A new perspective on deep neural network regularization using RKHS norms.
problem Improving deep neural network performance and robustness.
method Using the norm of a reproducing kernel Hilbert space (RKHS) for regularization, with practical approximations.
result Effective regularization strategies for deep neural networks, including new penalties and hybrid approaches.
Kernel methods are among the most popular techniques in machine learning. From a frequentist/discriminative perspective they play a central role in regularization theory as they provide a natural choice for the hypotheses space and the regularization functional through the notion of reproducing kernel Hilbert spaces. F…
This paper explains why ResNets generalize better than FFNets using neural tangent kernels.
problem Understanding why deep ResNets generalize better than deep FFNets.
method Using neural tangent kernels to compare the learnability of functions induced by the kernels of ResNets and FFNets.
result The kernel of ResNets does not exhibit degeneracy as depth increases, unlike FFNets.
Unified theory for adaptive image convolutions using metric perspectives.
problem Fixed kernels in convolutions limit adaptability in image processing.
method Metric perspective on images as 2D manifolds with local distances, proposing metric convolutions.
result Metric convolutions provide better generalisation and competitive performance.
The paper analyzes a kernel perspective for group invariant feature learning.
problem Signal classification with group invariant features.
method Random feature map based on invariance I-theory, Haar integration kernel.
result Non-linear random feature map approximates group invariant kernel uniformly.
Since their emergence in the 1990's, the support vector machine and the AdaBoost algorithm have spawned a wave of research in statistical machine learning. Much of this new research falls into one of two broad categories: kernel methods and ensemble methods. In this expository article, I discuss the main ideas behind t…
Kernel methods linked to feature subspaces and maximal correlation kernels.
problem Understanding kernel methods and their relationship to feature extraction.
method Established a correspondence between feature subspaces and kernels, introduced maximal correlation kernels, and demonstrated their optimality.
result Kernel SVM on maximal correlation kernel achieves minimum prediction error.
New kernel-based models improve on traditional neural methods in sequence modeling.
problem Sequence modeling challenges in natural language processing and neuroscience.
method Kernel-based recurrent neural networks and convolutional neural networks.
result Kernel-based models perform on par or better than traditional neural methods.
Researchers approximate conditional expectation operators using kernel methods.
problem Statistical approximation of conditional expectation operators under minimal assumptions.
method Modifying the domain of the operator, approximating it by Hilbert-Schmidt operators in a reproducing kernel Hilbert space.
result The nonparametric estimate of the operator converges to a specific limiting object.
The paper establishes concentration bounds for embeddings of generative models.
problem Analyzing statistical properties of generative models.
method High probability concentration bounds on sample vector embeddings using Data Kernel Perspective Space.
result Determines the number of samples needed for accurate approximation of generative model embeddings.
A new kernel improves Gaussian process performance for non-stationary data.
problem Poor prediction and uncertainty quantification with standard GPs.
method Study and comparison of non-stationary kernels, propose a new combined kernel.
result A new kernel outperforms existing stationary and non-stationary kernels.
Analyzes neural networks using spectral perspectives.
problem Understanding neural network initialization and training.
method Examines the Conjugate Kernel and Neural Tangent Kernel spectra.
result Lends insights into neural network initialization and training properties.
Optimizes kernel density ratios for better predictions and information measures.
problem Improving accuracy of kernel density estimates for density ratios.
method Derives an optimal weight function using calculus of variations.
result Reduces bias in kernel density estimates, leading to improved prediction posteriors and information-theoretic measures.
Cross-validation methods help learn dynamical systems from data.
problem Learning surrogate models for dynamical systems from limited data.
method Variants of cross-validation (Kernel Flows, MMD, Lyapunov exponents).
result Simple approaches for kernel selection in dynamical system emulators.
Abstract perspective on quadratic programming for optimal portfolio allocation.
problem Optimal allocation problems in long portfolio theory.
method Using maximum principles and distinguished boundaries in reproducing kernel Hilbert spaces.
result Support of an optimal distribution lies in a variety intersecting a distinguished boundary.
Investigates kernel regression rates without assuming polynomial eigenvalue decay.
problem Achieving minimax rates without strict assumptions on kernel eigenvalue decay.
method Examines kernel regularization methods under weak assumptions on eigenvalue decay.
result Achieves minimax convergence rates under less restrictive conditions.
