Paper shows how gradient concentration helps in learning from inexact data.
problem Learning from inexact and stochastic training data.
method Combines probabilistic gradient concentration with inexact optimization techniques.
result Derives sharp test error guarantees for learning.
Concerns about interpretability, computational resources, and principled inductive priors have motivated efforts to engineer sparse neural models for NLP tasks. If sparsity is important for NLP, might well-trained neural models naturally become roughly sparse? Using the Taxi-Euclidean norm to measure sparsity, we find …
The study characterizes heat flow and concentration on directed graphs with a lower Ricci curvature bound.
problem Understanding heat flow and concentration on directed graphs with a specific curvature bound.
method Characterization via gradient estimate and transportation inequality for the heat semigroup.
result Concentration of measure inequality for directed graphs with positive Ricci curvature.
Stochastic approximation algorithms show exponential progress bounds.
problem Analyzing the convergence of stochastic approximation algorithms.
method Developed geometric ergodicity proofs to establish exponential concentration bounds.
result Proved faster convergence rates for specific algorithms.
Quantile gradient boosted trees outperform other models in predicting NO2 concentration distributions.
problem Forecasting high NO2 concentration episodes for effective air quality management.
method Compared 10 probabilistic forecasting models for NO2 concentration prediction.
result Quantile gradient boosted trees model outperformed others in predicting NO2 concentration distributions.
Study Ricci-Deturck flow from rough metrics, proving short-time existence.
problem Short-time existence of Ricci-Deturck flow from rough metrics.
method Ricci-Deturck flow, bi-Lipschitz metrics, small gradient concentration.
result Proved short-time existence of Ricci-Deturck flow.
SGD converges to an invariant distribution with sub-Gaussian or sub-exponential properties.
problem Optimizing smooth and strongly convex objectives using SGD.
method Analysis through Markov chains, focusing on convergence and concentration properties.
result SGD iterates and their invariant limit distribution inherit sub-Gaussian or sub-exponential concentration properties.
We study the concentration of NTK for MLPs at EOC, proving finite-width approximation of gradient independence.
problem Understanding the concentration of Neural Tangent Kernel (NTK) for MLPs at the Edge of Chaos (EOC).
method Proved approximate gradient independence holds at finite width, using maximal inequalities to show NTK matrix concentrates around its infinitely wide limit.
result The NTK matrix of MLPs at EOC concentrates around its infinitely wide limit, requiring hidden layer widths to grow quadratically.
New algorithm improves heavy-tailed statistical estimation in streaming data.
problem Heavy-tailed statistical estimation in streaming data.
method Clipped stochastic gradient descent algorithm with improved analysis.
result Guarantees exponential concentration with O(1) batch size for mean estimation and linear regression. We consider unreliable distributed learning systems wherein the training data is kept confidential by external workers, and the learner has to interact closely with those workers to train a model. In particular, we assume that there exists a system adversary that can adaptively compromise some workers; the compromised …
Proves convergence of gradient Ricci shrinkers with uniform bounds.
problem Compactness and energy concentration in gradient Ricci shrinkers.
method Bubble-tree convergence and local energy analysis.
result No energy concentrates in neck regions, leading to a local diffeomorphism finiteness theorem.
Paper develops a new local convexity condition for non-isolated minima in non-convex optimization.
problem Lack of theory for non-isolated minima in non-convex optimization.
method Formulates a new local convexity condition and studies SGD convergence under this condition.
result Shows SGD can converge locally under the new condition.
Gradient descent dynamics in nonconvex models explained with universality.
problem Understanding long-time behavior of nonconvex gradient descent.
method Developed a state evolution system for tracking gradient descent iterates.
result Gradient descent iterates are approximately independent of data and strongly incoherent with feature vectors.
Softmax policy gradient achieves global optimality in wide neural networks with entropy regularization.
problem Optimizing softmax policies with neural networks in the mean-field regime.
method Modeling neural networks as Wasserstein gradient flows and proving global optimality of fixed points.
result Global optimality of softmax policy gradient in wide single hidden layer neural networks with entropy regularization.
We develop a new approach to learn the parameters of regression models with hidden variables. In a nutshell, we estimate the gradient of the regression function at a set of random points, and cluster the estimated gradients. The centers of the clusters are used as estimates for the parameters of hidden units. We justif…
Logit dynamics formula reveals self-regulation in softmax policy gradient methods.
problem Understanding the stability and convergence of softmax policy gradient methods.
method Deriving the exact formula for the L2 norm of the logit update vector.
result Logit update magnitudes are modulated by action probability and policy concentration.
