English
Related papers

Related papers: Avoiding Spurious Local Minima in Deep Quadratic N…

200 papers

Inspired by neuronal diversity in the biological neural system, a plethora of studies proposed to design novel types of artificial neurons and introduce neuronal diversity into artificial neural networks. Recently proposed quadratic neuron,…

Machine Learning · Computer Science 2023-03-14 Feng-Lei Fan , Hang-Cheng Dong , Zhongming Wu , Lecheng Ruan , Tieyong Zeng , Yiming Cui , Jing-Xiao Liao

We show that for any convex differentiable loss, a deep linear network has no spurious local minima as long as it is true for the two layer case. This reduction greatly simplifies the study on the existence of spurious local minima in deep…

Machine Learning · Computer Science 2020-01-07 Li Zhang

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…

Machine Learning · Computer Science 2019-02-06 Simon S. Du , Xiyu Zhai , Barnabas Poczos , Aarti Singh

The critical locus of the loss function of a neural network is determined by the geometry of the functional space and by the parameterization of this space by the network's weights. We introduce a natural distinction between pure critical…

Machine Learning · Computer Science 2020-04-06 Matthew Trager , Kathlén Kohn , Joan Bruna

We identify a class of over-parameterized deep neural networks with standard activation functions and cross-entropy loss which provably have no bad local valley, in the sense that from any point in parameter space there exists a continuous…

Machine Learning · Computer Science 2018-12-27 Quynh Nguyen , Mahesh Chandra Mukkamala , Matthias Hein

Learning in Deep Neural Networks (DNN) takes place by minimizing a non-convex high-dimensional loss function, typically by a stochastic gradient descent (SGD) strategy. The learning process is observed to be able to find good minimizers…

Machine Learning · Computer Science 2020-03-12 Carlo Baldassi , Fabrizio Pittorino , Riccardo Zecchina

One of the major concerns for neural network training is that the non-convexity of the associated loss functions may cause bad landscape. The recent success of neural networks suggests that their loss landscape is not too bad, but what…

Machine Learning · Computer Science 2023-07-19 Ruoyu Sun , Dawei Li , Shiyu Liang , Tian Ding , R Srikant

Neural networks trained via gradient descent with random initialization and without any regularization enjoy good generalization performance in practice despite being highly overparametrized. A promising direction to explain this phenomenon…

Machine Learning · Computer Science 2022-05-17 Hancheng Min , Salma Tarmoun , Rene Vidal , Enrique Mallada

A central challenge to many fields of science and engineering involves minimizing non-convex error functions over continuous, high dimensional spaces. Gradient descent or quasi-Newton methods are almost ubiquitously used to perform such…

Machine Learning · Computer Science 2014-06-11 Yann Dauphin , Razvan Pascanu , Caglar Gulcehre , Kyunghyun Cho , Surya Ganguli , Yoshua Bengio

Low-rank matrix sensing is a fundamental yet challenging nonconvex problem whose optimization landscape typically contains numerous spurious local minima, making it difficult for gradient-based optimizers to converge to the global optimum.…

Machine Learning · Computer Science 2026-02-12 Tianqi Shen , Jinji Yang , Junze He , Kunhan Gao , Ziye Ma

Quantification of the stationary points and the associated basins of attraction of neural network loss surfaces is an important step towards a better understanding of neural network loss surfaces at large. This work proposes a novel method…

Machine Learning · Computer Science 2019-01-10 Anna Sergeevna Bosman , Andries Engelbrecht , Mardé Helbig

The proliferation of saddle points, rather than poor local minima, is increasingly understood to be a primary obstacle in large-scale non-convex optimization for machine learning. Variable elimination algorithms, like Variable Projection…

Machine Learning · Computer Science 2025-11-04 Min Gan , Guang-Yong Chen , Yang Yi , Lin Yang

Due to the success of deep learning to solving a variety of challenging machine learning tasks, there is a rising interest in understanding loss functions for training neural networks from a theoretical aspect. Particularly, the properties…

Machine Learning · Statistics 2017-11-01 Yi Zhou , Yingbin Liang

Overparameterization refers to the important phenomenon where the width of a neural network is chosen such that learning algorithms can provably attain zero loss in nonconvex training. The existing theory establishes such global convergence…

Machine Learning · Computer Science 2021-11-04 Chaehwan Song , Ali Ramezani-Kebrya , Thomas Pethick , Armin Eftekhari , Volkan Cevher

We study the high-dimensional asymptotics of empirical risk minimization (ERM) in over-parametrized two-layer neural networks with quadratic activations trained on synthetic data. We derive sharp asymptotics for both training and test…

Machine Learning · Statistics 2026-02-03 Vittorio Erba , Emanuele Troiani , Lenka Zdeborová , Florent Krzakala

Inspired by the diversity of biological neurons, quadratic artificial neurons can play an important role in deep learning models. The type of quadratic neurons of our interest replaces the inner-product operation in the conventional neuron…

Machine Learning · Computer Science 2023-09-12 Feng-Lei Fan , Mengzhou Li , Fei Wang , Rongjie Lai , Ge Wang

It is widely observed that deep learning models with learned parameters generalize well, even with much more model parameters than the number of training samples. We systematically investigate the underlying reasons why deep neural networks…

Machine Learning · Computer Science 2017-11-29 Lei Wu , Zhanxing Zhu , Weinan E

In this paper, we analyze the landscape of the true loss of neural networks with one hidden layer and ReLU, leaky ReLU, or quadratic activation. In all three cases, we provide a complete classification of the critical points in the case…

Machine Learning · Computer Science 2022-07-07 Patrick Cheridito , Arnulf Jentzen , Florian Rossmannek

A recent line of work has shown that an overparametrized neural network can perfectly fit the training data, an otherwise often intractable nonconvex optimization problem. For (fully-connected) shallow networks, in the best case scenario,…

Machine Learning · Computer Science 2019-10-30 Armin Eftekhari , ChaeHwan Song , Volkan Cevher

The fundamental learning theory behind neural networks remains largely open. What classes of functions can neural networks actually learn? Why doesn't the trained network overfit when it is overparameterized? In this work, we prove that…

Machine Learning · Computer Science 2020-06-02 Zeyuan Allen-Zhu , Yuanzhi Li , Yingyu Liang