Study on stochastic hypergradient computation for machine learning problems.
problem Efficient computation of hypergradients in machine learning models.
method Stochastic approximation schemes for hypergradient computation, focusing on empirical risk minimization.
result Bounds for the mean square error of hypergradient approximation under contraction assumptions.
Proposes glocal hypergradient estimation for hyperparameter optimization.
problem Combining reliability and efficiency in hyperparameter optimization.
method Uses Koopman operator theory to approximate global hypergradients from local ones.
result Achieves both reliability and efficiency in hyperparameter optimization.
Study compares methods for computing hypergradients in machine learning problems.
problem Computing exact hypergradients in machine learning is difficult.
method Investigates reverse mode iterative differentiation and approximate implicit differentiation methods.
result Unified analysis provides iteration complexity bounds and hierarchy of methods.
NHGD solves bilevel optimization problems with reduced computational time.
problem Solving bilevel optimization problems with high computational cost.
method Exploits statistical structure of inner optimization to use empirical Fisher matrix as Hessian surrogate, enabling parallel optimization and approximation.
result NHGD achieves error bounds and sample complexity guarantees matching state-of-the-art methods, with significantly reduced computational time.
Paper proposes AggITD for efficient federated hypergradient computation.
problem Computing hypergradient in federated settings is challenging due to distributed and nonlinear construction of global Hessian matrices.
method AggITD: a novel communication-efficient federated hypergradient estimator via aggregated iterative differentiation.
result AggITD achieves the same sample complexity as AID-based approaches but with fewer communication rounds, especially in heterogeneous data environments.
We study the problem of fitting task-specific learning rate schedules from the perspective of hyperparameter optimization, aiming at good generalization. We describe the structure of the gradient of a validation error w.r.t. the learning rate schedule -- the hypergradient. Based on this, we introduce MARTHE, a novel on…
One-pass optimisation for high-dimensional hyperparameters.
problem Efficient optimisation of hyperparameters in machine learning models.
method Approximate hypergradient-based optimisation for any continuous hyperparameter, requiring only one training episode.
result Competitive performance on various datasets without hyperparameter restarts.
First-order method solves stochastic bilevel optimization with linear constraints.
problem Stochastic bilevel optimization with linear constraints and noise.
method Developed a novel framework using gradient-based techniques and smoothed penalty functions.
result Achieved finite-time convergence guarantees for (δ,ε)-Goldstein stationary points. Semiparametric method removes bias in functional bilevel gradient estimation.
problem First-order bias in plug-in hypergradient when lower-level problem is nonparametric.
method Semiparametric debiasing theory based on efficient influence function leads to cross-fitted orthogonal hypergradient estimator.
result Asymptotic normality and uniform control over outer parameter established for the estimator.
Framework solves bilevel optimization on manifolds.
problem Solving bilevel optimization problems with manifold constraints.
method Hypergradient estimation strategies on manifolds, convergence and complexity analyses.
result Efficacy demonstrated through various applications.
New Hessian-free method improves bilevel optimization for meta-learning.
problem Efficiently solving bilevel optimization problems with limited second-order information.
method Proposes a new Hessian-free method that approximates the response Jacobian matrix via optimization path differences.
result Demonstrates superior performance on meta-learning tasks compared to baseline methods.
ROBOT framework solves regression without correspondence for large data and complex models.
problem Regression without known correspondence in large datasets.
method ROBOT framework reformulates regression as a continuous optimization problem and uses hypergradient approach.
result ROBOT achieves better performance than existing methods in linear and nonlinear regression tasks.
New method finds stationary points in bilevel optimization problems.
problem Solving nonconvex-strongly-convex bilevel optimization problems.
method Restarted Accelerated HyperGradient Descent (RAHGD) method.
result Achieves best-known theoretical guarantees for finding stationary points in bilevel optimization.
EvoGrad improves efficiency in meta-learning and hyperparameter optimization.
problem Efficiently compute hypergradients for larger network architectures.
method Uses evolutionary techniques to estimate hypergradients without second-order derivatives or longer computational graphs.
result Significant improvements in efficiency, enabling scaling to bigger architectures.
