New findings show fixed-kernel discriminators are weaker than feature-learning ones.
problem Comparing performance of fixed-kernel and feature-learning discriminators.
method Using function classes F2 and F1, constructing pairs of distributions, and linking IPMs with sliced Wasserstein distances. result Fixed-kernel IPM and SD cannot discriminate certain distributions that feature-learning IPM and SD can.
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.
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.
Theory explains why neural nets better learn Calabi-Yau metrics.
problem Learning Calabi-Yau metrics with neural networks.
method Developed a theory of metric flows in neural network space.
result Finite-width neural networks learn Calabi-Yau metrics better than fixed kernel methods.
This paper proposes Deep Hyperalignment (DHA) as a regularized, deep extension, scalable Hyperalignment (HA) method, which is well-suited for applying functional alignment to fMRI datasets with nonlinearity, high-dimensionality (broad ROI), and a large number of subjects. Unlink previous methods, DHA is not limited by …
Bayesian neural networks explore rare fluctuations for better feature learning.
problem Understanding rare but dominant fluctuations in Bayesian neural networks.
method Large-deviation theory and joint optimization over predictors and internal kernels.
result Posterior rate function optimization reveals data-dependent kernel selection.
Develops a new theory for neural systems stability and width effects.
problem Stability and finite-width effects in deep neural systems.
method Gauge-covariant stochastic effective field theory using classical commuting fields.
result Predicts the edge of chaos and low-frequency spectral deformation.
The paper analyzes V-statistics and variance estimation under varying kernel sizes.
problem Analyzing V-statistics and variance estimation under varying kernel sizes. method Develops a general framework for asymptotics of V-statistics, reducing to U-statistics and providing a unified variance estimation method. result Demonstrates asymptotic normality of V-statistics when kernel size grows with sample size. Deep generative models can learn to generate realistic-looking images, but many of the most effective methods are adversarial and involve a saddlepoint optimization, which requires a careful balancing of training between a generator network and a critic network. Maximum mean discrepancy networks (MMD-nets) avoid this i…
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.
Wide neural networks can benefit from multi-task learning in their infinite-width limit.
problem The generalization behavior of wide neural networks in multi-task learning settings.
method Optimizing wide ReLU neural networks with L2-regularization promotes multi-task learning in the infinite-width limit.
result An exact quantitative characterization of multi-task learning in the infinite-width limit of wide ReLU neural networks.
HKConv learns hyperbolic features by aggregating kernel points.
problem Challenges in learning good hyperbolic representations using Euclidean operations.
method Proposes HKConv, a trainable hyperbolic convolution that correlates local features with kernel points and aggregates them.
result HKConv learns expressive local features according to hyperbolic geometry and enjoys equivariance to permutation and invariance to parallel transport.
A determinantal point process (DPP) is a probabilistic model of set diversity compactly parameterized by a positive semi-definite kernel matrix. To fit a DPP to a given task, we would like to learn the entries of its kernel matrix by maximizing the log-likelihood of the available data. However, log-likelihood is non-co…
Attention-only transformers learn from context via two stages of inference.
problem Learning from corrupted token sequences in minimal transformers.
method Two-stage empirical Bayes interpretation: kernel-weighted posterior mean and particle dynamics.
result Effective denoising without explicit noise schedules, showing posterior-mean recovery under asymptotic conditions.
The paper shows how multi-task learning in neural networks is similar to kernel regression and Hilbert spaces.
problem Understanding the solutions to multi-task shallow ReLU neural network learning problems.
method Analyzing the properties of solutions to multi-task shallow ReLU neural network learning problems, proving uniqueness and equivalence to minimum-norm interpolation problems in Hilbert spaces.
result The solutions to multi-task neural network interpolation problems are almost always unique and coincide with the solution to a minimum-norm interpolation problem in a Sobolev (Reproducing Kernel) Hilbert Space.
Infinite neural networks lack key flexibility, finite ones learn better.
problem Theoretical limitations of infinite neural networks and their inferior performance.
method Analytic results and empirical evidence on finite deep linear networks and SOTA architectures.
result Finite deep linear networks perform better and learn representations, unlike infinite networks.
Improved standard parameterization yields well-defined neural tangent kernel.
problem Extrapolation of standard parameterization to infinite width is problematic.
method Proposed an improved extrapolation of the standard parameterization.
result Improved standard parameterization yields similar accuracy to NTK parameterization but with better correspondence to finite width networks.
In this paper we introduce and analyze the learning scenario of \emph{coupled nonlinear dimensionality reduction}, which combines two major steps of machine learning pipeline: projection onto a manifold and subsequent supervised learning. First, we present new generalization bounds for this scenario and, second, we int…
KSDAgg combines multiple KSD tests to improve goodness-of-fit testing without splitting data.
problem Improving goodness-of-fit testing without data splitting.
method KSDAgg aggregates multiple KSD tests with different kernels to maximize power.
result KSDAgg achieves the smallest uniform separation rate of the collection, up to a logarithmic term.
Ad-SVGD optimizes kernel parameters for SVGD, improving inference performance.
problem Efficiently approximating posterior distributions in Bayesian inference.
method Adaptive kernel selection for SVGD dynamics.
result Ad-SVGD outperforms standard heuristics in various tasks.
Adaptive RBF-KAN improves KANs by dynamically adjusting kernel parameters.
problem Efficiently approximating multivariate functions using learnable univariate edge functions.
method Integrates LOOCV-based kernel scale estimation with adaptive kernel learning.
result Adaptive RBF-KAN outperforms fixed kernel KANs on various benchmark functions.
A novel approach uses an ensemble of Gaussian processes for robust and adaptive reinforcement learning.
problem Adaptive reinforcement learning in large or continuous state spaces.
method Online scalable (OS) approach with a weighted ensemble of Gaussian processes.
result The ensemble approach improves performance in adversarial settings.
A new method optimizes MMD test power by dynamically selecting kernels, overcoming traditional trade-offs.
problem Fixed kernels fail to distinguish certain distributions, leading to overfitting and variance collapse.
method Complexity-Penalized MMD (CP-MMD) criterion, derived from concentration inequality, optimizes kernel selection.
result CP-MMD maximizes true test power while ensuring unconditional Type-I validity, matching or exceeding state-of-the-art performance.
Study on how adversarial training affects neural network kernels and robustness.
problem Understanding and improving adversarial robustness in neural networks.
method Empirical study of the evolution of the empirical Neural Tangent Kernel (NTK) under standard and adversarial training.
result Adversarial training leads to a new kernel that provides robustness, even when non-robust training is performed on top of it.
Enhanced CNN kernels improve image classification accuracy.
problem Improving classification accuracy of CNNs.
method Local Average Pooling and random image patch representation.
result 89% accuracy on CIFAR-10, matching AlexNet performance.
Gradient-flow optimization is reinterpreted as a statistical inference problem.
problem Optimizing training duration and assessing model performance in deep learning.
method Develops a statistical framework for gradient-flow training, treating it as a random-effects model.
result Establishes asymptotic optimality for prediction and reduces reliance on validation splits.
Two-sample tests using MMD control type I error and achieve optimal power.
problem Developing reliable nonparametric two-sample tests for small sample sizes.
method Maximum Mean Discrepancy (MMD) for constructing novel nonparametric tests, proving non-asymptotic error control and optimality.
result MMDAgg test controls type I error and achieves minimax rate over Sobolev balls, outperforming other tests.