The study explains how neural networks align their kernels to target functions during training.
problem Understanding how neural networks align their kernels to target functions during training.
method Theoretical analysis of kernel evolution in toy models and deep networks.
result Kernel alignment naturally emerges during training to accelerate convergence and improve generalization.
Kernel alignment measures the degree of similarity between two kernels. In this paper, inspired from kernel alignment, we propose a new Linear Discriminant Analysis (LDA) formulation, kernel alignment LDA (kaLDA). We first define two kernels, data kernel and class indicator kernel. The problem is to find a subspace to …
The study investigates kernel-target alignment in tree ensemble kernels.
problem The degree of kernel-target alignment affects the performance of tree ensemble kernels in kernel learning.
method Eigenanalysis of the kernel matrix and sensitivity analysis via landmark learning.
result Good performance of tree ensemble kernels is associated with strong kernel-target alignment.
MKA incorporates manifold geometry into kernel alignment for more robust representation comparison.
problem Inadequate accounting for manifold geometry in kernel alignment metrics.
method Derives a theoretical framework for Manifold Approximated Kernel Alignment (MKA).
result MKA provides a more robust foundation for measuring representations.
DHGAK aligns substructures for better graph kernel performance.
problem Limited performance of traditional graph kernels due to missing substructure similarities.
method Hierarchically aligns relational substructures in deep embedding space, assigning same feature maps in RKHS.
result DHGAK outperforms state-of-the-art graph kernels on various benchmarks.
TKRR improves KRR performance by aligning target functions with kernels.
problem Improving kernel ridge regression performance through target alignment.
method Focuses on truncated kernel ridge regression (TKRR) with an additional spectral truncation parameter.
result TKRR can achieve faster rates than full KRR, reaching parametric rates.
CKA with Gaussian RBF kernels converges linearly as bandwidth increases.
problem Understanding the behavior of CKA with large bandwidth Gaussian kernels.
method Analyzing the convergence of CKA based on Gaussian RBF kernels in the large-bandwidth limit.
result CKA based on Gaussian RBF kernels converges linearly as bandwidth increases.
Neural networks can learn kernel machines with a data-dependent kernel.
problem Can neural networks in the rich feature learning regime learn a kernel machine?
method Demonstrated silent alignment effect in neural networks, showing they can learn a kernel machine with a data-dependent kernel.
result Neural networks in the rich feature learning regime can learn a kernel machine with a data-dependent kernel due to silent alignment.
In this paper we introduce the deep kernelized autoencoder, a neural network model that allows an explicit approximation of (i) the mapping from an input space to an arbitrary, user-specified kernel space and (ii) the back-projection from such a kernel space to input space. The proposed method is based on traditional a…
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.
NTKs explain GNNs' alignment for graph prediction.
problem Understanding GNNs' alignment for graph prediction.
method Analyzing NTKs and alignment in GNNs, focusing on cross-covariance.
result Optimizing alignment in GNNs optimizes graph representation.
Regularization effect found in neural feature alignment.
problem Implicit regularization in deep learning models.
method Geometrical viewpoint and analysis of Rademacher complexity.
result Neural features align along task-relevant directions, leading to regularization.
We introduce a kernel method for manifold alignment (KEMA) and domain adaptation that can match an arbitrary number of data sources without needing corresponding pairs, just few labeled examples in all domains. KEMA has interesting properties: 1) it generalizes other manifold alignment methods, 2) it can align manifold…
Graph kernels are widely used for measuring the similarity between graphs. Many existing graph kernels, which focus on local patterns within graphs rather than their global properties, suffer from significant structure information loss when representing graphs. Some recent global graph kernels, which utilizes the align…
This paper uses MIO to select features for kernel SVM classification.
problem Feature selection for kernel SVM classification.
method Mixed-integer optimization (MIO) for feature subset selection.
result The MIO approach can often outperform linear-SVM-based methods in prediction performance.
String kernels are attractive data analysis tools for analyzing string data. Among them, alignment kernels are known for their high prediction accuracies in string classifications when tested in combination with SVM in various applications. However, alignment kernels have a crucial drawback in that they scale poorly du…
End-to-end kernel learning using generative RFFs for improved performance.
problem Improving kernel learning performance and generalization.
method Develops a generative network via RFFs to implicitly learn the kernel, followed by a linear classifier, jointly trained by ERM.
result Shows superior generalization performance over classical methods in real-world tasks.
