We investigate the role of noise in optimization algorithms for learning over-parameterized models. Specifically, we consider the recovery of a rank one matrix $Y^*\in R^{d\times d}$ from a noisy observation $Y$ using an over-parameterization model. We parameterize the rank one matrix $Y^*$ by $XX^\top$, where $X\in R^{d\times d}$. We then show that under mild conditions, the estimator, obtained by the randomly perturbed gradient descent algorithm using the square loss function, attains a mean square error of $O(σ^2/d)$, where $σ^2$ is the variance of the observational noise. In contrast, the estimator obtained by gradient descent without random perturbation only attains a mean square error of $O(σ^2)$. Our result partially justifies the implicit regularization effect of noise when learning over-parameterized models, and provides new understanding of training over-parameterized neural networks.
The Blavatnik School of Computer Science, Tel Aviv University（特拉维夫大学布拉瓦特尼克计算机科学学院）；School of Computer Science and Engineering, The Hebrew University（希伯来大学计算机科学与工程学院）
Neural networks exhibit good generalization behavior in the over-parameterized regime, where the number of network parameters exceeds the number of observations. Nonetheless, current generalization bounds for neural networks fail to explain this phenomenon. In an attempt to bridge this gap, we study the problem of learning a two-layer over-parameterized neural network, when the data is generated by a linearly separable function. In the case where the network has Leaky ReLU activations, we provide both optimization and generalization guarantees for over-parameterized networks. Specifically, we …
Carnegie Mellon University（卡内基梅隆大学）；Massachusetts Institute of Technology（麻省理工学院）
One of the mysteries in the success of neural networks is randomly initialized first order methods like gradient descent can achieve zero training loss even though the objective function is non-convex and non-smooth. This paper demystifies this surprising phenomenon for two-layer fully connected ReLU activated neural networks. For an $m$ hidden node shallow neural network with ReLU activation and $n$ training data, we show as long as $m$ is large enough and no two inputs are parallel, randomly initialized gradient descent converges to a globally optimal solution at a linear convergence rate fo…
In overparametrized models, the noise in stochastic gradient descent (SGD) implicitly regularizes the optimization trajectory and determines which local minimum SGD converges to. Motivated by empirical studies that demonstrate that training with noisy labels improves generalization, we study the implicit regularization effect of SGD with label noise. We show that SGD with label noise converges to a stationary point of a regularized loss $L(θ) +λR(θ)$, where $L(θ)$ is the training loss, $λ$ is an effective regularization parameter depending on the step size, strength of the label noise, and the…
Langevin Monte-Carlo Provably Learns Depth Two Neural Nets at Any Size and Data
Department of Computer Science, The University of Manchester（曼彻斯特大学计算机科学系）；Department of Mathematics, Indian Institute of Technology, Bombay（印度理工学院孟买分校数学系）
University of California, Riverside（加州大学河滨分校）；University of Southern California（南加州大学）
Modern neural networks are typically trained in an over-parameterized regime where the parameters of the model far exceed the size of the training data. Such neural networks in principle have the capacity to (over)fit any set of labels including pure noise. Despite this, somewhat paradoxically, neural network models trained via first-order methods continue to predict well on yet unseen test data. This paper takes a step towards demystifying this phenomena. Under a rich dataset model, we show that gradient descent is provably robust to noise/corruption on a constant fraction of the labels despi…
Amazon Alexa（亚马逊Alexa）；Toyota Technological Institute at Chicago（芝加哥丰田技术研究所）
We study stochastic optimization of nonconvex loss functions, which are typical objectives for training neural networks. We propose stochastic approximation algorithms which optimize a series of regularized, nonlinearized losses on large minibatches of samples, using only first-order gradient information. Our algorithms provably converge to an approximate critical point of the expected objective with faster rates than minibatch stochastic gradient descent, and facilitate better parallelization by allowing larger minibatches.
