method Sketch-and-project analysis of SNG.
result Global convergence and rate characterization for SNG.
A new method reduces variance in SGMCMC by preferentially subsampling data.
problem High variance in stochastic gradient estimates impacts sampler performance.
method Use a non-uniform probability distribution to preferentially subsample data points and adaptively adjust subsample size.
result Maintains accuracy while substantially reducing average subsample size.
Enhances financial optimization under model uncertainty using subsampling.
problem Model uncertainty in financial decision-making from limited data.
method Superimposes uncertainty measure on model space, uses subsampling for model distribution approximation, adapts SGD for efficiency.
result Uncertainty measures outperform traditional methods and achieve comparable performance to Bayesian methods.
SGLDiff approximates Bayesian posterior distributions with subsampling error.
problem Approximating Bayesian posterior distributions in large-scale data settings.
method Stochastic Gradient Langevin Diffusion (SGLDiff) with subsampling.
result The Wasserstein distance between the posterior and SGLDiff's limiting distribution is bounded by a fractional power of the mean waiting time.
Hamiltonian Monte Carlo (HMC) samples efficiently from high-dimensional posterior distributions with proposed parameter draws obtained by iterating on a discretized version of the Hamiltonian dynamics. The iterations make HMC computationally costly, especially in problems with large datasets, since it is necessary to c…
The paper studies the solution of stochastic optimization problems in which approximations to the gradient and Hessian are obtained through subsampling. We first consider Newton-like methods that employ these approximations and discuss how to coordinate the accuracy in the gradient and Hessian to yield a superlinear ra…
A new approach RA improves stochastic optimization by executing multiple steps between subsample updates.
problem Improving the efficiency and effectiveness of stochastic optimization methods.
method Developed Retrospective Approximation (RA) which executes multiple steps between subsample updates using a deterministic solver.
result RA achieves almost sure and consistency under weak conditions and optimizes iteration and oracle complexity.
New method reduces inference variance for faster optimization.
problem High variance in black-box variational inference.
method Joint control variate addressing both data subsampling and Monte Carlo noise.
result Significantly reduced gradient variance, leading to faster optimization.
Many Markov Chain Monte Carlo (MCMC) methods leverage gradient information of the potential function of target distribution to explore sample space efficiently. However, computing gradients can often be computationally expensive for large scale applications, such as those in contemporary machine learning. Stochastic Gr…
New method improves inference for hierarchical models.
problem Challenges in inference for large hierarchical models.
method Locally enhanced variational bounds with subsampling.
result Better posterior approximations than baselines.
Paper bridges statistical inference for DP-SGD, a privacy-preserving machine learning method.
problem Asymptotic statistical inference for Differentially Private Stochastic Gradient Descent (DP-SGD).
method Established asymptotic properties of SGD under randomized subsampling, extended to DP-SGD, proposed methods for constructing valid confidence intervals.
result Valid confidence intervals for DP-SGD output achieve nominal coverage rates while maintaining privacy.
New research finds many coreset methods for logistic regression are not better than simple sampling.
problem Evaluation of coreset methods for reducing data size in logistic regression.
method Comparison of multiple coreset and optimal subsampling methods for logistic regression.
result Many coreset methods do not outperform simple uniform subsampling.
A scalable algorithm for GP regression selects relevant covariates efficiently.
problem Scalable variable selection in large GP regression models.
method VGPR algorithm using Vecchia approximation for sparse precision matrix, mini-batch subsampling.
result Improved scalability and accuracy in selecting relevant covariates.
Large batch sizes reduce gradient variance in DP-SGD, improving privacy.
problem Understanding why large batch sizes work in DP-SGD.
method Decomposed total gradient variance into subsampling and noise-induced variances, proving batch size independence in the limit.
result Large batch sizes reduce effective total gradient variance, improving privacy in DP-SGD.
Sketching, a dimensionality reduction technique, has received much attention in the statistics community. In this paper, we study sketching in the context of Newton's method for solving finite-sum optimization problems in which the number of variables and data points are both large. We study two forms of sketching that…
New method audits DP guarantees without noise or subsampling info.
problem Auditing DP guarantees of ML models without prior info.
method Histogram-based density estimation for lower bounds.
result Natural generalization of membership inference auditing.
