Chaoyue Liu

Libin Zhu, Chaoyue Liu, Adityanarayanan Radhakrishnan et al.

In this paper, we first present an explanation regarding the common occurrence of spikes in the training loss when neural networks are trained with stochastic gradient descent (SGD). We provide evidence that the spikes in the training loss of SGD are "catapults", an optimization phenomenon originally observed in GD with large learning rates in [Lewkowycz et al. 2020]. We empirically show that these catapults occur in a low-dimensional subspace spanned by the top eigenvectors of the tangent kernel, for both GD and SGD. Second, we posit an explanation for how catapults lead to better generalization by demonstrating that catapults promote feature learning by increasing alignment with the Average Gradient Outer Product (AGOP) of the true predictor. Furthermore, we demonstrate that a smaller batch size in SGD induces a larger number of catapults, thereby improving AGOP alignment and test performance.

Chaoyue Liu, Dmitriy Drusvyatskiy, Mikhail Belkin et al.

Modern machine learning paradigms, such as deep learning, occur in or close to the interpolation regime, wherein the number of model parameters is much larger than the number of data samples. In this work, we propose a regularity condition within the interpolation regime which endows the stochastic gradient method with the same worst-case iteration complexity as the deterministic gradient method, while using only a single sampled gradient (or a minibatch) in each iteration. In contrast, all existing guarantees require the stochastic gradient method to take small steps, thereby resulting in a much slower linear rate of convergence. Finally, we demonstrate that our condition holds when training sufficiently wide feedforward neural networks with a linear output layer.

Libin Zhu, Chaoyue Liu, Adityanarayanan Radhakrishnan et al.

While neural networks can be approximated by linear models as their width increases, certain properties of wide neural networks cannot be captured by linear models. In this work we show that recently proposed Neural Quadratic Models can exhibit the "catapult phase" [Lewkowycz et al. 2020] that arises when training such models with large learning rates. We then empirically show that the behaviour of neural quadratic models parallels that of neural networks in generalization, especially in the catapult phase regime. Our analysis further demonstrates that quadratic models can be an effective tool for analysis of neural networks.

Chaoyue Liu, Libin Zhu, Mikhail Belkin

Wide neural networks with linear output layer have been shown to be near-linear, and to have near-constant neural tangent kernel (NTK), in a region containing the optimization path of gradient descent. These findings seem counter-intuitive since in general neural networks are highly complex models. Why does a linear structure emerge when the networks become wide? In this work, we provide a new perspective on this "transition to linearity" by considering a neural network as an assembly model recursively built from a set of sub-models corresponding to individual neurons. In this view, we show that the linearity of wide neural networks is, in fact, an emerging property of assembling a large number of diverse "weak" sub-models, none of which dominate the assembly.

Chaoyue Liu, Amirhesam Abedsoltan, Mikhail Belkin

When random label noise is added to a training dataset, the prediction error of a neural network on a label-noise-free test dataset initially improves during early training but eventually deteriorates, following a U-shaped dependence on training time. This behaviour is believed to be a result of neural networks learning the pattern of clean data first and fitting the noise later in the training, a phenomenon that we refer to as clean-priority learning. In this study, we aim to explore the learning dynamics underlying this phenomenon. We theoretically demonstrate that, in the early stage of training, the update direction of gradient descent is determined by the clean subset of training data, leaving the noisy subset has minimal to no impact, resulting in a prioritization of clean learning. Moreover, we show both theoretically and experimentally, as the clean-priority learning goes on, the dominance of the gradients of clean samples over those of noisy samples diminishes, and finally results in a termination of the clean-priority learning and fitting of the noisy samples.

Libin Zhu, Chaoyue Liu, Mikhail Belkin

In this paper we show that feedforward neural networks corresponding to arbitrary directed acyclic graphs undergo transition to linearity as their "width" approaches infinity. The width of these general networks is characterized by the minimum in-degree of their neurons, except for the input and first layers. Our results identify the mathematical structure underlying transition to linearity and generalize a number of recent works aimed at characterizing transition to linearity or constancy of the Neural Tangent Kernel for standard architectures.

Marc Duquesnoy, Chaoyue Liu, Vishank Kumar et al.

The optimization of the electrode manufacturing process is important for upscaling the application of Lithium Ion Batteries (LIBs) to cater for growing energy demand. In particular, LIB manufacturing is very important to be optimized because it determines the practical performance of the cells when the latter are being used in applications such as electric vehicles. In this study, we tackled the issue of high-performance electrodes for desired battery application conditions by proposing a powerful data-driven approach supported by a deterministic machine learning (ML)-assisted pipeline for bi-objective optimization of the electrochemical performance. This ML pipeline allows the inverse design of the process parameters to adopt in order to manufacture electrodes for energy or power applications. The latter work is an analogy to our previous work that supported the optimization of the electrode microstructures for kinetic, ionic, and electronic transport properties improvement. An electrochemical pseudo-two-dimensional model is fed with the electrode properties characterizing the electrode microstructures generated by manufacturing simulations and used to simulate the electrochemical performances. Secondly, the resulting dataset was used to train a deterministic ML model to implement fast bi-objective optimizations to identify optimal electrodes. Our results suggested a high amount of active material, combined with intermediate values of solid content in the slurry and calendering degree, to achieve the optimal electrodes.