Regularized nonlinear acceleration (RNA) estimates the minimum of a function by post-processing iterates from an algorithm such as the gradient method. It can be seen as a regularized version of Anderson acceleration, a classical acceleration scheme from numerical analysis. The new scheme provably improves the rate of convergence of fixed step gradient descent, and its empirical performance is comparable to that of quasi-Newton methods. However, RNA cannot accelerate faster multistep algorithms like Nesterov's method and often diverges in this context. Here, we adapt RNA to overcome these issu…
The University of Chicago（芝加哥大学）；Princeton University（普林斯顿大学）；Google Research, Brain Team（谷歌研究院大脑团队）；Northwestern University（西北大学）
Structural equation models (SEMs) are widely used in sciences, ranging from economics to psychology, to uncover causal relationships underlying a complex system under consideration and estimate structural parameters of interest. We study estimation in a class of generalized SEMs where the object of interest is defined as the solution to a linear operator equation. We formulate the linear operator equation as a min-max game, where both players are parameterized by neural networks (NNs), and learn the parameters of these neural networks using the stochastic gradient descent. We consider both 2-l…
University of Cambridge（剑桥大学）；Max Planck Institute for Intelligent Systems（马克斯·普朗克智能系统研究所）；Emory University（埃默里大学）；Mila（米拉人工智能研究所）；Université de Montréal（蒙特利尔大学）；Georgia Institute of Technology（佐治亚理工学院）；Alan Turing Institute（艾伦·图灵研究所）
The inductive bias of a neural network is largely determined by the architecture and the training algorithm. To achieve good generalization, how to effectively train a neural network is of great importance. We propose a novel orthogonal over-parameterized training (OPT) framework that can provably minimize the hyperspherical energy which characterizes the diversity of neurons on a hypersphere. By maintaining the minimum hyperspherical energy during training, OPT can greatly improve the empirical generalization. Specifically, OPT fixes the randomly initialized weights of the neurons and learns …
University of Bath（巴斯大学）；CNRS（法国国家科学研究中心）；ENS（巴黎高等师范学院）；PSL Université（巴黎文理研究大学）
Non-smooth optimization is a core ingredient of many imaging or machine learning pipelines. Non-smoothness encodes structural constraints on the solutions, such as sparsity, group sparsity, low-rank and sharp edges. It is also the basis for the definition of robust loss functions and scale-free functionals such as square-root Lasso. Standard approaches to deal with non-smoothness leverage either proximal splitting or coordinate descent. These approaches are effective but usually require parameter tuning, preconditioning or some sort of support pruning. In this work, we advocate and study a dif…
Fairness-Regularized Online Optimization with Switching Costs
School of Information（信息学院）；Rochester Institute of Technology（罗切斯特技术学院）；Electrical and Computer Engineering（电气与计算机工程）；University of California, Riverside（加州大学河滨分校）；Computing & Mathematical Sciences（计算与数学科学）；California Institute of Technology（加州理工学院）
Google Brain（谷歌大脑）；Institute for Quantum Computing, University of Waterloo（滑铁卢大学量子计算研究所）；UC Berkeley（加州大学伯克利分校）；Stanford University（斯坦福大学）；Princeton University（普林斯顿大学）；University of Waterloo（滑铁卢大学）
The dual tasks of quantum Hamiltonian learning and quantum Gibbs sampling are relevant to many important problems in physics and chemistry. In the low temperature regime, algorithms for these tasks often suffer from intractabilities, for example from poor sample- or time-complexity. With the aim of addressing such intractabilities, we introduce a generalization of quantum natural gradient descent to parameterized mixed states, as well as provide a robust first-order approximating algorithm, Quantum-Probabilistic Mirror Descent. We prove data sample efficiency for the dual tasks using tools fro…
Univ. de Rouen（鲁昂大学）；Criteo AI Lab（Criteo人工智能实验室）；Univ Côte d’Azur（蔚蓝海岸大学）；CNRS（法国国家科学研究中心）；OCA Lagrange（拉格朗日蔚蓝海岸天文台）；INSA de Rouen（鲁昂国立应用科学学院）；IMAG（蒙彼利埃亚历山大·格罗滕迪克研究所）；Université de Montpellier（蒙彼利埃大学）
Owing to their statistical properties, non-convex sparse regularizers have attracted much interest for estimating a sparse linear model from high dimensional data. Given that the solution is sparse, for accelerating convergence, a working set strategy addresses the optimization problem through an iterative algorithm by incre-menting the number of variables to optimize until the identification of the solution support. While those methods have been well-studied and theoretically supported for convex regularizers, this paper proposes a working set algorithm for non-convex sparse regularizers with…
In over-parameterized deep neural networks there can be many possible parameter configurations that fit the training data exactly. However, the properties of these interpolating solutions are poorly understood. We argue that over-parameterized neural networks trained with stochastic gradient descent are subject to a Geometric Occam's Razor; that is, these networks are implicitly regularized by the geometric model complexity. For one-dimensional regression, the geometric model complexity is simply given by the arc length of the function. For higher-dimensional settings, the geometric model comp…
