# The Power of Interpolation: Understanding the Effectiveness of SGD in Modern Over-parametrized Learning

arXiv: https://arxiv.org/abs/1712.06559 · HTML: https://mlsanity.com/paper/1712.06559/

## Key facts

- Cited by 150 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Dec 2017.

## Built on by (most cited first, 30 of 150)

- [A Geometric Analysis of Neural Collapse with Unconstrained Features](https://mlsanity.com/paper/2105.02375/) (2021)
- [Unlocking High-Accuracy Differentially Private Image Classification through Scale](https://mlsanity.com/paper/2204.13650/) (2022)
- [On the Origin of Implicit Regularization in Stochastic Gradient Descent](https://mlsanity.com/paper/2101.12176/) (2021)
- [Exploring Deep Neural Networks via Layer-Peeled Model: Minority Collapse in Imbalanced Training](https://mlsanity.com/paper/2101.12699/) (2021)
- [Fit without fear: remarkable mathematical phenomena of deep learning through the prism of interpolation](https://mlsanity.com/paper/2105.14368/) (2021)
- [Extended Unconstrained Features Model for Exploring Deep Neural Collapse](https://mlsanity.com/paper/2202.08087/) (2022)
- [Neural Collapse: A Review on Modelling Principles and Generalization](https://mlsanity.com/paper/2206.04041/) (2022)
- [Information-Theoretic Generalization Bounds for Stochastic Gradient Descent](https://mlsanity.com/paper/2102.00931/) (2021)
- [On Large-Cohort Training for Federated Learning](https://mlsanity.com/paper/2106.07820/) (2021)
- [How Does Critical Batch Size Scale in Pre-training?](https://mlsanity.com/paper/2410.21676/) (2024)
- [In Defense of the Unitary Scalarization for Deep Multi-Task Learning](https://mlsanity.com/paper/2201.04122/) (2022)
- [Stochastic Training is Not Necessary for Generalization](https://mlsanity.com/paper/2109.14119/) (2021)
- [From high-dimensional & mean-field dynamics to dimensionless ODEs: A unifying approach to SGD in two-layers networks](https://mlsanity.com/paper/2302.05882/) (2023)
- [Power Lines: Scaling Laws for Weight Decay and Batch Size in LLM Pre-training](https://mlsanity.com/paper/2505.13738/) (2025)
- [Last iterate convergence of SGD for Least-Squares in the Interpolation regime](https://mlsanity.com/paper/2102.03183/) (2021)
- [Batch size-invariance for policy optimization](https://mlsanity.com/paper/2110.00641/) (2021)
- [Perturbation Analysis of Neural Collapse](https://mlsanity.com/paper/2210.16658/) (2022)
- [Drawing Multiple Augmentation Samples Per Image During Training Efficiently Decreases Test Error](https://mlsanity.com/paper/2105.13343/) (2021)
- [Implicit Gradient Alignment in Distributed and Federated Learning](https://mlsanity.com/paper/2106.13897/) (2021)
- [There Will Be a Scientific Theory of Deep Learning](https://mlsanity.com/paper/2604.21691/) (2026)
- [Convergence Bound and Critical Batch Size of Muon Optimizer](https://mlsanity.com/paper/2507.01598/) (2025)
- [MoMo: Momentum Models for Adaptive Learning Rates](https://mlsanity.com/paper/2305.07583/) (2023)
- [ScaleCom: Scalable Sparsified Gradient Compression for Communication-Efficient Distributed Training](https://mlsanity.com/paper/2104.11125/) (2021)
- [SGD: The Role of Implicit Regularization, Batch-size and Multiple-epochs](https://mlsanity.com/paper/2107.05074/) (2021)
- [Stochastic Mirror Descent: Convergence Analysis and Adaptive Variants via the Mirror Stochastic Polyak Stepsize](https://mlsanity.com/paper/2110.15412/) (2021)
- [Random matrix analysis of deep neural network weight matrices](https://mlsanity.com/paper/2203.14661/) (2022)
- [Beyond Lipschitz: Sharp Generalization and Excess Risk Bounds for Full-Batch GD](https://mlsanity.com/paper/2204.12446/) (2022)
- [Trajectory of Mini-Batch Momentum: Batch Size Saturation and Convergence in High Dimensions](https://mlsanity.com/paper/2206.01029/) (2022)
- [Beyond Uniform Lipschitz Condition in Differentially Private Optimization](https://mlsanity.com/paper/2206.10713/) (2022)
- [When and Why Momentum Accelerates SGD:An Empirical Study](https://mlsanity.com/paper/2306.09000/) (2023)

## Abstract (arXiv)

In this paper we aim to formally explain the phenomenon of fast convergence of SGD observed in modern machine learning. The key observation is that most modern learning architectures are over-parametrized and are trained to interpolate the data by driving the empirical loss (classification and regression) close to zero. While it is still unclear why these interpolated solutions perform well on test data, we show that these regimes allow for fast convergence of SGD, comparable in number of iterations to full gradient descent. For convex loss functions we obtain an exponential convergence bound for it mini-batch SGD parallel to that for full gradient descent. We show that there is a critical batch size m^* such that: (a) SGD iteration with mini-batch size m≤ m^* is nearly equivalent to m iterations of mini-batch size 1 (linear scaling regime). (b) SGD iteration with mini-batch m> m^* is nearly equivalent to a full gradient descent iteration (saturation regime). Moreover, for the quadratic loss, we derive explicit expressions for the optimal mini-batch and step size and explicitly characterize the two regimes above. The critical mini-batch size can be viewed as the limit for effective mini-batch parallelization. It is also nearly independent of the data size, implying O(n) acceleration over GD per unit of computation. We give experimental evidence on real data which closely follows our theoretical analyses. Finally, we show how our results fit in the recent developments in training deep neural networks and discuss connections to adaptive rates for SGD and variance reduction.