Improved privacy-preserving methods for convex optimization with heavy-tailed data.
problem Privacy-preserving optimization of convex functions with heavy-tailed data.
method Developed algorithms for private mean estimation and convex optimization under concentrated differential privacy constraints.
result Achieved improved upper bounds on excess population risk for convex and strongly convex loss functions.
Recent years have seen increased interest in performance guarantees of gradient descent algorithms for non-convex optimization. A number of works have uncovered that gradient noise plays a critical role in the ability of gradient descent recursions to efficiently escape saddle-points and reach second-order stationary p…
For stationary harmonic maps between Riemannian manifolds, we provide a necessary and sufficient condition for the uniform interior and boundary gradient estimates in terms of the total energy of maps. We also show that if analytic target manifolds do not carry any harmonic S^2, then the singular sets of stationary map…
Although stochastic gradient descent (SGD) is a driving force behind the recent success of deep learning, our understanding of its dynamics in a high-dimensional parameter space is limited. In recent years, some researchers have used the stochasticity of minibatch gradients, or the signal-to-noise ratio, to better char…
Quantum neural networks converge to Gaussian processes as they grow.
problem Understanding the convergence of quantum neural networks to Gaussian processes.
method Analyzing Haar random unitary and orthogonal deep QNNs, considering input states, measurement observables, and non-independence of unitary matrix entries.
result Quantum neural networks outputs converge to Gaussian processes in the limit of large Hilbert space dimension.
New algorithm resists Byzantine attacks in distributed SGD for heterogeneous data.
problem Byzantine attacks in distributed SGD for heterogeneous data.
method Polynomial-time outlier-filtering for robust mean estimation, new matrix concentration result.
result Tolerates up to 25% Byzantine workers and achieves optimal convergence rates.
New method improves sampling efficiency in complex stochastic systems.
problem Sampling efficiency in nonconvex stochastic gradient cases.
method Reflection coupling for unadjusted generalized Hamiltonian Monte Carlo.
result Quantitative Gaussian concentration bounds and convergence rates established.
We study learning properties of accelerated gradient descent methods for linear least-squares in Hilbert spaces. We analyze the implicit regularization properties of Nesterov acceleration and a variant of heavy-ball in terms of corresponding learning error bounds. Our results show that acceleration can provides faster …
Paper develops a federated learning method to protect privacy without sacrificing model utility.
problem Privacy leakage in federated learning due to information exchange between edge devices and server.
method Combines local gradient perturbation, secure aggregation, and zCDP for privacy protection.
result Demonstrates superior trade-off between privacy and model utility through extensive experiments.
The Langevin Algorithm's stationary distribution is shown to be sub-exponential or sub-Gaussian under certain conditions.
problem Understanding the properties of the Langevin Algorithm's stationary distribution.
method Analysis using a rotation-invariant moment generating function (Bessel function) to study the stationary dynamics of the Langevin Algorithm.
result Concentration results for the Langevin Algorithm's stationary distribution πη are established, showing it is sub-exponential or sub-Gaussian under convex or strongly convex potential conditions. Deep networks can perfectly classify two low-dimensional manifolds on a sphere with large depth and width.
problem Binary classification of two low-dimensional submanifolds on a sphere.
method Analysis of a deep fully-connected neural network trained to separate two submanifolds of the unit sphere.
result Randomly-initialized gradient descent can perfectly classify the two manifolds with high probability when the network depth is large relative to certain geometric and statistical properties of the data.
Many policy gradient methods are variants of Actor-Critic (AC), where a value function (critic) is learned to facilitate updating the parameterized policy (actor). The update to the actor involves a log-likelihood update weighted by the action-values, with the addition of entropy regularization for soft variants. In th…
Study gradient flow of phase transitions with fixed contact angle.
problem Understanding phase transitions with fixed contact angle.
method Gradient flow of the Allen-Cahn equation with fixed boundary contact angle.
result Established interior and boundary convergence properties for solutions and energy measures.
We study the gradient flow of the L2−norm of the second fundamental form of smooth immersions of two-dimensional surfaces into compact Riemannian manifolds. By analogy with the results obtained for the Willmore flow in Riemannian manifolds, we prove lifespan estimates in terms of the L2−concentration of the secon…
This paper analyzes minibatch optimal transport distances and their applications.
problem Optimal transport distances are complex and impractical for large datasets.
method Extended analysis of minibatch optimal transport distances, focusing on various kernels and debiased functions.
result Minibatch optimal transport distances are unbiased estimators and have statistical and optimisation properties.