New networks interpret kernel decompositions for signal analysis.
problem Mode decomposition in signal analysis.
method Programmable and interpretable regression networks using kernels and data.
result Near machine precision recovery of signal modes under regularity and separation assumptions.
Overparametrized neural networks can generalize well with proper regularization.
problem Generalization guarantee for noisy data in overparametrized neural networks.
method Nonparametric analysis of ℓ2-regularized GD trajectories. result Achieving minimax optimal rate of L2 estimation error with ℓ2 regularization. This paper improves neural network generalization by dynamically learning kernel parameters.
problem Improving neural network generalization and adaptability.
method Diagonal adaptive kernel model that learns kernel eigenvalues and output coefficients during training.
result The diagonal adaptive kernel model significantly improves generalization over fixed-kernel methods.
New kernels use sliced Wasserstein for better probability distribution learning.
problem Learning from probability distributions efficiently.
method Sliced Wasserstein kernels based on optimal transport distances.
result Improved performance in various machine learning tasks.
Develops RKHS framework for analyzing tree ensembles.
problem Analyzing the theoretical properties of tree ensembles.
method Reproducing Kernel Hilbert Spaces (RKHS) for tree ensembles.
result Characterizes Random Forest predictor as minimizer of a penalized empirical risk functional in RKHS.
Optimal Biweight kernel and computationally efficient Epanechnikov kernel for modal linear regression.
problem Finding the best kernel for modal linear regression.
method Refined analysis of asymptotic statistical behavior and IRLS algorithm convergence.
result Biweight kernel minimizes asymptotic mean squared error, Epanechnikov kernel guarantees IRLS convergence.
Random forests improve probability estimates through kernel regression.
problem Improving the principled approach to random forest probability estimation.
method Forge a connection between random forests and kernel regression, develop a proximity kernel model.
result Improves statistical footing of random forest probability estimation.
We study the mixtures of factorizing probability distributions represented as visible marginal distributions in stochastic layered networks. We take the perspective of kernel transitions of distributions, which gives a unified picture of distributed representations arising from Deep Belief Networks (DBN) and other netw…
Although operator-valued kernels have recently received increasing interest in various machine learning and functional data analysis problems such as multi-task learning or functional regression, little attention has been paid to the understanding of their associated feature spaces. In this paper, we explore the potent…
We study the problem of structured output learning from a regression perspective. We first provide a general formulation of the kernel dependency estimation (KDE) problem using operator-valued kernels. We show that some of the existing formulations of this problem are special cases of our framework. We then propose a c…
Kernel-based Bayesian filter for nonlinear systems using infinite-dimensional operators.
problem Modeling and predicting nonlinear dynamical systems.
method Functional Bayesian perspective, reproducing kernel Hilbert space, Gaussian kernel.
result Effective approximation and accurate results for nonlinear systems.
Kernel methods benefit from enforcing invariance, reducing generalization error.
problem Improving generalization in kernel methods.
method Function space perspective and feature averaging for invariance.
result Strict non-zero generalization benefit for kernel ridge regression with invariant targets.
New approach to supervised learning in RKHS and vvRKHS using C∗-algebras.
problem Traditional supervised learning in RKHS and vvRKHS.
method Generalizing supervised learning to RKHM using C∗-algebras. result Constructing RKHMs with enhanced representation power.
The paper proposes a learning-theoretic perspective on representation alignment.
problem Understanding how AI models' representations become aligned as they scale.
method Reviewing and connecting different notions of alignment, focusing on stitching.
result Relating properties of stitching to kernel alignment of representations.
Sparse Kernel Flows learns dynamical systems from data.
problem Learning dynamical systems from limited data.
method Sparse Kernel Flows: trains optimal kernel from a dictionary of kernels.
result Sparse Kernel Flows can learn from 132 chaotic systems.
Proposes DAK model for improved GP computations.
problem Challenges in high-dimensional GP layers in DKL.
method Additive structure and induced prior approximation for GP units.
result Outperforms state-of-the-art DKL methods in regression and classification.
Unified perspective on score matching and new estimators designed.
problem Infeasibility of maximum likelihood estimation in complex models.
method Minimum Stein discrepancy estimators, diffusion kernel Stein discrepancy (DKSD), diffusion score matching (DSM).
result Consistency, asymptotic normality, and robustness of DKSD and DSM estimators.
