Statistical inverse learning problems with random observations

We provide an overview of recent progress in statistical inverse problems with random experimental design, covering both linear and nonlinear inverse problems. Different regularization schemes have been studied to produce robust and stable solutions. We discuss recent results in spectral regularization methods and regularization by projection, exploring both approaches within the context of Hilbert scales and presenting new insights particularly in regularization by projection. Additionally, we overview recent advancements in regularization using convex penalties. Convergence rates are analyzed in terms of the sample size in a probabilistic sense, yielding minimax rates in both expectation and probability. To achieve these results, the structure of reproducing kernel Hilbert spaces is leveraged to establish minimax rates in the statistical learning setting. We detail the assumptions underpinning these key elements of our proofs. Finally, we demonstrate the application of these concepts to nonlinear inverse problems in pharmacokinetic/pharmacodynamic (PK/PD) models, where the task is to predict changes in drug concentrations in patients.

How many Neurons do we need? A refined Analysis for Shallow Networks trained with Gradient Descent

We analyze the generalization properties of two-layer neural networks in the neural tangent kernel (NTK) regime, trained with gradient descent (GD). For early stopped GD we derive fast rates of convergence that are known to be minimax optimal in the framework of non-parametric regression in reproducing kernel Hilbert spaces. On our way, we precisely keep track of the number of hidden neurons required for generalization and improve over existing results. We further show that the weights during training remain in a vicinity around initialization, the radius being dependent on structural assumptions such as degree of smoothness of the regression function and eigenvalue decay of the integral operator associated to the NTK.

Random feature approximation for general spectral methods

Random feature approximation is arguably one of the most popular techniques to speed up kernel methods in large scale algorithms and provides a theoretical approach to the analysis of deep neural networks. We analyze generalization properties for a large class of spectral regularization methods combined with random features, containing kernel methods with implicit regularization such as gradient descent or explicit methods like Tikhonov regularization. For our estimators we obtain optimal learning rates over regularity classes (even for classes that are not included in the reproducing kernel Hilbert space), which are defined through appropriate source conditions. This improves or completes previous results obtained in related settings for specific kernel algorithms.

Learning linear operators: Infinite-dimensional regression as a well-behaved non-compact inverse problem

We consider the problem of learning a linear operator $\theta$ between two Hilbert spaces from empirical observations, which we interpret as least squares regression in infinite dimensions. We show that this goal can be reformulated as an inverse problem for $\theta$ with the undesirable feature that its forward operator is generally non-compact (even if $\theta$ is assumed to be compact or of $p$-Schatten class). However, we prove that, in terms of spectral properties and regularisation theory, this inverse problem is equivalent to the known compact inverse problem associated with scalar response regression. Our framework allows for the elegant derivation of dimension-free rates for generic learning algorithms under H\"older-type source conditions. The proofs rely on the combination of techniques from kernel regression with recent results on concentration of measure for sub-exponential Hilbertian random variables. The obtained rates hold for a variety of practically-relevant scenarios in functional regression as well as nonlinear regression with operator-valued kernels and match those of classical kernel regression with scalar response.

Data splitting improves statistical performance in overparametrized regimes

While large training datasets generally offer improvement in model performance, the training process becomes computationally expensive and time consuming. Distributed learning is a common strategy to reduce the overall training time by exploiting multiple computing devices. Recently, it has been observed in the single machine setting that overparametrization is essential for benign overfitting in ridgeless regression in Hilbert spaces. We show that in this regime, data splitting has a regularizing effect, hence improving statistical performance and computational complexity at the same time. We further provide a unified framework that allows to analyze both the finite and infinite dimensional setting. We numerically demonstrate the effect of different model parameters.

From inexact optimization to learning via gradient concentration

Optimization was recently shown to control the inductive bias in a learning process, a property referred to as implicit, or iterative regularization. The estimator obtained iteratively minimizing the training error can generalise well with no need of further penalties or constraints. In this paper, we investigate this phenomenon in the context of linear models with smooth loss functions. In particular, we investigate and propose a proof technique combining ideas from inexact optimization and probability theory, specifically gradient concentration. The proof is easy to follow and allows to obtain sharp learning bounds. More generally, it highlights a way to develop optimization results into learning guarantees.

Stochastic Gradient Descent Meets Distribution Regression

Stochastic gradient descent (SGD) provides a simple and efficient way to solve a broad range of machine learning problems. Here, we focus on distribution regression (DR), involving two stages of sampling: Firstly, we regress from probability measures to real-valued responses. Secondly, we sample bags from these distributions for utilizing them to solve the overall regression problem. Recently, DR has been tackled by applying kernel ridge regression and the learning properties of this approach are well understood. However, nothing is known about the learning properties of SGD for two stage sampling problems. We fill this gap and provide theoretical guarantees for the performance of SGD for DR. Our bounds are optimal in a mini-max sense under standard assumptions.

Stochastic Gradient Descent in Hilbert Scales: Smoothness, Preconditioning and Earlier Stopping

Stochastic Gradient Descent (SGD) has become the method of choice for solving a broad range of machine learning problems. However, some of its learning properties are still not fully understood. We consider least squares learning in reproducing kernel Hilbert spaces (RKHSs) and extend the classical SGD analysis to a learning setting in Hilbert scales, including Sobolev spaces and Diffusion spaces on compact Riemannian manifolds. We show that even for well-specified models, violation of a traditional benchmark smoothness assumption has a tremendous effect on the learning rate. In addition, we show that for miss-specified models, preconditioning in an appropriate Hilbert scale helps to reduce the number of iterations, i.e. allowing for "earlier stopping".

Reproducing kernel Hilbert spaces on manifolds: Sobolev and Diffusion spaces

We study reproducing kernel Hilbert spaces (RKHS) on a Riemannian manifold. In particular, we discuss under which condition Sobolev spaces are RKHS and characterize their reproducing kernels. Further, we introduce and discuss a class of smoother RKHS that we call diffusion spaces. We illustrate the general results with a number of detailed examples.

Global Minima of DNNs: The Plenty Pantry

A common strategy to train deep neural networks (DNNs) is to use very large architectures and to train them until they (almost) achieve zero training error. Empirically observed good generalization performance on test data, even in the presence of lots of label noise, corroborate such a procedure. On the other hand, in statistical learning theory it is known that over-fitting models may lead to poor generalization properties, occurring in e.g. empirical risk minimization (ERM) over too large hypotheses classes. Inspired by this contradictory behavior, so-called interpolation methods have recently received much attention, leading to consistent and optimally learning methods for some local averaging schemes with zero training error. However, there is no theoretical analysis of interpolating ERM-like methods so far. We take a step in this direction by showing that for certain, large hypotheses classes, some interpolating ERMs enjoy very good statistical guarantees while others fail in the worst sense. Moreover, we show that the same phenomenon occurs for DNNs with zero training error and sufficiently large architectures.