Ruchirinkil Marreddy, Chaoyue Liu

The Neural Tangent Kernel (NTK) framework explains optimization in over-parameterized neural networks via approximately linearized dynamics, yielding exponential convergence guarantees. However, existing results are often overly pessimistic and do not match the fast training in practice, as they depend on the smallest NTK eigenvalue, which is typically extremely small in practice. In this work, we develop sharper convergence guarantees by characterizing the interaction between data labels and the NTK eigen-spectrum. We identify two key phenomena, Label-NTK alignment and Residual-NTK alignment, showing that projections of labels and residuals onto NTK eigenvectors scale with the corresponding eigenvalues. We provide empirical evidence and theoretical justification under mild data assumptions. Exploiting these alignment properties, we derive a refined convergence bound that depends on the full spectrum and closely matches practical training dynamics, significantly improving over classical worst-case results. We further obtain improved generalization bounds. Experiments on MLPs and CNNs across multiple datasets validate our theory.

Atul Sharma, Kavindu Herath, Saurabh Bagchi et al.

We introduce the Hubs and Spokes Learning (HSL) framework, a novel paradigm for collaborative machine learning that combines the strengths of Federated Learning (FL) and Decentralized Learning (P2PL). HSL employs a two-tier communication structure that avoids the single point of failure inherent in FL and outperforms the state-of-the-art P2PL framework, Epidemic Learning Local (ELL). At equal communication budgets (total edges), HSL achieves higher performance than ELL, while at significantly lower communication budgets, it can match ELL's performance. For instance, with only 400 edges, HSL reaches the same test accuracy that ELL achieves with 1000 edges for 100 peers (spokes) on CIFAR-10, demonstrating its suitability for resource-constrained systems. HSL also achieves stronger consensus among nodes after mixing, resulting in improved performance with fewer training rounds. We substantiate these claims through rigorous theoretical analyses and extensive experimental results, showcasing HSL's practicality for large-scale collaborative learning.

Jianing Zhang, Evan Chen, Chaoyue Liu et al.

Vertical Federated Learning (VFL) enables collaborative training with feature-partitioned data, yet remains vulnerable to privacy leakage through gradient transmissions. Standard differential privacy (DP) techniques such as DP-SGD are difficult to apply in this setting due to VFL's distributed nature and the high variance incurred by vector-valued noise. On the other hand, zeroth-order (ZO) optimization techniques can avoid explicit gradient exposure but lack formal privacy guarantees. In this work, we propose DPZV, the first ZO optimization framework for VFL that achieves tunable DP with performance guarantees. DPZV overcomes these limitations by injecting low-variance scalar noise at the server, enabling controllable privacy with reduced memory overhead. We conduct a comprehensive theoretical analysis showing that DPZV matches the convergence rate of first-order optimization methods while satisfying formal ($ε, δ$)-DP guarantees. Experiments on image and language benchmarks demonstrate that DPZV outperforms several baselines in terms of accuracy under a wide range of privacy constraints ($ε\le 10$), thereby elevating the privacy-utility tradeoff in VFL.

Evan Chen, Shiqiang Wang, Jianing Zhang et al.

In federated learning (FL), model training performance is strongly impacted by data heterogeneity across clients. Client-drift compensation methods have recently emerged as a solution to this issue, introducing correction terms into local model updates. To date, these methods have only been considered under stochastic gradient descent (SGD)-based model training, while modern FL frameworks also employ adaptive optimizers (e.g., Adam) for improved convergence. However, due to the complex interplay between first and second moments found in most adaptive optimization methods, naively injecting correction terms can lead to performance degradation in heterogeneous settings. In this work, we propose {\tt FAdamGC}, the first algorithm to integrate drift compensation into adaptive federated optimization. The key idea of {\tt FAdamGC} is injecting a pre-estimation correction term that aligns with the moment structure of adaptive methods. We provide a rigorous convergence analysis of our algorithm under non-convex settings, showing that {\tt FAdamGC} results in better rate and milder assumptions than naively porting SGD-based correction algorithms into adaptive optimizers. Our experimental results demonstrate that {\tt FAdamGC} consistently outperform existing methods in total communication and computation cost across varying levels of data heterogeneity, showing the efficacy of correcting gradient information in federated adaptive optimization.

Chaoyue Liu, Han Bi, Like Hui et al.

