Paper tackles adversarial attacks on A3C path finding, proposing Gradient Band-based Adversarial Training.
problem Adversarial attacks on A3C path finding.
method Gradient Band-based Adversarial Training with CDG method.
result Gradient Band-based Adversarial Training achieves high attack immunity.
New method optimizes multiple objectives using particle dynamics and gradient flow.
problem Optimizing multiple conflicting objectives in complex scenarios.
method Interacting particle method combining Langevin and birth-death dynamics with a dominance potential.
result Method effectively relocates dominated particles, improving Pareto optimality.
The paper characterizes neural network landscapes for gradient dominance and regularity.
problem Understanding the landscape of neural network loss functions.
method Characterization of gradient dominance and regularity conditions for neural networks.
result Explicit characterization of global minimizers and landscape properties for different neural network types.
SGD updates align with a low-rank subspace but do not lead to further loss reduction.
problem Understanding the training dynamics of deep neural networks, particularly the role of the dominant subspace.
method Exploring whether neural networks can be trained within the dominant subspace of the loss Hessian.
result SGD updates, when projected onto the dominant subspace, do not decrease the training loss further, suggesting spurious alignment.
Develops a new solver for optimizing with stochastic dominance constraints.
problem Optimizing with stochastic dominance constraints is computationally expensive and impractical.
method Introduces Light Stochastic Dominance Solver (light-SD) that uses Lagrangian properties and surrogate approximation.
result The light-SD solver demonstrates superior performance on various problems.
New algorithm improves convergence of gradient boosting trees.
problem Global convergence of Newton boosting in tabular machine learning.
method Introduces Gradient Regularized Newton Descent for GBDTs, proving linear convergence for smooth, strongly convex losses and O(k21) rate for general convex losses. result Achieves globally convergent second-order GBDT algorithm with rate matching first-order boosting.
GD outperforms ridge regression and SGD in linear regression problems.
problem Comparing the risks of GD, ridge regression, and SGD in linear regression problems.
method Instance-wise finite-sample risk analysis of GD, ridge regression, and SGD.
result GD outperforms ridge regression and is incomparable with SGD in some cases.
New method accelerates neural network training by focusing on flat directions.
problem Improving neural network training speed and stability.
method Bulk-SGD, interpolated gradient methods.
result Updates along the Dominant subspace can accelerate convergence but compromise stability.
GENIE balances domain-invariant feature learning and gradient alignment for improved DG performance.
problem Domain Generalization (DG) overfitting to domain-specific features
method GENIE (Generalization-ENhancing Iterative Equalizer) optimizer
result Prevents a small subset of parameters from dominating optimization, promoting domain-invariant feature learning
In this paper, we study and analyze the mini-batch version of StochAstic Recursive grAdient algoritHm (SARAH), a method employing the stochastic recursive gradient, for solving empirical loss minimization for the case of nonconvex losses. We provide a sublinear convergence rate (to stationary points) for general noncon…
A new RL method using SSD compares action uncertainties to manage aleatoric uncertainty.
problem Managing aleatoric uncertainty in RL environments.
method Distributional RL based on SSD, mapping to Wasserstein gradient flow.
result Optimal particle-based algorithm for SSD policy demonstrates better uncertainty balancing.
This paper tackles robust control of LQR systems with multiplicative noise using policy gradient methods.
problem Robustness in reinforcement learning control of complex systems with multiplicative noise.
method Policy gradient algorithms with gradient domination property for non-convex cost functions.
result Global convergence of policy gradient algorithms to the globally optimum control policy.
Looped Transformers learn to implement multi-step gradient descent for in-context learning.
problem Understanding the learnability of multi-step algorithms in multi-layer Transformers.
method Training weight-sharing looped Transformers for in-context linear regression, proving gradient dominance condition for convergence.
result Looped Transformers implement multi-step preconditioned gradient descent, converging to global minimizer.
Stochastic Gradient Descent phases explained for deep networks.
problem Understanding the different regimes of SGD in deep learning.
method Teacher-student perceptron model, phase diagram analysis.
result SGD phases separated by batch size B∗, scaling with training set size P. TOPPO improves PPO for MTRL by balancing critic gradients, outperforming SAC.
problem Critic-side gradient ill-conditioning in PPO for MTRL.
method Critic Balancing modules to improve gradient conditioning and balance task updates.
result TOPPO achieves stronger mean and tail-task performance than SAC-family and ARS-family baselines.