A Validation Approach to Over-parameterized Matrix and Image Recovery
Department of Mathematics, University of California San Diego（加州大学圣地亚哥分校数学系）；Department of Computer Science and Engineering, Ohio State University（俄亥俄州立大学计算机科学与工程系）；H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology（佐治亚理工学院H. Milton Stewart工业与系统工程学院）
Facebook AI Research（Facebook人工智能研究）；Univ Rennes（雷恩大学）；Inria（法国国家信息与自动化研究所）；CNRS（法国国家科学研究中心）；IRISA（信息科学与自动化研究所）
Modern neural networks are over-parametrized. In particular, each rectified linear hidden unit can be modified by a multiplicative factor by adjusting input and output weights, without changing the rest of the network. Inspired by the Sinkhorn-Knopp algorithm, we introduce a fast iterative method for minimizing the L2 norm of the weights, equivalently the weight decay regularizer. It provably converges to a unique solution. Interleaving our algorithm with SGD during training improves the test accuracy. For small batches, our approach offers an alternative to batch-and group-normalization on CI…
This paper considers two-player zero-sum finite-horizon Markov games with simultaneous moves. The study focuses on the challenging settings where the value function or the model is parameterized by general function classes. Provably efficient algorithms for both decoupled and {coordinated} settings are developed. In the {decoupled} setting where the agent controls a single player and plays against an arbitrary opponent, we propose a new model-free algorithm. The sample complexity is governed by the Minimax Eluder dimension -- a new dimension of the function class in Markov games. As a special …
University of California, Berkeley（加州大学伯克利分校）；University of Denver（丹佛大学）；New York University（纽约大学）
Recent advances have shown that implicit bias of gradient descent on over-parameterized models enables the recovery of low-rank matrices from linear measurements, even with no prior knowledge on the intrinsic rank. In contrast, for robust low-rank matrix recovery from grossly corrupted measurements, over-parameterization leads to overfitting without prior knowledge on both the intrinsic rank and sparsity of corruption. This paper shows that with a double over-parameterization for both the low-rank matrix and sparse corruption, gradient descent with discrepant learning rates provably recovers t…
Deep neural networks (DNNs) are capable of perfectly fitting the training data, including memorizing noisy data. It is commonly believed that memorization hurts generalization. Therefore, many recent works propose mitigation strategies to avoid noisy data or correct memorization. In this work, we step back and ask the question: Can deep learning be robust against massive label noise without any mitigation? We provide an affirmative answer for the case of symmetric label noise: We find that certain DNNs, including under-parameterized and over-parameterized models, can tolerate massive symmetric…
University of Chicago（芝加哥大学）；University of Texas at Austin（得克萨斯大学奥斯汀分校）；University of Washington（华盛顿大学）
We propose a computationally-friendly adaptive learning rate schedule, "AdaLoss", which directly uses the information of the loss function to adjust the stepsize in gradient descent methods. We prove that this schedule enjoys linear convergence in linear regression. Moreover, we provide a linear convergence guarantee over the non-convex regime, in the context of two-layer over-parameterized neural networks. If the width of the first-hidden layer in the two-layer networks is sufficiently large (polynomially), then AdaLoss converges robustly \emph{to the global minimum} in polynomial time. We nu…
LieTrunc-QNN: Lie Algebra Truncation and Quantum Expressivity Phase Transition from LiePrune to Provably Stable Quantum Neural Networks
School of Computer, Jiangsu University of Science and Technology（江苏科技大学计算机学院）；Department of Electrical and Computer Engineering, University of Nevada, Las Vegas（内华达大学拉斯维加斯分校电气与计算机工程系）
Provably Robust Training of Quantum Circuit Classifiers Against Parameter Noise
Computer Science, University of California Los Angeles（计算机科学，加州大学洛杉矶分校）；Electrical and Computer Engineering, University of California Los Angeles（电气与计算机工程，加州大学洛杉矶分校）
The classical low rank approximation problem is to find a rank $k$ matrix $UV$ (where $U$ has $k$ columns and $V$ has $k$ rows) that minimizes the Frobenius norm of $A - UV$. Although this problem can be solved efficiently, we study an NP-hard variant of this problem that involves weights and regularization. A previous paper of [Razenshteyn et al. '16] derived a polynomial time algorithm for weighted low rank approximation with constant rank. We derive provably sharper guarantees for the regularized version by obtaining parameterized complexity bounds in terms of the statistical dimension rath…
Recent works have demonstrated that increasing model capacity through width in over-parameterized neural networks leads to a decrease in test risk. For neural networks, however, model capacity can also be increased through depth, yet understanding the impact of increasing depth on test risk remains an open question. In this work, we demonstrate that the test risk of over-parameterized convolutional networks is a U-shaped curve (i.e. monotonically decreasing, then increasing) with increasing depth. We first provide empirical evidence for this phenomenon via image classification experiments usin…