New kernel interprets 3D anisotropic data with rotations and improved predictions.
problem Capturing rotated anisotropy in 3D spatial fields.
method Introduces a Lie-algebraic kernel with three principal length-scales and an explicit rotation.
result Posterior recovers rotated anisotropy and improves prediction over axis-aligned kernels.
Autoencoders learn data representations (codes) in such a way that the input is reproduced at the output of the network. However, it is not always clear what kind of properties of the input data need to be captured by the codes. Kernel machines have experienced great success by operating via inner-products in a theoret…
Expressiveness and generalization of deep models was recently addressed via the connection between neural networks (NNs) and kernel learning, where first-order dynamics of NN during a gradient-descent (GD) optimization were related to gradient similarity kernel, also known as Neural Tangent Kernel (NTK). In the majorit…
The paper explains generalization in kernel regression and deep neural networks using spectral bias and task-model alignment.
problem Understanding generalization in machine learning models, especially deep neural networks.
method Analytical expression for generalization error derived from statistical mechanics, applied to various kernels and data distributions.
result Spectral bias and task-model alignment explain generalization in kernel regression and deep neural networks.
Recent results on optimization and generalization properties of neural networks showed that in a simple two-layer network, the alignment of the labels to the eigenvectors of the corresponding Gram matrix determines the convergence of the optimization during training. Such analyses also provide upper bounds on the gener…
Can neural networks learn to compare graphs without feature engineering? In this paper, we show that it is possible to learn representations for graph similarity with neither domain knowledge nor supervision (i.e.\ feature engineering or labeled graphs). We propose Deep Divergence Graph Kernels, an unsupervised method …
Gradient descent reshapes the function space of neural networks.
problem Understanding how feature learning affects the function space of neural networks.
method Characterized the evolution of the feature space during training using a two-layer neural network.
result Gradient descent induces a data-adaptive deformation that selectively enhances signal-aligned directions.
The paper explores how kernel eigenalignments affect generalization in KRR.
problem Achieving robust generalization in kernel methods.
method Direct connection between generalization and matrix eigenvectors/eigenvalues, focusing on finite-sample settings.
result Strong generalization requires increasing eigenvector alignment, eigenvalue magnitude, or gaps between eigenvalues.
New learning rules for wide neural networks without backpropagation.
problem Training wide neural networks efficiently and without backpropagation.
method Input-weight alignment driven by gradient descent in the NTK regime.
result Biologically-motivated learning rules equivalent to backpropagation in wide networks.
Quantum kernel methods can lead to trivial models due to exponential concentration of kernel values.
problem Exponential concentration of quantum kernel values can lead to trivial models in QML.
method Analyzing the resources needed to accurately estimate quantum kernel values and identifying four sources of concentration.
result Quantum kernel values can be exponentially concentrated, leading to trivial models.
SimpleMKKM improves multi-kernel clustering efficiency.
problem Efficient multi-kernel clustering.
method Re-formulated minimization-maximization problem into a smooth minimization, solved with gradient descent.
result Outperforms state-of-the-art multi-kernel clustering alternatives.
TopoNTK kernel captures higher-order interactions in simplicial complexes.
problem Graph neural networks miss higher-order interactions in relational systems.
method Introduces TopoNTK, an infinite-width kernel for simplicial message passing.
result TopoNTK captures topology invisible to graph kernels, improving expressivity and interpretability.
NTK-SAP improves neural network pruning by aligning training dynamics.
problem Improving neural network pruning to reduce training time and memory.
method Prune connections based on the spectrum of the Neural Tangent Kernel (NTK), using multiple random weight realizations and random inputs.
result Empirically, NTK-SAP achieves better performance than all baselines on multiple datasets.
A new method to rescale ReLU neural networks based on path-lifting.
problem Lack of principled ways to leverage rescaling symmetries in ReLU neural networks.
method Introduces a geometrically motivated criterion to rescale neural network parameters, aligning a kernel in the path-lifting space with a chosen reference.
result Proposed method can speed up training and aligns a kernel in the path-lifting space with a chosen reference.
Estimates KRR risk from training data for various kernels and hyperparameters.
problem Predicting the generalization error of Kernel Ridge Regression.
method Introduces SCT and KARE to approximate KRR risk from training data.
result KARE provides an excellent approximation of KRR risk and helps select good kernels.
