This paper presents a compact, matrix-based representation of neural networks in a self-contained tutorial fashion. Specifically, we develop neural networks as a composition of several vector-valued functions. Although neural networks are well-understood pictorially in terms of interconnected neurons, neural networks are mathematical nonlinear functions constructed by composing several vector-valued functions. Using basic results from linear algebra, we represent a neural network as an alternating sequence of linear maps and scalar nonlinear functions, also known as activation functions. The training of neural networks requires the minimization of a cost function, which in turn requires the computation of a gradient. Using basic multivariable calculus results, the cost gradient is also shown to be a function composed of a sequence of linear maps and nonlinear functions. In addition to the analytical gradient computation, we consider two gradient-free training methods and compare the three training methods in terms of convergence rate and prediction accuracy.

A recent goal in the theory of deep learning is to identify how neural networks can escape the "lazy training," or Neural Tangent Kernel (NTK) regime, where the network is coupled with its first order Taylor expansion at initialization. While the NTK is minimax optimal for learning dense polynomials (Ghorbani et al, 2021), it cannot learn features, and hence has poor sample complexity for learning many classes of functions including sparse polynomials. Recent works have thus aimed to identify settings where gradient based algorithms provably generalize better than the NTK. One such example is the "QuadNTK" approach of Bai and Lee (2020), which analyzes the second-order term in the Taylor expansion. Bai and Lee (2020) show that the second-order term can learn sparse polynomials efficiently; however, it sacrifices the ability to learn general dense polynomials. In this paper, we analyze how gradient descent on a two-layer neural network can escape the NTK regime by utilizing a spectral characterization of the NTK (Montanari and Zhong, 2020) and building on the QuadNTK approach. We first expand upon the spectral analysis to identify "good" directions in parameter space in which we can move without harming generalization. Next, we show that a wide two-layer neural network can jointly use the NTK and QuadNTK to fit target functions consisting of a dense low-degree term and a sparse high-degree term -- something neither the NTK nor the QuadNTK can do on their own. Finally, we construct a regularizer which encourages our parameter vector to move in the "good" directions, and show that gradient descent on the regularized loss will converge to a global minimizer, which also has low test error. This yields an end to end convergence and generalization guarantee with provable sample complexity improvement over both the NTK and QuadNTK on their own.

Deep neural networks (DNNs) for supervised learning can be viewed as a pipeline of the feature extractor (i.e., last hidden layer) and a linear classifier (i.e., output layer) that are trained jointly with stochastic gradient descent (SGD) on the loss function (e.g., cross-entropy). In each epoch, the true gradient of the loss function is estimated using a mini-batch sampled from the training set and model parameters are then updated with the mini-batch gradients. Although the latter provides an unbiased estimation of the former, they are subject to substantial variances derived from the size and number of sampled mini-batches, leading to noisy and jumpy updates. To stabilize such undesirable variance in estimating the true gradients, we propose In-Training Representation Alignment (ITRA) that explicitly aligns feature distributions of two different mini-batches with a matching loss in the SGD training process. We also provide a rigorous analysis of the desirable effects of the matching loss on feature representation learning: (1) extracting compact feature representation; (2) reducing over-adaption on mini-batches via an adaptive weighting mechanism; and (3) accommodating to multi-modalities. Finally, we conduct large-scale experiments on both image and text classifications to demonstrate its superior performance to the strong baselines.

As one of the central tasks in machine learning, regression finds lots of applications in different fields. An existing common practice for solving regression problems is the mean square error (MSE) minimization approach or its regularized variants which require prior knowledge about the models. Recently, Yi et al., proposed a mutual information based supervised learning framework where they introduced a label entropy regularization which does not require any prior knowledge. When applied to classification tasks and solved via a stochastic gradient descent (SGD) optimization algorithm, their approach achieved significant improvement over the commonly used cross entropy loss and its variants. However, they did not provide a theoretical convergence analysis of the SGD algorithm for the proposed formulation. Besides, applying the framework to regression tasks is nontrivial due to the potentially infinite support set of the label. In this paper, we investigate the regression under the mutual information based supervised learning framework. We first argue that the MSE minimization approach is equivalent to a conditional entropy learning problem, and then propose a mutual information learning formulation for solving regression problems by using a reparameterization technique. For the proposed formulation, we give the convergence analysis of the SGD algorithm for solving it in practice. Finally, we consider a multi-output regression data model where we derive the generalization performance lower bound in terms of the mutual information associated with the underlying data distribution. The result shows that the high dimensionality can be a bless instead of a curse, which is controlled by a threshold. We hope our work will serve as a good starting point for further research on the mutual information based regression.

Stein Variational Gradient Descent (SVGD) is a popular sampling algorithm used in various machine learning tasks. It is well known that SVGD arises from a discretization of the kernelized gradient flow of the Kullback-Leibler divergence $D_{KL}\left(\cdot\mid\pi\right)$, where $\pi$ is the target distribution. In this work, we propose to enhance SVGD via the introduction of importance weights, which leads to a new method for which we coin the name $\beta$-SVGD. In the continuous time and infinite particles regime, the time for this flow to converge to the equilibrium distribution $\pi$, quantified by the Stein Fisher information, depends on $\rho_0$ and $\pi$ very weakly. This is very different from the kernelized gradient flow of Kullback-Leibler divergence, whose time complexity depends on $D_{KL}\left(\rho_0\mid\pi\right)$. Under certain assumptions, we provide a descent lemma for the population limit $\beta$-SVGD, which covers the descent lemma for the population limit SVGD when $\beta\to 0$. We also illustrate the advantages of $\beta$-SVGD over SVGD by experiments.