We show that asymptotically, completely asynchronous stochastic gradient procedures achieve optimal (even to constant factors) convergence rates for the solution of convex optimization problems under nearly the same conditions required for asymptotic optimality of standard stochastic gradient procedures. Roughly, the n…
The Knowledge Gradient (KG) policy was originally proposed for online ranking and selection problems but has recently been adapted for use in online decision making in general and multi-armed bandit problems (MABs) in particular. We study its use in a class of exponential family MABs and identify weaknesses, including …
This work ensures policy gradient methods converge to global optima for certain control problems.
problem Non-convex optimization challenges in policy gradient methods for complex control problems.
method Identifies structural properties ensuring non-convex objective functions have no suboptimal stationary points.
result Policy gradient methods converge to global optima under certain conditions, satisfying a Polyak-Lojasiewicz condition.
New convergence guarantees for learning with unknown nuisance parameters.
problem Learning problems with unknown nuisance parameters.
method Stochastic gradient optimization with Neyman orthogonality and approximately orthogonalized updates.
result Stochastic gradient algorithms can converge under conditions of nuisance parameters.
Method estimates M-matrices in graphical models with improved accuracy.
problem Estimating M-matrices as precision matrices in Gaussian graphical models.
method Adaptive multiple-stage estimation method solving weighted ℓ1-regularized problems.
result Method outperforms state-of-the-art methods in precision matrix estimation and graph edge identification.
We consider the least-squares regression problem and provide a detailed asymptotic analysis of the performance of averaged constant-step-size stochastic gradient descent (a.k.a. least-mean-squares). In the strongly-convex case, we provide an asymptotic expansion up to explicit exponentially decaying terms. Our analysis…
New method improves MMD estimation without convexity assumptions.
problem Lack of theoretical guarantees for MMD estimation algorithms.
method Preconditioned gradient descent (PGD) scheme for MMD optimization.
result PGD scheme converges globally under specific conditions.
We propose a stochastic gradient Markov chain Monte Carlo (SG-MCMC) algorithm for scalable inference in mixed-membership stochastic blockmodels (MMSB). Our algorithm is based on the stochastic gradient Riemannian Langevin sampler and achieves both faster speed and higher accuracy at every iteration than the current sta…
Graph neural networks are explained through energy gradient flow and framelet decomposition.
problem Understanding and improving graph neural networks.
method Viewing framelet-based models as gradient flows of energy, proposing a generalized energy via framelet decomposition.
result The proposed model leads to more flexible dynamics, enhancing graph neural networks.
Lazy neural networks are vulnerable to adversarial attacks.
problem Adversarial robustness issues in lazy training models.
method Extending recent work on adversarial examples to lazy training models.
result Over-parametrized neural networks that generalize well remain vulnerable to single-step gradient ascent attacks.
Improves stochastic gradient methods for faster convergence.
problem Low asymptotic convergence of stochastic gradient methods in nonconvex optimization.
method Predictive Local Smoothness (PLS) method to adaptively adjust learning rates based on local smoothness predictions.
result New variants of SGD, AccSGD, and AMSGrad achieve faster linear convergence.
Random Reshuffling outperforms Stochastic Gradient Descent in smooth convex optimization.
problem Theoretical limitations of Random Reshuffling in smooth convex optimization.
method Random Reshuffling (RR) as a variant of Shuffling Stochastic Gradient Descent (Shuffling SGD).
result Random Reshuffling (RR) dominates Stochastic Gradient Descent (SGD) in smooth convex optimization under any reasonable stepsize after any finite number of epochs.
Deep forecasting models show output heads significantly improve performance on fat-tailed financial returns.
problem Improving deep learning models for forecasting fat-tailed financial returns.
method Comparison of backbone architectures and output heads (point, Gaussian, Gaussian mixture) on S&P 500 monthly log-returns.
result Switching from point to Gaussian heads improves CRPS by about 1.3 percent, and from Gaussian to mixture adds another 2.4 percent.
BoostTransformer uses boosting to improve transformer efficiency and accuracy.
problem Heavy computational resources and hyperparameter tuning in transformer architectures.
method Augments transformers with boosting principles through subgrid token selection and importance-weighted sampling, incorporating a least square boosting objective directly into the pipeline.
result BoostTransformer demonstrates faster convergence and higher accuracy compared to standard transformers.
Paper removes bounded gradient assumption for SGD in nonconvex learning.
problem Existing theoretical results for SGD in nonconvex learning require uniform boundedness of gradients, which is hard to verify.
method Establishes sufficient conditions for SGD convergence without bounded gradient assumption.
result SGD achieves optimal convergence rates for nonconvex and gradient-dominated objectives.
SpecGD mitigates misalignment in phase retrieval models with anisotropic inputs.
problem Misalignment during gradient descent in phase retrieval models with anisotropic inputs.
method Spectral gradient descent modifies gradient updates to preserve directional information and remove spike amplification.
result SpecGD removes spike amplification, leading to stable alignment and accelerated noise contraction.