The stable under iterated tessellation (STIT) process is a stochastic process that produces a recursive partition of space with cut directions drawn independently from a distribution over the sphere. The case of random axis-aligned cuts is known as the Mondrian process. Random forests and Laplace kernel approximations …
The ability of a human being to extrapolate previously gained knowledge to other domains inspired a new family of methods in machine learning called transfer learning. Transfer learning is often based on the assumption that objects in both target and source domains share some common feature and/or data space. In this p…
New kernels allow learning from non-separable data.
problem Learning from non-separable data.
method Introducing entangled kernels and a two-step algorithm.
result Efficient algorithm for learning entangled kernels.
The paper addresses rigid alignment of noisy patches, providing a polynomial time algorithm and convergence conditions.
problem Finding a rigid alignment of overlapping local views (patches) that minimizes alignment error in a noisy setting.
method Characterizes non-degeneracy based on kernel and positivity of a matrix, provides polynomial time algorithm for testing non-degeneracy, and uses Riemannian gradient descent for alignment.
result The algorithm converges locally linearly to a non-degenerate perfect alignment under certain conditions.
The success of kernel-based learning methods depend on the choice of kernel. Recently, kernel learning methods have been proposed that use data to select the most appropriate kernel, usually by combining a set of base kernels. We introduce a new algorithm for kernel learning that combines a {\em continuous set of base …
Uniform heat kernel bounds lead to synthetic Ricci curvature conditions for Lipschitz manifolds.
problem Establishing synthetic Ricci curvature conditions for Lipschitz manifolds.
method Uniform heat kernel bounds and synthetic Ricci curvature conditions.
result Uniform heat kernel bounds lead to synthetic Ricci curvature conditions for Lipschitz manifolds.
New method separates graph structure from node attributes to recover lost signal.
problem Standard representation learning on attributed graphs merges incompatible metric spaces, leading to geometrically flawed alignment.
method Custom variational autoencoder that separates manifold learning from structural alignment.
result Transforms geometric conflict into interpretable structural descriptor, uncovering connectivity patterns and anomalies.
Time series data analytics has been a problem of substantial interests for decades, and Dynamic Time Warping (DTW) has been the most widely adopted technique to measure dissimilarity between time series. A number of global-alignment kernels have since been proposed in the spirit of DTW to extend its use to kernel-based…
The paper analyzes how deep models memorize spurious features.
problem Understanding how deep models memorize spurious features in training data.
method Characterizes spurious feature memorization via model stability and feature alignment.
result Memorization of spurious features weakens as generalization capability increases.
We present a novel framework for kernel learning with sequential data of any kind, such as time series, sequences of graphs, or strings. Our approach is based on signature features which can be seen as an ordered variant of sample (cross-)moments; it allows to obtain a "sequentialized" version of any static kernel. The…
ResNets and their GP generalization align ideas via function space warping, revealing robust properties and connections to image registration.
problem Aligning abstract shapes (ideas) using neural networks.
method Introducing a generalization of ResNets as a GP, showing convergence to image registration variational algorithms, and revealing properties via a Hamiltonian interpretation.
result ResNets and their GP generalization align ideas via function space warping, revealing robust properties and connections to image registration.
Kernel methods are successful approaches for different machine learning problems. This success is mainly rooted in using feature maps and kernel matrices. Some methods rely on the eigenvalues/eigenvectors of the kernel matrix, while for other methods the spectral information can be used to estimate the excess risk. An …
Propose an XMSE-aware mixed estimator for EB that interpolates between ML and EB shrinkage.
problem Kernel-based EB estimation may be worse than ML when the kernel is poorly aligned with the true parameter.
method An XMSE-aware mixed estimator that interpolates between ML and EB shrinkage.
result Fixed-weight XMSE is a scalar quadratic, yielding a closed-form oracle mixing weight that is no worse than both ML and the base EB estimator at the XMSE scale.
Phylogenetic tree reconstruction is traditionally based on multiple sequence alignments (MSAs) and heavily depends on the validity of this information bottleneck. With increasing sequence divergence, the quality of MSAs decays quickly. Alignment-free methods, on the other hand, are based on abstract string comparisons …
Introduces greedy feature selection for classifier-dependent feature ranking.
problem Feature selection for classification tasks.
method Greedy feature selection, identifying the most important feature at each step based on the selected classifier.
result Theoretical and numerical benefits of greedy feature selection.
We consider the kernel completion problem with the presence of multiple views in the data. In this context the data samples can be fully missing in some views, creating missing columns and rows to the kernel matrices that are calculated individually for each view. We propose to solve the problem of completing the kerne…