This work shows how penalising bias terms in norm regularisation leads to sparse solutions.
problem Understanding the relation between parameter norm regularization and the sparsity of neural network solutions.
method Analyzes one hidden ReLU layer networks with unidimensional data, showing the norm required for function representation and the importance of the bias term's norm.
result Penalising the bias terms in regularisation leads to sparse solutions, enforcing the uniqueness and sparsity of the minimal norm interpolator.
New IDS algorithm refines parameter norm bounds for better bandit performance.
problem Frequentist IDS requires tight norm bounds, which are often unavailable in practice.
method Iteratively refines a high-probability upper bound on true parameter norm using data.
result Regret bounds independent of assumed parameter norm, outperforming state-of-the-art algorithms.
Improved online PCA algorithm learns from evolving norm of parameter vector.
problem Discarding evolving norm in online PCA leads to suboptimal learning.
method Implicitly Normalized Online PCA (INO-PCA) removes unit-norm constraint.
result Parameter norm evolution leads to improved learning behavior.
We show that the log-likelihood of several probabilistic graphical models is Lipschitz continuous with respect to the lp-norm of the parameters. We discuss several implications of Lipschitz parametrization. We present an upper bound of the Kullback-Leibler divergence that allows understanding methods that penalize the …
The study analyzes robustness of estimators in linear models with adversarial errors.
problem Analyzing robustness of estimators in linear models with adversarial errors.
method Develops a general theory for minimum norm interpolating estimators and RERM in linear models without conditions on errors.
result Quantitative bound for the prediction error relating it to Rademacher complexity, norm of minimum norm interpolator of errors, and subdifferential size.
A Bayesian approach termed BAyesian Least Squares Optimization with Nonnegative L1-norm constraint (BALSON) is proposed. The error distribution of data fitting is described by Gaussian likelihood. The parameter distribution is assumed to be a Dirichlet distribution. With the Bayes rule, searching for the optimal parame…
We prove that the norm version of the adaptive stochastic gradient method (AdaGrad-Norm) achieves a linear convergence rate for a subset of either strongly convex functions or non-convex functions that satisfy the Polyak Lojasiewicz (PL) inequality. The paper introduces the notion of Restricted Uniform Inequality of Gr…
New method tightens sub-Gaussian concentration inequalities.
problem Estimating variance-type parameters of sub-Gaussian distributions.
method Using sub-Gaussian intrinsic moment norm to maximize normalized moments.
result Provides tighter sub-Gaussian concentration inequalities.
This technical report describes an efficient technique for computing the norm of the gradient of the loss function for a neural network with respect to its parameters. This gradient norm can be computed efficiently for every example.
PSiLON Net uses L1 weight normalization and 1-path-norm regularization for efficient learning and sparsity.
problem Efficient learning and sparsity in neural networks with limited data.
method PSiLON Net employs L1 weight normalization and 1-path-norm regularization to simplify the 1-path-norm and achieve efficient learning and near-sparse parameters. result PSiLON Net achieves reliable optimization and strong performance in the small data regime.
We model how Lipschitz continuity changes during neural network training.
problem Understanding how Lipschitz continuity evolves during training.
method We use a system of stochastic differential equations to capture the dynamics of Lipschitz continuity under SGD.
result We identify three factors driving the evolution of Lipschitz continuity: gradient flow projection, gradient noise, and Hessian projection.
The spectral k-support norm enjoys good estimation properties in low rank matrix learning problems, empirically outperforming the trace norm. Its unit ball is the convex hull of rank k matrices with unit Frobenius norm. In this paper we generalize the norm to the spectral (k,p)-support norm, whose additional para…
A l1-norm penalized orthogonal forward regression (l1-POFR) algorithm is proposed based on the concept of leaveone- out mean square error (LOOMSE). Firstly, a new l1-norm penalized cost function is defined in the constructed orthogonal space, and each orthogonal basis is associated with an individually tunable regulari…
A new method for learning function parameters in operators using data-adaptive RKHS.
problem Learning function parameters in operators with robustness to noise and numerical error.
method Data Adaptive RKHS Tikhonov Regularization (DARTR) method.
result DARTR leads to an accurate estimator robust to noise and numerical error, converging at a consistent rate as data refines.
Study shows how networks converge to minimum norm solutions with regularization.
problem Interpolating between known regions in shallow ReLU networks.
method Investigates empirical risk minimizers and weight decay regularizers.
result Empirical risk minimizers converge to minimum norm interpolants under specific conditions.
CNNs use a bottleneck structure to focus on a few frequencies, affecting function representation.
problem Understanding how CNNs focus on specific frequencies in their feature learning.
method Defined Convolution Bottleneck (CBN) structure, measured CBN rank, and analyzed parameter norms.
result Parameter norm scales with depth and CBN rank, and networks with optimal parameters exhibit this structure.
