The Sample Complexity of Gradient Descent in Stochastic Convex Optimization

• URL: http://arxiv.org/abs/2404.04931v2
• Date: Thu, 11 Apr 2024 02:32:43 GMT
• Title: The Sample Complexity of Gradient Descent in Stochastic Convex Optimization
• Authors: Roi Livni,
• Abstract summary: We show that the generalization error of full-batch Gradient Descent can be $tilde Theta(d/m + 1/sqrtm)$, where $d$ is the dimension and $m$ is the sample size. This matches the sample complexity of emphworst-case empirical risk minimizers.
• Score: 14.268363583731848
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: We analyze the sample complexity of full-batch Gradient Descent (GD) in the setup of non-smooth Stochastic Convex Optimization. We show that the generalization error of GD, with common choice of hyper-parameters, can be $\tilde \Theta(d/m + 1/\sqrt{m})$, where $d$ is the dimension and $m$ is the sample size. This matches the sample complexity of \emph{worst-case} empirical risk minimizers. That means that, in contrast with other algorithms, GD has no advantage over naive ERMs. Our bound follows from a new generalization bound that depends on both the dimension as well as the learning rate and number of iterations. Our bound also shows that, for general hyper-parameters, when the dimension is strictly larger than number of samples, $T=\Omega(1/\epsilon^4)$ iterations are necessary to avoid overfitting. This resolves an open problem by Schlisserman et al.23 and Amir er Al.21, and improves over previous lower bounds that demonstrated that the sample size must be at least square root of the dimension.

Related papers

• Dimension-free Private Mean Estimation for Anisotropic Distributions [55.86374912608193]
  Previous private estimators on distributions over $mathRd suffer from a curse of dimensionality. We present an algorithm whose sample complexity has improved dependence on dimension.
  arXiv  Detail & Related papers  (2024-11-01T17:59:53Z)
• Precise Asymptotics of Bagging Regularized M-estimators [5.165142221427928]
  We characterize the squared prediction risk of ensemble estimators obtained through subagging (subsample bootstrap aggregating) regularized M-estimators. Key to our analysis is a new result on the joint behavior of correlations between the estimator and residual errors on overlapping subsamples. Joint optimization of subsample size, ensemble size, and regularization can significantly outperform regularizer optimization alone on the full data.
  arXiv  Detail & Related papers  (2024-09-23T17:48:28Z)
• Measuring quantum relative entropy with finite-size effect [53.64687146666141]
  We study the estimation of relative entropy $D(rho|sigma)$ when $sigma$ is known. Our estimator attains the Cram'er-Rao type bound when the dimension $d$ is fixed.
  arXiv  Detail & Related papers  (2024-06-25T06:07:20Z)
• Projection by Convolution: Optimal Sample Complexity for Reinforcement Learning in Continuous-Space MDPs [56.237917407785545]
  We consider the problem of learning an $varepsilon$-optimal policy in a general class of continuous-space Markov decision processes (MDPs) having smooth Bellman operators. Key to our solution is a novel projection technique based on ideas from harmonic analysis. Our result bridges the gap between two popular but conflicting perspectives on continuous-space MDPs.
  arXiv  Detail & Related papers  (2024-05-10T09:58:47Z)
• The Dimension Strikes Back with Gradients: Generalization of Gradient Methods in Stochastic Convex Optimization [30.26365073195728]
  We study the generalization performance of gradient methods in the fundamental convex optimization setting. We show that an application of the same construction technique provides a similar $Omega(sqrtd)$ lower bound for the sample complexity of SGD to reach a non-trivial empirical error.
  arXiv  Detail & Related papers  (2024-01-22T15:50:32Z)
• The Sample Complexity Of ERMs In Stochastic Convex Optimization [13.896417716930687]
  We show that in fact $tildeO(fracdepsilon+frac1epsilon2)$ data points are also sufficient. We further generalize the result and show that a similar upper bound holds for all convex bodies.
  arXiv  Detail & Related papers  (2023-11-09T14:29:25Z)
• Stochastic Approximation Approaches to Group Distributionally Robust Optimization [96.26317627118912]
  Group distributionally robust optimization (GDRO) Online learning techniques to reduce the number of samples required in each round from $m$ to $1$, keeping the same sample. A novel formulation of weighted GDRO, which allows us to derive distribution-dependent convergence rates.
  arXiv  Detail & Related papers  (2023-02-18T09:24:15Z)
• Best Policy Identification in Linear MDPs [70.57916977441262]
  We investigate the problem of best identification in discounted linear Markov+Delta Decision in the fixed confidence setting under a generative model. The lower bound as the solution of an intricate non- optimization program can be used as the starting point to devise such algorithms.
  arXiv  Detail & Related papers  (2022-08-11T04:12:50Z)
• Robust High Dimensional Expectation Maximization Algorithm via Trimmed Hard Thresholding [24.184520829631587]
  We study the problem of estimating latent variable models with arbitrarily corrupted samples in high dimensional space. We propose a method called Trimmed (Gradient) Expectation Maximization which adds a trimming gradient step. We show that the algorithm is corruption-proofing and converges to the (near) optimal statistical rate geometrically.
  arXiv  Detail & Related papers  (2020-10-19T15:00:35Z)
• Hybrid Stochastic-Deterministic Minibatch Proximal Gradient: Less-Than-Single-Pass Optimization with Nearly Optimal Generalization [83.80460802169999]
  We show that HSDMPG can attain an $mathcalObig (1/sttnbig)$ which is at the order of excess error on a learning model. For loss factors, we prove that HSDMPG can attain an $mathcalObig (1/sttnbig)$ which is at the order of excess error on a learning model.
  arXiv  Detail & Related papers  (2020-09-18T02:18:44Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.