Improved text classification performance through conformal transformations of kernels.
problem Text document categorization in high-dimensional spaces.
method Introduced new Gaussian Cosine kernel and two conformal transformations.
result Conformal transformations significantly improve kernel performance, especially for sub-optimal kernels.
Modeling videos and image-sets as linear subspaces has proven beneficial for many visual recognition tasks. However, it also incurs challenges arising from the fact that linear subspaces do not obey Euclidean geometry, but lie on a special type of Riemannian manifolds known as Grassmannian. To leverage the techniques d…
Two-layer neural networks learn efficiently using kernel methods in mean-field analysis.
problem Feature learning ability of two-layer neural networks in the mean-field regime.
method Mean-field analysis through kernel methods, focusing on dynamics of the first layer's kernel.
result Two-layer neural networks can learn a union of multiple reproducing kernel Hilbert spaces more efficiently than kernel methods.
New measures link neural representation geometry to decoding ability.
problem Understanding how neural representations relate to decoding ability.
method Showed that popular similarity measures can be interpreted from a decoding perspective.
result Proved that measures like CKA and CCA quantify alignment between optimal linear readouts.
Study on deep neural networks using branching processes and Mehler's formula.
problem Understanding the mathematical role of activation functions in compositional neural networks.
method Connection between compositional kernels and branching processes via Mehler's formula; new random features algorithm.
result Explicit formulas for eigenvalues of compositional kernels quantify complexity.
This paper shows equivalence between SVGD and BBVI using kernel gradient flows.
problem Bayesian inference methods and their equivalence.
method Formalizes equivalence between SVGD and BBVI using kernel gradient flows.
result BBVI corresponds precisely to SVGD when using the neural tangent kernel.
New kernels for set optimization improve combinatorial optimization without jitter.
problem Optimizing with set inputs, especially in combinatorial problems.
method Developed new set kernels based on Reproducing Kernel Hilbert Space embeddings.
result Subclasses of Deep Embedding kernels are strictly positive definite, enabling combinatorial optimization.
This study connects Gaussian processes and RKHS, bridging two machine learning communities.
problem Understanding the relationship between Gaussian processes and RKHS.
method Examining connections and equivalences in regression, interpolation, and other topics.
result Established the equivalence between Gaussian Hilbert space and RKHS.
New approach to learning kernels from data using AIT principles.
problem Learning kernels from data in machine learning.
method Sparse Kernel Flows method based on AIT principles.
result Sparse Kernel Flows aligns with MDL principle and offers a robust theoretical foundation.
The neural tangent kernel equivalence theorem fails in practice.
problem Does the neural tangent kernel (NTK) equivalence theorem hold in practical neural network training?
method Rigorously derived NTK and conducted numerical experiments to evaluate the equivalence theorem.
result Adding a layer to a neural network and the corresponding updated NTK do not yield matching changes in predictor error.
This paper improves neural tangent kernels for better generalization and local elasticity.
problem Performance gap between neural tangent kernels and real-world neural networks.
method Introduces label-aware kernels using Hoeffding decomposition.
result Models trained with proposed kernels simulate NNs better in terms of generalization and local elasticity.
Interest in multioutput kernel methods is increasing, whether under the guise of multitask learning, multisensor networks or structured output data. From the Gaussian process perspective a multioutput Mercer kernel is a covariance function over correlated output functions. One way of constructing such kernels is based …
Quantum kernel improves probabilistic time series forecasting.
problem Quantifying uncertainty in probabilistic time series predictions.
method Integrates quantum kernel with Gaussian process regression.
result Quantum kernel enhances forecasting performance.
Paper develops Gaussian process for distributions using Wasserstein distances.
problem Forecasting Gaussian processes indexed by probability distributions.
method Developed positive definite kernels based on Wasserstein distances.
result Efficient forecasting of Gaussian processes indexed by distributions.
Flow Matching improves statistical guarantees through kernel density estimation.
problem Improving statistical guarantees for generative models.
method Connecting Flow Matching to kernel density estimation and verifying optimal rates of convergence.
result Flow Matching achieves optimal rates up to logarithmic factors for large networks and on lower-dimensional manifolds.