We introduce a general method for improving the convergence rate of gradient-based optimizers that is easy to implement and works well in practice. We demonstrate the effectiveness of the method in a range of optimization problems by applying it to stochastic gradient descent, stochastic gradient descent with Nesterov …
Working with any gradient-based machine learning algorithm involves the tedious task of tuning the optimizer's hyperparameters, such as its step size. Recent work has shown how the step size can itself be optimized alongside the model parameters by manually deriving expressions for "hypergradients" ahead of time. We sh…
Standard gradient descent methods are susceptible to a range of issues that can impede training, such as high correlations and different scaling in parameter space.These difficulties can be addressed by second-order approaches that apply a pre-conditioning matrix to the gradient to improve convergence. Unfortunately, s…
OSGM uses online learning to adapt stepsize for faster convergence.
problem Improving convergence rates of first-order methods.
method OSGM combines online learning and feedback functions to adjust stepsize.
result OSGM achieves convergence rates asymptotically no worse than optimal.
La-MAML improves fast online continual learning with a look-ahead approach.
problem Fast online continual learning with limited model capacity.
method Optimisation-based meta-learning with look-ahead and episodic memory.
result Superior performance on visual classification benchmarks.
This work speeds up hyperparameter selection for non-smooth convex models using implicit differentiation.
problem Optimizing hyperparameters of non-smooth convex models.
method Implicit differentiation of proximal gradient and coordinate descent methods.
result Implicit differentiation can speed up hyperparameter optimization, especially for non-smooth problems.
New methods solve complex optimization problems in machine learning.
problem Challenges in stochastic bilevel optimization with constraints and high variables.
method Inexact bilevel stochastic gradient methods for constrained and unconstrained lower-level problems.
result Comprehensive convergence theory for both unconstrained and constrained cases.
New algorithm achieves linear speedup in non-i.i.d. federated bilevel learning.
problem Linear speedup in convergence for non-i.i.d. datasets in federated bilevel optimization.
method Proposes FedMBO with a novel client sampling scheme for non-i.i.d. datasets.
result Achieves a convergence rate of O(1/√(nK) + 1/K + √n/K³/²).
CB-RL solves complex decision-making problems with contextual information and exogenous events.
problem Optimal policy in strategic decision-making problems that depend on environmental configuration and exogenous events.
method Contextual Bilevel Reinforcement Learning (CB-RL) with a stochastic Hyper Policy Gradient Descent (HPGD) algorithm.
result Demonstrated convergence and performance of the HPGD algorithm for reward shaping and tax design.
Solving a bilevel optimization problem is at the core of several machine learning problems such as hyperparameter tuning, data denoising, meta- and few-shot learning, and training-data poisoning. Different from simultaneous or multi-objective optimization, the steepest descent direction for minimizing the upper-level c…
The paper analyzes convergence rates of bilevel optimization algorithms and introduces a new stochastic algorithm.
problem Nonconvex-strongly-convex bilevel optimization problems in machine learning.
method Comprehensive convergence rate analysis for deterministic bilevel optimization using AID and ITD, and a novel stochastic algorithm stocBiO.
result Theoretical convergence rates for AID and ITD methods, and stocBiO's superior performance.
Fewer data weight updates lead to faster convergence in machine learning models.
problem Improving robustness of machine learning models through data mixing.
method Analyzing convergence behavior of data mixing with a finite number of inner steps.
result The optimal number of inner steps scales with the budget and type of gradients used.
Paper tackles bilevel optimization problems using penalty methods.
problem Unconstrained and constrained bilevel optimization problems with nonsmooth lower levels.
method Introduces first-order penalty methods and O(ε−4logε−1) and O(ε−7logε−1) operation complexities. result Establishes operation complexities for finding ε-KKT solutions. New method tackles bilevel optimization with polyhedral constraints.
problem Challenges in bilevel optimization with active-set changes and expensive Hessian inversions.
method Logarithmic barrier smoothing and proxy-gradient algorithm for differentiable approximation.
result Stationarity rates of O(K−2/3) in deterministic setting and O(K−2/5) under stochastic noise. A new stochastic method tackles bi-level optimization problems in deep learning.
problem Bi-level optimization problems in deep learning, including hyperparameter optimization and meta learning.
method Turning a BLO problem into a stochastic optimization, using SGLD MCMC and a recurrent algorithm to compute MC-estimated hypergradient.
result Our method is more robust to suboptimal inner optimization and non-unique inner minima, leading to more reliable solutions.
New algorithm tackles stochastic bilevel optimization under relaxed smoothness conditions.
problem Optimal algorithms for stochastic bilevel optimization under relaxed smoothness conditions.
method Introduces a novel fully single-loop and Hessian-inversion-free algorithmic framework for stochastic bilevel optimization.
result Demonstrates state-of-the-art oracle complexity results for multi-objective robust bilevel optimization.
