Efficiently samples sequences without replacement for machine learning models.
problem Generating diverse outputs from sequential models without duplicates.
method Incremental sampling procedure for randomized programs, including neural models.
result Efficacy and flexibility of incremental sampling for large output spaces.
The well-known Gumbel-Max trick for sampling from a categorical distribution can be extended to sample k elements without replacement. We show how to implicitly apply this 'Gumbel-Top-k' trick on a factorized distribution over sequences, allowing to draw exact samples without replacement using a Stochastic Beam Sea…
New sketches for weighted ℓp sampling without replacement improve accuracy and efficiency.
problem Efficiently sampling weighted data with high accuracy and minimal redundancy.
method Design of novel composable sketches for WOR ℓp sampling, based on CountSketch. result First to provide WOR sampling for p>1 and signed updates. Sampling without replacement speeds up optimization in minimax problems.
problem Optimizing minimax problems with faster convergence rates.
method Analysis of gradient descent ascent and proximal point method with two sampling strategies.
result Sampling without replacement leads to faster convergence rates in minimax optimization.
The paper introduces methods to quantify uncertainty in sampling without replacement.
problem Accurately estimating parameters from finite populations sampled without replacement.
method Develops confidence sequences using Bayesian and empirical methods.
result Improved confidence intervals and sequences for sampling without replacement.
New estimator reduces variance in discrete random variables.
problem Estimating gradients for discrete random variables with reduced variance.
method Sampling without replacement and Rao-Blackwellization.
result Our estimator is the most consistent gradient estimator across different entropy settings.
Stochastic gradient methods for machine learning and optimization problems are usually analyzed assuming data points are sampled \emph{with} replacement. In practice, however, sampling \emph{without} replacement is very common, easier to implement in many cases, and often performs better. In this paper, we provide comp…
Paper closes convergence gap for SGD without replacement.
problem Establishing convergence rate for SGD without replacement.
method Analyzes convergence rates for strongly convex and smooth functions.
result Achieves a rate of O(1/T^2 + n^2/T^3) for quadratic sums.
New research disproves a key conjecture in optimization.
problem Comparison of sampling methods in stochastic optimization.
method Reduction to noncommutative arithmetic-geometric mean inequality and application of noncommutative Positivstellensatz.
result The Recht-Ré conjecture is false for general n.
SGD without replacement decouples into curvature-following and flatness-regularizing steps.
problem Theoretical analysis of SGD without replacement for large-scale neural networks.
method Analysis of SGD without replacement in a realistic regime, considering high curvature and flatness.
result Optimizing with SGD without replacement is locally equivalent to an additional regularizer step.
We introduce a variant of Shepp's classical urn problem in which the optimal stopper does not know whether sampling from the urn is done with or without replacement. By considering the problem's continuous-time analog, we provide bounds on the value function and in the case of a balanced urn (with an equal number of ea…
Randomized algorithms that base iteration-level decisions on samples from some pool are ubiquitous in machine learning and optimization. Examples include stochastic gradient descent and randomized coordinate descent. This paper makes progress at theoretically evaluating the difference in performance between sampling wi…
Crowdsourcing platforms are now extensively used for conducting subjective pairwise comparison studies. In this setting, a pairwise comparison dataset is typically gathered via random sampling, either \emph{with} or \emph{without} replacement. In this paper, we use tools from random graph theory to analyze these two ra…
Unsupervised algorithm parses CSG images into CFG without pretraining.
problem Sparse reward problem in unsupervised program synthesis for images.
method Grammar-encoded tree LSTM, entropy regularization, sampling without replacement.
result Recover meaningful programs in large search spaces (up to 3.8imes1028). RLFA estimates misstated monetary fraction with weighted sampling without replacement.
problem Estimating misstated monetary fraction with given accuracy and confidence.
method Developed new confidence sequences for weighted average of unknown values using randomized weighted sampling and side information.
result Adaptive methods improve accuracy of estimates based on side information's predictive power.