We unify subsampling methods for network embeddings and prove their asymptotic distribution.
problem Understanding and improving the performance of network embeddings learned via subsampling.
method Unified framework for node2vec-like methods, proving asymptotic distribution under exchangeable graph assumption.
result Asymptotic distribution of learned embedding vectors decouples and provides rates of convergence.
In the setting of nonparametric regression, we propose and study a combination of stochastic gradient methods with Nyström subsampling, allowing multiple passes over the data and mini-batches. Generalization error bounds for the studied algorithm are provided. Particularly, optimal learning rates are derived considerin…
Stochastic Gradient Hamiltonian Monte Carlo (SGHMC) is a momentum version of stochastic gradient descent with properly injected Gaussian noise to find a global minimum. In this paper, non-asymptotic convergence analysis of SGHMC is given in the context of non-convex optimization, where subsampling techniques are used o…
New insights into how randomization affects greedy model selection.
problem Understanding the impact of feature subsampling on greedy model selection.
method Investigated greedy forward selection with feature subsampling, proving effects on bias and variance.
result Ensembling with feature subsampling reduces both bias and variance, unlike convex base learners.
SNG-DBSCAN clusters data faster with subsampled similarity queries.
problem Efficiently clustering large datasets with DBSCAN's high complexity.
method Subsampled -neighborhood graph for similarity queries.
result Subsampling of the graph leads to 200x speedup and 250x RAM reduction.
Deep reinforcement learning boosts commodities trading performance.
problem Improving algorithmic trading performance in commodities markets.
method Formulated as a stochastic dynamical system, employed actor-based and actor-critic-based policy gradient algorithms with CNN and LSTM function approximators.
result DRL models increase Sharpe ratio by 83% compared to buy-and-hold.
Doubly SGD improves convergence for intractable objective optimization.
problem Optimizing objectives in sum of intractable expectations.
method Doubly SGD with doubly stochastic gradients and independent minibatching.
result Established convergence of doubly SGD under general conditions, including dependent component gradient estimators.
Paper develops methods for statistical inference in SGD with infinite variance.
problem Challenges in statistical inference for SGD with infinite variance.
method Model-agnostic methodology based on weak convergence and subsampling calibration.
result Asymptotically valid confidence regions for SGD in both finite and infinite variance regimes.
A new method selects a representative subsample for efficient kernel density estimation.
problem Selecting a representative subsample without model assumptions.
method Optimal transport techniques for model-free subsampling with an efficient algorithm.
result The selected subsample can be used for efficient density estimation with derived convergence rates and optimal bandwidth.
Tempering by Subsamplingstat.ML
In this paper we demonstrate that tempering Markov chain Monte Carlo samplers for Bayesian models by recursively subsampling observations without replacement can improve the performance of baseline samplers in terms of effective sample size per computation. We present two tempering by subsampling algorithms, subsampled…
Modern computer vision algorithms often rely on very large training datasets. However, it is conceivable that a carefully selected subsample of the dataset is sufficient for training. In this paper, we propose a gradient-based importance measure that we use to empirically analyze relative importance of training images …
Structured subsampling improves privacy in deep time series forecasting.
problem Incompatible privacy guarantees with time series forecasting.
method Structured subsampling of sequential data for privacy amplification.
result Structured subsampling enables training with strong privacy guarantees.
This paper optimizes subsampling for large datasets using Poisson distribution.
problem Efficiently subsample large datasets for quasi-likelihood estimation.
method Derives optimal Poisson subsampling probabilities and develops a distributed subsampling framework.
result Consistent and asymptotically normal estimators are obtained.
Large sample size brings the computation bottleneck for modern data analysis. Subsampling is one of efficient strategies to handle this problem. In previous studies, researchers make more fo- cus on subsampling with replacement (SSR) than on subsampling without replacement (SSWR). In this paper we investigate a kind of…
Safe, Scalable, and Accurate Bayes Posterior Sampling for Large-Data Generalized Linear Mixed Modelsstat.ME
New method for Bayesian inference on large datasets.
problem Scalable sampling for Bayesian generalized linear mixed models on large datasets.
method Mirror Langevin dynamics with data subsampling, post-processing for variance estimation.
result Asymptotic, order-wise correct estimation of posterior variance.