SmoothFBO tackles non-stationary functional bilevel optimization.
problem Current FBO methods are limited to static offline settings and perform poorly in online, non-stationary scenarios.
method SmoothFBO introduces a time-smoothed stochastic hypergradient estimator with a window parameter to handle non-stationarity.
result SmoothFBO achieves sublinear regret and outperforms existing methods in non-stationary hyperparameter optimization and model-based reinforcement learning.
New method optimizes fairness in predictive models for continuous sensitive attributes.
problem Enforcing full statistical independence on continuous sensitive attributes is too restrictive.
method Functional bilevel optimization (FBO) and ITD algorithms.
result Achieves lowest or near-lowest fairness-accuracy regret on synthetic and real datasets.
This thesis analyzes and improves convergence rates of bilevel optimization algorithms in machine learning.
problem Convergence analysis and algorithm design for bilevel optimization in machine learning.
method Comprehensive convergence rate analysis for both problem-based and algorithm-based bilevel optimization formulations.
result First lower bounds and matching upper bounds for bilevel optimization, and new stochastic algorithms with lower complexity.
A new method solves bilevel optimization problems in competitive Markov games.
problem Capturing competitive structures in RL with multiple interacting policies.
method Penalty-augmented Nikaido-Isoda descent-ascent (PANDA) method.
result PANDA converges to stationary points without convexity assumptions.
The study provides conditions for approximating Riemannian manifolds with polyhedral metrics.
problem Approximating Riemannian manifolds with polyhedral metrics.
method Conditions on curvature tensors for Lipschitz and local polyhedral approximations.
result Conditions are sufficient for local polyhedral approximations, conjectured to be sufficient for global approximations.
Geometric Gaussian approximations capture any distribution.
problem Approximating complex probability distributions.
method Geometric Gaussian approximations through diffeomorphisms or exponential maps.
result Geometric Gaussian approximations are universal, capturing any distribution.
Method approximates Riemannian barycenter on manifolds.
problem Computing the exact Riemannian barycenter is computationally expensive.
method Uses under- and over-approximations of Riemannian distance to compute an approximate barycenter.
result Approximation method is more efficient than exact methods and steepest descent.
We consider in this paper the optimal approximations of convex univariate functions with feed-forward Relu neural networks. We are interested in the following question: what is the minimal approximation error given the number of approximating linear pieces? We establish the necessary and sufficient conditions and uniqu…
Efficiently reduces tensor ranks using mean-field approximation.
problem Low-rank approximation of non-negative tensors.
method Mean-field approximation of tensor rank reduction.
result Our algorithm achieves faster and competitive tensor rank reduction.
Study approximates unknown function levels with queries.
problem Approximating unknown function levels through sequential queries.
method Introduce Bisect and Approximate algorithms to reduce to local function approximation.
result Rate-optimal sample complexity guarantees for H{ö}lder functions.
We study sparse approximate solutions to convex optimization problems. It is known that in many engineering applications researchers are interested in an approximate solution of an optimization problem as a linear combination of elements from a given system of elements. There is an increasing interest in building such …
Softmax attention approximates complex functions and subsumes many known universal approximators.
problem Universal approximation of continuous sequence-to-sequence functions.
method Interpolation-based analysis of attention's internal mechanism, showing its ability to approximate ReLU functions.
result Softmax attention is a universal approximator for continuous sequence-to-sequence functions.
Improved matrix approximation using randomized algorithms.
problem Finding better approximations of given matrices.
method Randomized algorithms to compute (HT) as an improved approximation. result Computed (HT) provides a better approximation than given F∗. Deviation inequalities for stochastic approximation methods.
problem Establishing bounds on the deviation of stochastic approximation methods.
method Martingale approximation method for separately Lipschitz functions.
result Established various deviation inequalities for stochastic approximation by averaging and minimization.
Neural approximate computing gains enormous energy-efficiency at the cost of tolerable quality-loss. A neural approximator can map the input data to output while a classifier determines whether the input data are safe to approximate with quality guarantee. However, existing works cannot maximize the invocation of the a…
Approximate symmetries of geodesic equations on 2-spheres are studied. These are the symmetries of the perturbed geodesic equations which represent approximate path of a particle rather than exact path. After giving the exact symmetries of the geodesic equations, two different approaches to study the approximate symmet…
We are concerned with an approximation problem for a symmetric positive semidefinite matrix due to motivation from a class of nonlinear machine learning methods. We discuss an approximation approach that we call {matrix ridge approximation}. In particular, we define the matrix ridge approximation as an incomplete matri…
Transformers use ReLUs to approximate softmax efficiently.
problem Analyzing resource usage in softmax transformer models.
method Translating ReLU approximation results to softmax attention mechanisms.
result Economic resource bounds for softmax attention mechanisms.