We study stochastic gradient descent {\em without replacement} (\sgdwor) for smooth convex functions. \sgdwor is widely observed to converge faster than true \sgd where each sample is drawn independently {\em with replacement} \cite{bottou2009curiously} and hence, is more popular in practice. But it's convergence prope…
Optimal algorithms for online convex optimization with random order.
problem Online convex optimization with random order and non-convex loss functions.
method Stochastic gradient descent and algorithmic stability analysis.
result Achieves optimal bounds and significantly outperforms previous methods.
Differential privacy is a useful tool to build machine learning models which do not release too much information about the training data. We study the Rényi differential privacy of stochastic gradient descent when each training example is sampled without replacement (also known as cyclic SGD). Cyclic SGD is typically f…
Formulates a Dueling Bandits problem for eliciting Kemeny rankings.
problem Eliciting preferences to find a Kemeny ranking.
method Formulates the problem as a Dueling Bandits problem, considering sampling with and without replacement.
result Approximation bounds and algorithms for finding PAC solutions with sample complexity.
New research shows CI in few-shot learning is misleading due to sampling with replacement.
problem Misleading confidence intervals in few-shot learning due to sampling with replacement.
method Comparative analysis of CIs computed with and without replacement.
result Significant underestimation of CI by the predominant method.
New insights into privacy guarantees for subsampled mechanisms under composition.
problem Tight privacy guarantees for the composition of subsampled differentially private mechanisms.
method Addressed confusion points in privacy accounting for subsampled mechanisms, providing examples and counterexamples.
result Privacy guarantees for subsampled mechanisms differ significantly between Poisson subsampling and sampling without replacement.
Optimized sampling scheme for compressed sensing combining randomness and determinism.
problem Improving compressed sensing performance with deterministic sampling.
method Optimized sampling scheme combining random and deterministic selection of rows.
result Measurable improvements in image compressed sensing for generative and sparse priors.
Paper improves CI and CS for bounded means using betting and mixtures.
problem Estimating means of bounded random variables.
method Composite nonnegative martingales, testing by betting, method of mixtures.
result Empirically outperforms existing CI and CS methods.
Improved convergence for VIPs with SEG-RR, a variant of SEG with random reshuffling.
problem Solving variational inequality problems (VIPs) in machine learning.
method Stochastic Extragradient with Random Reshuffling (SEG-RR).
result SEG-RR achieves faster convergence rates than with-replacement variants for certain VIP classes.
Random forests perform bootstrap-aggregation by sampling the training samples with replacement. This enables the evaluation of out-of-bag error which serves as a internal cross-validation mechanism. Our motivation lies in using the unsampled training samples to improve each decision tree in the ensemble. We study the e…
Paper introduces S-SSE for stable sparse subspace embedding.
problem Inefficient sparse random projection matrices with uneven non-zero distribution.
method Uses uniform sampling without replacement to create a stable sparse subspace embedded matrix (S-SSE).
result S-SSE maintains Euclidean distance better after dimension reduction.
Enhances mixture models with classifier-defined weights.
problem Density evaluation and sampling in mixture models.
method Introduces Classifier Weighted Mixtures (CWM) with functional weights.
result Improves expressivity in variational estimation without increasing complexity.
This study examines biases in flow matching samplers using finite-sample estimation.
problem Biases in flow matching samplers when using finite-sample surrogates.
method Finite-sample plug-in estimation and hierarchy of empirical FM models.
result Exact empirical minimizer and smoothed plug-in regime identified for affine conditional flows.
New convergence bounds for shuffling-based SGD methods in distributed learning.
problem Analyzing the performance of shuffling-based variants of SGD in distributed learning.
method Study of minibatch and local Random Reshuffling methods, proving convergence bounds and lower bounds.
result Shuffling-based variants converge faster than with-replacement sampling methods, and the bounds are tight.
New priors improve Bayesian neural networks without cooling.
problem Bayesian neural networks underfit on clean datasets.
method Introduce DirClip and confidence priors to replace cooling.
result DirClip and confidence priors outperform cold posterior.