Machine learning algorithms with empirical risk minimization are vulnerable under distributional shifts due to the greedy adoption of all the correlations found in training data. There is an emerging literature on tackling this problem by minimizing the worst-case risk over an uncertainty set. However, existing methods mostly construct ambiguity sets by treating all variables equally regardless of the stability of their correlations with the target, resulting in the overwhelmingly-large uncertainty set and low confidence of the learner. In this paper, we propose a novel Stable Adversarial Learning (SAL) algorithm that leverages heterogeneous data sources to construct a more practical uncertainty set and conduct differentiated robustness optimization, where covariates are differentiated according to the stability of their correlations with the target. We theoretically show that our method is tractable for stochastic gradient-based optimization and provide the performance guarantees for our method. Empirical studies on both simulation and real datasets validate the effectiveness of our method in terms of uniformly good performance across unknown distributional shifts.

K are sparsity promoting, such as the black-box adversarial attack [4], model agnostic methods for explaining machine learning models [37] and sparse cox regression [34]. Despite the low dimensional structure restricted by r and K, standard stochastic mirror descent methods [27] and the conditional gradient methods [19] have oracle complexity depending linearly on d and are not optimal for high dimensional problems. The gradient descent algorithm is dimensionality independent when the first-order information is available [38]. For black-box objective functions, stronger dependence of the oracle complexity on dimensionality is caused by the biased gradient estimation [21]. In [50], the authors have proposed a LASSO-based gradient estimator for zerothorder optimisation of unconstrained convex objective functions.

This paper proposes a Decentralized Stochastic Gradient Descent (DSGD) algorithm to solve distributed machine-learning tasks over wirelessly-connected systems, without the coordination of a base station. It combines local stochastic gradient descent steps with a Non-Coherent Over-The-Air (NCOTA) consensus scheme at the receivers, that enables concurrent transmissions by leveraging the waveform superposition properties of the wireless channels. With NCOTA, local optimization signals are mapped to a mixture of orthogonal preamble sequences and transmitted concurrently over the wireless channel under half-duplex constraints. Consensus is estimated by non-coherently combining the received signals with the preamble sequences and mitigating the impact of noise and fading via a consensus stepsize. NCOTA-DSGD operates without channel state information (typically used in over-the-air computation schemes for channel inversion) and leverages the channel pathloss to mix signals, without explicit knowledge of the mixing weights (typically known in consensus-based optimization). It is shown that, with a suitable tuning of decreasing consensus and learning stepsizes, the error (measured as Euclidean distance) between the local and globally optimum models vanishes with rate $\mathcal O(k^{-1/4})$ after $k$ iterations. NCOTA-DSGD is evaluated numerically by solving an image classification task on the MNIST dataset, cast as a regularized cross-entropy loss minimization. Numerical results depict faster convergence vis-\`a-vis running time than implementations of the classical DSGD algorithm over digital and analog orthogonal channels, when the number of learning devices is large, under stringent delay constraints.

Stochastic differential equations (SDEs) have been shown recently to well characterize the dynamics of training machine learning models with SGD. This provides two opportunities for better understanding the generalization behaviour of SGD through its SDE approximation. First, under the SDE characterization, SGD may be regarded as the full-batch gradient descent with Gaussian gradient noise. This allows the application of the generalization bounds developed by Xu & Raginsky (2017) to analyzing the generalization behaviour of SGD, resulting in upper bounds in terms of the mutual information between the training set and the training trajectory. Second, under mild assumptions, it is possible to obtain an estimate of the steady-state weight distribution of SDE. Using this estimate, we apply the PAC-Bayes-like information-theoretic bounds developed in both Xu & Raginsky (2017) and Negrea et al. (2019) to obtain generalization upper bounds in terms of the KL divergence between the steady-state weight distribution of SGD with respect to a prior distribution. Among various options, one may choose the prior as the steady-state weight distribution obtained by SGD on the same training set but with one example held out. In this case, the bound can be elegantly expressed using the influence function (Koh & Liang, 2017), which suggests that the generalization of the SGD is related to the stability of SGD. Various insights are presented along the development of these bounds, which are subsequently validated numerically.

Parallel tempering (PT), also known as replica exchange, is the go-to workhorse for simulations of multi-modal distributions. The key to the success of PT is to adopt efficient swap schemes. The popular deterministic even-odd (DEO) scheme exploits the non-reversibility property and has successfully reduced the communication cost from $O(P^2)$ to $O(P)$ given sufficiently many $P$ chains. However, such an innovation largely disappears in big data due to the limited chains and few bias-corrected swaps. To handle this issue, we generalize the DEO scheme to promote non-reversibility and propose a few solutions to tackle the underlying bias caused by the geometric stopping time. Notably, in big data scenarios, we obtain an appealing communication cost $O(P\log P)$ based on the optimal window size. In addition, we also adopt stochastic gradient descent (SGD) with large and constant learning rates as exploration kernels. Such a user-friendly nature enables us to conduct approximation tasks for complex posteriors without much tuning costs.