Proposes mGBDTs for learning hierarchical representations in gradient boosting decision trees.
problem Inability of gradient boosting decision trees to learn hierarchical representations.
method Introduces multi-layered GBDT forest (mGBDTs) with explicit emphasis on hierarchical learning.
result Jointly trained mGBDTs can learn hierarchical representations effectively without backpropagation.
Connected domination numbers found for plane triangulations up to 13 vertices.
problem Finding connected domination numbers for plane triangulations.
method Analyzing triangulations of up to 13 vertices and proving the difference between connected and regular domination numbers can be arbitrarily large.
result Connected domination numbers for triangulations up to 13 vertices and upper bound for larger triangulations.
Study on PG learning for LQ MFC problems with common noise, proving convergence and sample complexity.
problem Optimal policy learning in LQ MFC problems with common noise and entropy regularization.
method Comprehensive error analysis of PG algorithms in both model-based and model-free settings.
result Global linear convergence and sample complexity of PG algorithms in model-free setting.
A new Adamize method improves multi-objective recommender systems.
problem Improving recommendation systems with multiple conflicting objectives.
method Developed a multi-objective model-agnostic Adamize method that corrects and stabilizes gradients.
result Significant improvements in recommendation systems, measured by hypervolume, coverage, and spacing.
Information that is stored in an encrypted format is, by definition, usually not amenable to statistical analysis or machine learning methods. In this paper we present detailed analysis of coordinate and accelerated gradient descent algorithms which are capable of fitting least squares and penalised ridge regression mo…
Private adaptive methods improve on traditional SGD for convex optimization.
problem Differential privacy constraints in gradient optimization.
method Differentially private variants of SGD and AdaGrad with adaptive stepsizes and non-isotropic clipping.
result Private AdaGrad outperforms private SGD in high-dimensional problems.
Bayesian optimization algorithm with preference constraints on objectives.
problem Finding Pareto-optimal solutions with user-defined preference over objectives.
method A multi-objective Bayesian optimization algorithm that incorporates user-defined preference constraints on objectives. The algorithm selects Pareto-optimal points that satisfy these constraints and uses a new acquisition function based on expected improvement in dominated hypervolume (EHI).
result The algorithm efficiently explores the Pareto front satisfying user-defined preference constraints.
Wider neural networks have predominantly positive curvature, aiding optimization.
problem Understanding the convex behavior of deep neural networks with varying layer widths.
method Hessian decomposition and gradient analysis of over-parameterized networks.
result For wide networks, the Hessian is dominated by the positive component G, leading to positive curvature.
New algorithm identifies dominant arm with high probability.
problem Identifying the arm with the highest realized reward in multi-armed bandits.
method Dominance score criterion and joint mixing and recycling mechanism.
result Identifies the best dominant arm with nearly optimal sample complexity.
Improved stochastic gradient estimation for deep learning in high dimensions.
problem Inadmissibility of mini-batch gradients in high-dimensional settings.
method Stein-rule shrinkage applied to gradient computation.
result The proposed SR-Adam outperforms Adam in large-batch settings.
Manifolds can be dominated by hypersurfaces in a sphere.
problem Dominating manifolds with hypersurfaces.
method Proving any smooth, closed, oriented manifold can be dominated by a codimension 1 submanifold of the sphere.
result Any smooth, closed, oriented manifold can be dominated by a codimension 1 submanifold of the sphere.
New algorithm reduces variance in nonconvex optimization problems.
problem Finite-sum nonconvex optimization problems.
method Stochastic gradient descent with nested variance reduction.
result Converges to an ε-stationary point with improved complexity.
New method ranks multivariate distributions in SMOOP using q-dominance.
problem Lack of reliable methods to rank multivariate distributions in SMOOP.
method Introduces center-outward q-dominance and develops empirical test procedures.
result Proves q-dominance implies FSD and establishes a sample size threshold.
A new method for learning gradient flows from population dynamics.
problem Reconstructing population dynamics from limited data.
method Residual approach to enforce continuity equations, combining with data-fitting divergence.
result Demonstrated state-of-the-art performance across trajectory inference benchmarks.
New framework for ranking distributions using variable fractional parameters.
problem Ordering distributions with varying steepness and local non-concavities.
method Introducing a function γ:Ro[0,1] to replace the fixed parameter in fractional SD. result Enables ranking of a broader range of distributions and incorporates dynamic greediness.
We show that non-domination results for targets that are not dominated by products are stable under Cartesian products.
Minibatch SGD outperforms Local SGD in heterogeneous distributed learning.
problem Optimizing a combined convex objective with stochastic gradient estimates from different machines.
method Analysis of Minibatch SGD and Local SGD in a heterogeneous distributed setting.
result Minibatch SGD dominates Local SGD in the heterogeneous distributed setting.