SGD-trained deep nets have bounds on their generalization error.
problem Bounding generalization error for deep neural networks trained by SGD.
method Combining dynamical control of parameter norms and Rademacher complexity estimates.
result Explicit bounds depend on loss trajectory, work for various architectures.
Muon dynamics study uses spectral Wasserstein flow for optimization stability.
problem Optimizing deep learning models with gradient normalization.
method Introduces Spectral Wasserstein distances for matrix flows, proving equivalence with Benamou--Brenier formulation.
result Gradient-flow interpretation of mean-field normalized training dynamics.
Unified analysis of parameter norms in overparameterized linear models, revealing scaling laws and thresholds.
problem Understanding the scaling of parameter norms in overparameterized linear models.
method Simple dual-ray analysis revealing competition between signal spike and bulk of null coordinates.
result Unified closed-form predictions for parameter norm scaling, including elbow and threshold laws.
Using the ℓ1-norm to regularize the estimation of the parameter vector of a linear model leads to an unstable estimator when covariates are highly correlated. In this paper, we introduce a new penalty function which takes into account the correlation of the design matrix to stabilize the estimation. This norm, ca…
This paper presents a normalization mechanism called Instance-Level Meta Normalization (ILM~Norm) to address a learning-to-normalize problem. ILM~Norm learns to predict the normalization parameters via both the feature feed-forward and the gradient back-propagation paths. ILM~Norm provides a meta normalization mechanis…
An elementary proof found for the double bubble problem in a specific norm.
problem Finding the shapes that minimize perimeter for given volumes in a specific norm.
method Direct comparison to a small family of parameterized sets for analysis.
result Existence of minimizing sets for any volume ratio parameter.
So-called sparse estimators arise in the context of model fitting, when one a priori assumes that only a few (unknown) model parameters deviate from zero. Sparsity constraints can be useful when the estimation problem is under-determined, i.e. when number of model parameters is much higher than the number of data point…
Deep networks learn sparse hierarchical features without CoD.
problem Overparameterized deep networks struggle with the curse of dimensionality.
method Norm-constrained neural networks for sparse compositional functions.
result Deep networks can learn sparse hierarchical features efficiently.
New bounds show current methods overestimate system parameter errors.
problem Current bounds overestimate parameter errors in system identification.
method Utilized asymptotic normality and second-order decomposition.
result Obtained finite-sample bounds matching optimal rates up to constants.
Subspace identification is a classical and very well studied problem in system identification. The problem was recently posed as a convex optimization problem via the nuclear norm relaxation. Inspired by robust PCA, we extend this framework to handle outliers. The proposed framework takes the form of a convex optimizat…
Improved 2-bit covariance estimator with reduced operator norm error and no tuning needed.
problem Improving 2-bit covariance estimation with reduced operator norm error and no tuning needed.
method Proposed a new 2-bit covariance matrix estimator using triangular dithering scales.
result Improved operator norm error rate that depends on effective rank of covariance matrix, closing theoretical gap.
GraN detects adversarial and misclassified examples efficiently.
problem Detecting adversarial and misclassified examples in deep neural networks.
method GraN uses the layer-wise norm of the DNN's gradient regarding the loss of the current input-output combination.
result GraN achieves state-of-the-art performance on numerous problem set-ups.
Hexagonal norm double bubble problem solved with minimal configurations.
problem Finding the optimal shapes for minimizing perimeter in hexagonal geometry.
method Elementary proof and geometric exclusions to simplify minimizer search.
result Existence of minimizing sets for volume ratio parameter α in (0,1].
We show generalisation error bounds for deep learning with two main improvements over the state of the art. (1) Our bounds have no explicit dependence on the number of classes except for logarithmic factors. This holds even when formulating the bounds in terms of the L2-norm of the weight matrices, where previous bo…
Recently theoretical guarantees have been obtained for matrix completion in the non-uniform sampling regime. In particular, if the sampling distribution aligns with the underlying matrix's leverage scores, then with high probability nuclear norm minimization will exactly recover the low rank matrix. In this article, we…
In this paper, we propose an unifying view of several recently proposed structured sparsity-inducing norms. We consider the situation of a model simultaneously (a) penalized by a set- function de ned on the support of the unknown parameter vector which represents prior knowledge on supports, and (b) regularized in Lp-n…
We propose a Generalized Dantzig Selector (GDS) for linear models, in which any norm encoding the parameter structure can be leveraged for estimation. We investigate both computational and statistical aspects of the GDS. Based on conjugate proximal operator, a flexible inexact ADMM framework is designed for solving GDS…
The trace norm is widely used in multi-task learning as it can discover low-rank structures among tasks in terms of model parameters. Nowadays, with the emerging of big datasets and the popularity of deep learning techniques, tensor trace norms have been used for deep multi-task models. However, existing tensor trace n…
A new framework uses matrix flows to unify frequentist and Bayesian approaches for sparse GGMs.
problem Challenges in studying conditional independence among many variables with few observations.
method General framework for variational inference with matrix-variate Normalizing Flow in Gaussian Graphical Models.
result Unified benefits of frequentist and Bayesian frameworks for sparse GGMs.