Nonlinear activation functions are widely recognized for enhancing the expressivity of neural networks, which is the primary reason for their widespread implementation. In this work, we focus on ReLU activation and reveal a novel and intriguing property of nonlinear activations. By comparing enabling and disabling the nonlinear activations in the neural network, we demonstrate their specific effects on wide neural networks: (a) better feature separation, i.e., a larger angle separation for similar data in the feature space of model gradient, and (b) better NTK conditioning, i.e., a smaller condition number of neural tangent kernel (NTK). Furthermore, we show that the network depth (i.e., with more nonlinear activation operations) further amplifies these effects; in addition, in the infinite-width-then-depth limit, all data are equally separated with a fixed angle in the model gradient feature space, regardless of how similar they are originally in the input space. Note that, without the nonlinear activation, i.e., in a linear neural network, the data separation remains the same as for the original inputs and NTK condition number is equivalent to the Gram matrix, regardless of the network depth. Due to the close connection between NTK condition number and convergence theories, our results imply that nonlinear activation helps to improve the worst-case convergence rates of gradient based methods.

Chaoyue Liu, Yulai Zhang

Hyper-parameter optimization is a crucial problem in machine learning as it aims to achieve the state-of-the-art performance in any model. Great efforts have been made in this field, such as random search, grid search, Bayesian optimization. In this paper, we model hyper-parameter optimization process as a Markov decision process, and tackle it with reinforcement learning. A novel hyper-parameter optimization method based on soft actor critic and hierarchical mixture regularization has been proposed. Experiments show that the proposed method can obtain better hyper-parameters in a shorter time.

Chaoyue Liu, Libin Zhu, Mikhail Belkin

The goal of this work is to shed light on the remarkable phenomenon of transition to linearity of certain neural networks as their width approaches infinity. We show that the transition to linearity of the model and, equivalently, constancy of the (neural) tangent kernel (NTK) result from the scaling properties of the norm of the Hessian matrix of the network as a function of the network width. We present a general framework for understanding the constancy of the tangent kernel via Hessian scaling applicable to the standard classes of neural networks. Our analysis provides a new perspective on the phenomenon of constant tangent kernel, which is different from the widely accepted "lazy training". Furthermore, we show that the transition to linearity is not a general property of wide neural networks and does not hold when the last layer of the network is non-linear. It is also not necessary for successful optimization by gradient descent.

Chaoyue Liu, Libin Zhu, Mikhail Belkin

The success of deep learning is due, to a large extent, to the remarkable effectiveness of gradient-based optimization methods applied to large neural networks. The purpose of this work is to propose a modern view and a general mathematical framework for loss landscapes and efficient optimization in over-parameterized machine learning models and systems of non-linear equations, a setting that includes over-parameterized deep neural networks. Our starting observation is that optimization problems corresponding to such systems are generally not convex, even locally. We argue that instead they satisfy PL$^*$, a variant of the Polyak-Lojasiewicz condition on most (but not all) of the parameter space, which guarantees both the existence of solutions and efficient optimization by (stochastic) gradient descent (SGD/GD). The PL$^*$ condition of these systems is closely related to the condition number of the tangent kernel associated to a non-linear system showing how a PL$^*$-based non-linear theory parallels classical analyses of over-parameterized linear equations. We show that wide neural networks satisfy the PL$^*$ condition, which explains the (S)GD convergence to a global minimum. Finally we propose a relaxation of the PL$^*$ condition applicable to "almost" over-parameterized systems.

Chaoyue Liu, Mikhail Belkin

Nesterov SGD is widely used for training modern neural networks and other machine learning models. Yet, its advantages over SGD have not been theoretically clarified. Indeed, as we show in our paper, both theoretically and empirically, Nesterov SGD with any parameter selection does not in general provide acceleration over ordinary SGD. Furthermore, Nesterov SGD may diverge for step sizes that ensure convergence of ordinary SGD. This is in contrast to the classical results in the deterministic scenario, where the same step size ensures accelerated convergence of the Nesterov's method over optimal gradient descent. To address the non-acceleration issue, we introduce a compensation term to Nesterov SGD. The resulting algorithm, which we call MaSS, converges for same step sizes as SGD. We prove that MaSS obtains an accelerated convergence rates over SGD for any mini-batch size in the linear setting. For full batch, the convergence rate of MaSS matches the well-known accelerated rate of the Nesterov's method. We also analyze the practically important question of the dependence of the convergence rate and optimal hyper-parameters on the mini-batch size, demonstrating three distinct regimes: linear scaling, diminishing returns and saturation. Experimental evaluation of MaSS for several standard architectures of deep networks, including ResNet and convolutional networks, shows improved performance over SGD, Nesterov SGD and Adam.

Chaoyue Liu, Mikhail Belkin

Analyses of accelerated (momentum-based) gradient descent usually assume bounded condition number to obtain exponential convergence rates. However, in many real problems, e.g., kernel methods or deep neural networks, the condition number, even locally, can be unbounded, unknown or mis-estimated. This poses problems in both implementing and analyzing accelerated algorithms. In this paper, we address this issue by proposing parametrized accelerated methods by considering the condition number as a free parameter. We provide spectral-level analysis for several important accelerated algorithms, obtain explicit expressions and improve worst case convergence rates. Moreover, we show that those algorithm converge exponentially even when the condition number is unknown or mis-estimated.