Feature selection from a large number of covariates (aka features) in a regression analysis remains a challenge in data science, especially in terms of its potential of scaling to ever-enlarging data and finding a group of scientifically meaningful features. For example, to develop new, responsive drug targets for ovar…
A significant hurdle for analyzing large sample data is the lack of effective statistical computing and inference methods. An emerging powerful approach for analyzing large sample data is subsampling, by which one takes a random subsample from the original full sample and uses it as a surrogate for subsequent computati…
We study the problem of subsampling in differential privacy (DP), a question that is the centerpiece behind many successful differentially private machine learning algorithms. Specifically, we provide a tight upper bound on the Rényi Differential Privacy (RDP) (Mironov, 2017) parameters for algorithms that: (1) subsamp…
A new model-free subsampling method using uniform designs is proposed.
problem Model-based subsampling methods are often dependent on model assumptions.
method Developed a criterion (GEFD) and a model-free subsampling method based on uniform designs.
result The proposed method outperforms random sampling and is robust under diverse model specifications.
This paper examines a novel gradient boosting framework for regression. We regularize gradient boosted trees by introducing subsampling and employ a modified shrinkage algorithm so that at every boosting stage the estimate is given by an average of trees. The resulting algorithm, titled Boulevard, is shown to converge …
Ensemble methods that average over a collection of independent predictors that are each limited to a subsampling of both the examples and features of the training data command a significant presence in machine learning, such as the ever-popular random forest, yet the nature of the subsampling effect, particularly of th…
State-of-the-art methods for Convolutional Sparse Coding usually employ Fourier-domain solvers in order to speed up the convolution operators. However, this approach is not without shortcomings. For example, Fourier-domain representations implicitly assume circular boundary conditions and make it hard to fully exploit …
The paper studies how adding an ℓ2 penalty affects network embeddings.
problem The impact of ℓ2 regularization on network embeddings.
method Analyzes the asymptotic behavior of ℓ2 regularized node2vec embeddings under graphon theory.
result The learned embeddings asymptotically form a graphon with a nuclear-norm-type penalty.
For massive data, the family of subsampling algorithms is popular to downsize the data volume and reduce computational burden. Existing studies focus on approximating the ordinary least squares estimate in linear regression, where statistical leverage scores are often used to define subsampling probabilities. In this p…
Unified framework for subsampling mechanisms with tighter privacy guarantees.
problem Improving privacy in machine learning models through subsampling.
method Conditional optimal transport for deriving mechanism-specific subsampling guarantees.
result Tighter privacy bounds for subsampled mechanisms compared to traditional methods.
Beyond Consistency: Inference for the Relative risk functional in Deep Nonparametric Cox Modelsstat.ML
Develops asymptotic theory for deep Cox models to enable valid inference.
problem Theoretical gaps in deep neural network estimators for Cox models.
method Asymptotic distribution theory linking in-sample optimization error to population risk.
result Pointwise and multivariate asymptotic normality for subsampled ensemble estimators.
The Sampled Gaussian Mechanism's noise level decreases with larger subsampling rates, improving privacy-utility trade-offs.
problem Improving privacy-utility trade-offs in differentially private stochastic optimization.
method Proof of a conjecture about the Sampled Gaussian Mechanism's noise level and subsampling rate relationship.
result A rigorous proof of the conjecture, completing the proof of Theorem 6.2 in the original paper.
CFIL uses coupled flows to model state distributions for imitation learning.
problem Lack of explicit modeling of state distributions in reinforcement and imitation learning.
method Coupled normalizing flows for state and state-action distributions.
result CFIL achieves state-of-the-art performance on benchmark tasks.
Group-equivariant subsampling layers improve CNNs' equivariance.
problem Non-translation equivariance in subsampling operations.
method Translation and group-equivariant subsampling/upsampling layers.
result Group-equivariant autoencoders learn equivariant representations.
Develops a method to optimize hyperparameters for subsampling methods.
problem Optimizing hyperparameters for subsampling methods to improve estimator efficiency.
method Careful theoretical analysis leading to an optimal choice of hyperparameters.
result Improves the statistical efficiency of subsampling estimators without extra CPU time.
Stochastic Stein Discrepancies improve inference efficiency.
problem Intractable computation of Stein discrepancies.
method Subsampled approximations of Stein operators.
result Stochastic Stein Discrepancies inherit convergence properties of standard SDs.