Enhanced privacy, utility, and efficiency through MUST subsampling.
problem Balancing privacy, utility, and computational efficiency in data analysis.
method MUltistage Sampling Technique (MUST) for privacy amplification in differential privacy.
result MUST offers stronger privacy guarantees (ϵ) than one-stage subsampling methods while maintaining similar utility and computational efficiency. New loss function equivalence reveals PER's uniform sampling can be improved.
problem Improving Prioritized Experience Replay (PER) for better learning efficiency.
method Transforming non-uniformly sampled data loss functions into uniformly sampled ones.
result Some environments can replace PER with a new loss function without performance loss.
Recent advances in optimization theory have shown that smooth strongly convex finite sums can be minimized faster than by treating them as a black box "batch" problem. In this work we introduce a new method in this class with a theoretical convergence rate four times faster than existing methods, for sums with sufficie…
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…
In many applications of relational learning, the available data can be seen as a sample from a larger relational structure (e.g. we may be given a small fragment from some social network). In this paper we are particularly concerned with scenarios in which we can assume that (i) the domain elements appearing in the giv…
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…
We show two novel concentration inequalities for suprema of empirical processes when sampling without replacement, which both take the variance of the functions into account. While these inequalities may potentially have broad applications in learning theory in general, we exemplify their significance by studying the t…
New MC simulation methods use classifiers to estimate pdf ratios without explicit pdfs.
problem Estimating ratios of probability density functions (pdfs) without explicit pdfs.
method Proposes classifier-based pdf-free versions of MC simulation algorithms.
result Enables pdf-free simulation algorithms using surrogate functions computed by classifiers.
A new model for graph sampling that preserves structure without explicit targeting.
problem Graphs are often not fully representative of true relationships, leading to biased machine learning models.
method Node copying model: randomly replaces each node's neighbors with those of a randomly sampled similar node.
result The model achieves higher accuracy in node classification and mitigates adversarial attacks.
Improved UIVI method shows better performance than state-of-the-art SIVI methods.
problem Estimating the likelihood of samples from complex distributions in high dimensions.
method Replaced the inner MCMC loop of UIVI with importance sampling and learned the optimal proposal distribution.
result The refined UIVI approach demonstrates superior performance or parity with state-of-the-art methods.
GAGA accelerates 3D molecular generation by replacing long trajectories with Gaussian approximations.
problem High computational cost of long generative trajectories in 3D molecular generation.
method GAGA identifies a characteristic step where molecular data becomes sufficiently Gaussian, replacing the trajectory with a Gaussian approximation.
result Significant improvement in both generation quality and computational efficiency.
We propose sequenced-replacement sampling (SRS) for training deep neural networks. The basic idea is to assign a fixed sequence index to each sample in the dataset. Once a mini-batch is randomly drawn in each training iteration, we refill the original dataset by successively adding samples according to their sequence i…
WMPG reduces policy gradient variance using world models.
problem Reducing variance in policy gradient estimates.
method Trains a world model online to estimate policy gradients and uses imagined trajectories as a baseline.
result WMPG achieves better sample efficiency compared to AC and MAC.
SimFBO simplifies FBO, making it more efficient and flexible.
problem Complex nested optimization in machine learning and edge computing.
method Proposes SimFBO, a simple and flexible FBO framework with improved communication efficiency.
result SimFBO and ShroFBO achieve linear convergence speedup and improved sample and communication complexities.
New simulation approaches to evaluating path-dependent options without matrix inversion issues nor Euler bias are evaluated. They employ three main contributions: Stochastic approximation replaces regression in the LSM algorithm; Explicit weak solutions to stochastic differential equations are developed and applied to …
New SGDA method speeds up nonconvex minimax optimization.
problem Improving convergence of nonconvex minimax optimization.
method SGDA with random reshuffling for nonconvex-PŁ objectives.
result Convergence rates faster than with-replacement SGDA.
Cubic regularization (CR) is an optimization method with emerging popularity due to its capability to escape saddle points and converge to second-order stationary solutions for nonconvex optimization. However, CR encounters a high sample complexity issue for finite-sum problems with a large data size. %Various inexact …
FMP sampling improves model calibration without sharing data.
problem Difficulties in specifying priors for modern neural networks.
method FMP sampling replaces prior and likelihood with a predictive distribution, running the sampler centrally.
result FMP sampling significantly improves calibration over baseline methods.