ResNets minimize circuit size for fitting data in HTMC regime.
problem Finding the simplest algorithm that fits data.
method Defining HTMC and ResNet norms to relate circuit size and function fitting.
result Minimizing ResNet norm is equivalent to finding a circuit with minimal nodes.
Proposes a new regression method using Lp-norms for non-Gaussian noise.
problem Non-Gaussian noise in residuals affects the performance of local least squares regression.
method Introduces local polynomial Lp-norm regression, replacing weighted least squares with weighted Lp-norm estimation. result Demonstrates superior performance over local least squares in one-dimensional data and higher dimensions.
Develops a new asymptotic efficiency theory for non-Euclidean parameter spaces.
problem Lack of a unified efficiency theory for non-Euclidean parameter spaces.
method Introduces a new theory for Riemannian manifolds with regularity conditions.
result Establishes efficiency bounds for non-Euclidean parameter spaces.
We propose ℓ1 norm regularized quadratic surface support vector machine models for binary classification in supervised learning. We establish their desired theoretical properties, including the existence and uniqueness of the optimal solution, reduction to the standard SVMs over (almost) linearly separable data s…
This study reveals the critical role of scale vectors in large language models, improving optimization and expressivity.
problem Understanding and optimizing the scale vectors in large language models.
method Systematic study of scale vectors from expressivity, optimization, and architectural perspectives; theoretical and empirical analysis of weight decay; proposing and evaluating improvements.
result Scale vectors improve optimization through a self-amplifying preconditioning effect and are beneficial for expressivity in certain architectures.
Study on neural networks' sample complexity with one hidden layer.
problem Understanding how sample complexity is affected by network architecture and norm constraints.
method Norm-based uniform convergence bounds for scalar-valued one-hidden-layer networks, focusing on spectral and Frobenius norms.
result Spectral norm control is insufficient for uniform convergence guarantees, but Frobenius norm control is sufficient, with conditions.
Advances robust principal component analysis with transformed ℓ1 regularization.
problem Recovering low-rank structures from noisy, partially observed data corrupted by sparse outliers.
method Proposes transformed ℓ1 (TL1) regularization to improve approximations of rank and ℓ0 functional.
result Achieves higher accuracy in estimating low-rank and sparse components compared to classical convex models, especially under non-uniform sampling schemes.
The paper predicts survival functions using random survival trees and concordance maximization.
problem Predicting conditional survival functions in right-censored data.
method The approach combines regression strategies with random survival trees and maximizes concordance.
result The proposed weighted predictor outperforms the usual survival cobra in terms of concordance.
New algorithm reduces regret from sqrt(T) to polylog(T) in stochastic contextual linear bandits.
problem Achieving logarithmic regret in stochastic contextual linear bandits.
method Low Regret Stochastic Contextual Bandits ( exttt{LR-SCB}) algorithm, exploiting stochastic contexts and parameter estimation.
result Logarithmic regret (polylog(T)) achieved, improving over sqrt(T) lower bound.
The paper examines curvature-dimension bounds on sub-Finsler Heisenberg groups.
problem Investigating synthetic curvature-dimension bounds in sub-Finsler Heisenberg groups.
method Study of measure contraction property (MCP) and curvature-dimension condition (CD).
result Sub-Finsler Heisenberg groups do not satisfy MCP or CD for any parameters.
Batch normalization (batch norm) is often used in an attempt to stabilize and accelerate training in deep neural networks. In many cases it indeed decreases the number of parameter updates required to achieve low training error. However, it also reduces robustness to small adversarial input perturbations and noise by d…
New insights into network generalization show learning rate affects both norm and sharpness.
problem Understanding the generalization of overparameterized networks.
method Empirical analysis and theoretical proof of the trade-off between norm and sharpness.
result Learning rate influences both norm and sharpness, neither alone minimizes generalization error.
There is growing body of learning problems for which it is natural to organize the parameters into matrix, so as to appropriately regularize the parameters under some matrix norm (in order to impose some more sophisticated prior knowledge). This work describes and analyzes a systematic method for constructing such matr…