We define functionals generalising the Seiberg-Witten functional on closed spinc manifolds, involving higher order derivatives of the curvature form and spinor field. We then consider their associated gradient flows and, using a gauge fixing technique, are able to prove short time existence for the flows. We then pr…
We show that the gradient norm ∥∇f(x)∥ for x∼exp(−f(x)), where f is strongly convex and smooth, concentrates tightly around its mean. This removes a barrier in the prior state-of-the-art analysis for the well-studied Metropolized Hamiltonian Monte Carlo (HMC) algorithm for sampling from a strongly l…
Gradient descent trains both layers of a ReLU network to fit a linear model.
problem Training dynamics of a ReLU network to fit a linear target function.
method Jointly training both layers of a one-hidden-layer ReLU network in a realizable setting with Gaussian inputs and labels.
result Gradient descent from a small random initialization converges to a global minimizer at a linear rate with optimal sample complexity.
New Langevin dynamics samples from entropy-regularized optimal transport.
problem Sampling from entropy-regularized optimal transport.
method Introduced analogous diffusion dynamics constrained to Π(μ,ν). result Long-time limit is the unique solution of an entropic optimal transport problem.
Theory for algebraic data on categories via concentration structures.
problem Defining algebraic structures on categories.
method Introducing concentration structures and concentration monoids.
result Every group can be represented as a concentration monoid of a trivial category.
The paper explores how data geometry influences generalization in neural networks.
problem Understanding generalization in overparameterized neural networks.
method Theoretical exploration of overparametrized two-layer ReLU networks trained below the edge of stability.
result Generalization bounds adapt to the intrinsic dimension of data distributions and deteriorate as data concentrates towards the unit sphere.
Iterative algorithms, like gradient descent, are common tools for solving a variety of problems, such as model fitting. For this reason, there is interest in creating differentially private versions of them. However, their conversion to differentially private algorithms is often naive. For instance, a fixed number of i…
Optimal transport distances are powerful tools to compare probability distributions and have found many applications in machine learning. Yet their algorithmic complexity prevents their direct use on large scale datasets. To overcome this challenge, practitioners compute these distances on minibatches {\em i.e.} they a…
Polyak step size GD reaches final radius of convergence after log iterations.
problem Statistical and computational complexities of Polyak step size GD.
method Generalized smoothness and Lojasiewicz conditions, stability of gradients.
result Polyak step size GD reaches final statistical radius of convergence after logarithmic number of iterations.
Paper addresses concentration of distances for fractional quasi p-norms, identifying conditions for concentration and anti-concentration.
problem Understanding concentration of distances for fractional quasi p-norms in high dimensions.
method Analyzes conditions for concentration and anti-concentration of distances for fractional quasi p-norms.
result Identifies conditions for concentration and anti-concentration of fractional quasi p-norms, ruling out some approaches and specifying conditions for control.
Study Finsler metric measure manifolds' concentration properties.
problem Understanding concentration properties in Finsler metric measure manifolds.
method Established relationships with observable diameter, isoperimetric inequalities, and first eigenvalue.
result Derived a Cheng type upper bound estimate for the first closed eigenvalue.
A new metric GNQ audits LLMs for privacy risks during training.
problem Auditing LLMs for privacy risks during training is computationally hard.
method Gradient Uniqueness (GNQ) metric derived from gradient descent, BS-Ghost GNQ for efficiency.
result GNQ successfully predicts sequence extractability and reveals risk heterogeneity.
Byzantine-resilient federated learning with local iterations and robust mean estimation.
problem Byzantine clients disrupt federated learning with local iterations.
method Local SGD iterations, robust mean estimation, and matrix concentration result.
result Convergence analysis for strongly-convex and non-convex smooth objectives in heterogeneous data settings.
New method improves missing mass concentration bounds.
problem Missing mass concentration problem
method New method of estimating concentration of heterogenic sums
result Slightly improved state-of-the-art bounds
New algorithm robustly optimizes data streams with heavy-tailed or infinite variance samples.
problem Optimizing data streams with heavy-tailed or infinite variance samples.
method Gradient quantile clipping for SGD, leveraging Markov chain connections.
result Algorithm converges to a concentrated distribution with high probability bounds.
Surfaces in 3-manifolds concentrate at curvature critical points.
problem Understanding concentration of surfaces in 3-manifolds.
method Proving surfaces concentrate at critical points of scalar curvature.
result Simply connected H-surfaces concentrate at curvature critical points.
Sharp concentration bounds for i.i.d. variables.
problem Controlling the tail probabilities of independent variables.
method Extension of Sanov's theorem using large deviations and information theory.
result Matching concentration and anti-concentration bounds for i.i.d. samples of any size.