University of Maryland（马里兰大学）；University of Siegen（锡根大学）
Data poisoning and backdoor attacks manipulate training data to induce security breaches in a victim model. These attacks can be provably deflected using differentially private (DP) training methods, although this comes with a sharp decrease in model performance. The InstaHide method has recently been proposed as an alternative to DP training that leverages supposed privacy properties of the mixup augmentation, although without rigorous guarantees. In this work, we show that strong data augmentations, such as mixup and random additive noise, nullify poison attacks while enduring only a small a…
Baidu Research（百度研究院）；Paradigm Inc.（第四范式公司）；Tsinghua University（清华大学）；Hong Kong University of Science and Technology（香港科技大学）
Matrix learning is at the core of many machine learning problems. A number of real-world applications such as collaborative filtering and text mining can be formulated as a low-rank matrix completion problem, which recovers incomplete matrix using low-rank assumptions. To ensure that the matrix solution has a low rank, a recent trend is to use nonconvex regularizers that adaptively penalize singular values. They offer good recovery performance and have nice theoretical properties, but are computationally expensive due to repeated access to individual singular values. In this paper, based on th…
Technical University of Munich（慕尼黑工业大学）；Rice University（莱斯大学）
An important problem in machine learning is the ability to learn tasks in a sequential manner. If trained with standard first-order methods most models forget previously learned tasks when trained on a new task, which is often referred to as catastrophic forgetting. A popular approach to overcome forgetting is to regularize the loss function by penalizing models that perform poorly on previous tasks. For example, elastic weight consolidation (EWC) regularizes with a quadratic form involving a diagonal matrix build based on past data. While EWC works very well for some setups, we show that, eve…
Modern tasks in reinforcement learning have large state and action spaces. To deal with them efficiently, one often uses predefined feature mapping to represent states and actions in a low-dimensional space. In this paper, we study reinforcement learning for discounted Markov Decision Processes (MDPs), where the transition kernel can be parameterized as a linear function of certain feature mapping. We propose a novel algorithm that makes use of the feature mapping and obtains a $\tilde O(d\sqrt{T}/(1-γ)^2)$ regret, where $d$ is the dimension of the feature space, $T$ is the time horizon and $γ…
Accelerating nuclear-norm regularized low-rank matrix optimization through Burer-Monteiro decomposition
Institute of Statistical Mathematics（统计数学研究所）；Department of Mathematics University of Maryland at College Park（大学数学系）；Institute of Operations Research and Analytics National University of Singapore（运营研究与分析国家大学研究所）；Department of Mathematics and Institute of Operations Research and Analytics National University of Singapore（大学数学系和运营研究与分析国家大学研究所）
Cornell University（康奈尔大学）；University of British Columbia（不列颠哥伦比亚大学）；Microsoft Research（微软研究院）
Algorithm configuration methods optimize the performance of a parameterized heuristic algorithm on a given distribution of problem instances. Recent work introduced an algorithm configuration procedure ("Structured Procrastination") that provably achieves near optimal performance with high probability and with nearly minimal runtime in the worst case. It also offers an $\textit{anytime}$ property: it keeps tightening its optimality guarantees the longer it is run. Unfortunately, Structured Procrastination is not $\textit{adaptive}$ to characteristics of the parameterized algorithm: it treats e…
We develop a convex analytic approach to analyze finite width two-layer ReLU networks. We first prove that an optimal solution to the regularized training problem can be characterized as extreme points of a convex set, where simple solutions are encouraged via its convex geometrical properties. We then leverage this characterization to show that an optimal set of parameters yield linear spline interpolation for regression problems involving one dimensional or rank-one data. We also characterize the classification decision regions in terms of a kernel matrix and minimum $\ell_1$-norm solutions.…
University of Cambridge（剑桥大学）；MPI-IS Tübingen（图宾根马克斯·普朗克智能系统研究所）；Emory University（埃默里大学）；Université de Montréal（蒙特利尔大学）；Alan Turing Institute（艾伦·图灵研究所）
Due to the over-parameterization nature, neural networks are a powerful tool for nonlinear function approximation. In order to achieve good generalization on unseen data, a suitable inductive bias is of great importance for neural networks. One of the most straightforward ways is to regularize the neural network with some additional objectives. L2 regularization serves as a standard regularization for neural networks. Despite its popularity, it essentially regularizes one dimension of the individual neuron, which is not strong enough to control the capacity of highly over-parameterized neural …
Rethinking Influence Functions of Neural Networks in the Over-parameterized Regime
Academy of Mathematics and Systems Science, Chinese Academy of Sciences（中国科学院数学与系统科学研究院）；School of Mathematical Sciences, University of Chinese Academy of Sciences（中国科学院大学数学科学学院）
