Global optimization algorithm finds sparse mixed membership matrix factorization's global optimum.
problem Sparse mixed membership matrix factorization problems with local optima.
method Derives a global optimization algorithm for sparse mixed membership matrix factorization.
result Guaranteed ε-global optimum across random initializations and multiple modes. Noise helps neural networks escape local optima.
problem Understanding the role of noise in neural network training.
method Perturbed gradient descent and noise annealing.
result Noise guarantees convergence to global optimum in polynomial time.
Community detection in graphs has been the subject of many algorithms. Recent methods want to optimize a modularity function which shows a maximum of relationships within communities and found a minimum of inter-community relations. these algorithms are applied to unipartite, multipartite and directed graphs. However, …
New findings on the max margin problem in neural networks.
problem Understanding the max margin problem in neural networks.
method Analyzing gradient flow and max margin problem in linear and ReLU networks.
result The KKT point is not always an optimum of the max margin problem.
Community detection using both graphs and social networks is the focus of many algorithms. Recent methods aimed at optimizing the so-called modularity function proceed by maximizing relations within communities while minimizing inter-community relations. However, given the NP-completeness of the problem, these algorith…
The Hidden Markov Model (HMM) is one of the mainstays of statistical modeling of discrete time series, with applications including speech recognition, computational biology, computer vision and econometrics. Estimating an HMM from its observation process is often addressed via the Baum-Welch algorithm, which is known t…
New method finds global optima in variational inference.
problem Uncertainty in finding global optima in variational inference.
method Deterministic optimization algorithm for variational inference.
result Always converges to globally optimal variational lower bound.
Efficient RNN algorithm guarantees convergence in online learning.
problem Online nonlinear regression with RNNs.
method First-order training algorithm with convergence guarantee.
result The algorithm converges to optimum network parameters.
Exact causal network discovery is polynomial for sparse networks.
problem Finding the optimal causal Bayesian network from data is computationally hard.
method Pruning the search space using network properties, combined with dynamic programming and shortest-path searches.
result Exact discovery is polynomial for sparse causal Bayesian networks.
Efficiently minimizes regret in non-convex games with gradient-based methods.
problem Computational intractability of standard regret minimization in non-convex games.
method Defining a new notion of regret and using gradient-based optimization methods.
result Achieves optimal regret, leading to convergence to equilibrium.
A new framework for performative prediction robust to distributional misspecification.
problem Performative prediction models can be influenced by their own predictions, leading to suboptimal outcomes.
method Introduces distributionally robust performative prediction (DRPO) to approximate the true performative optimum (PO) robustly.
result DRPO provides provable guarantees as a robust approximation to the true PO when the nominal distribution map is misspecified.
No-regret BO algorithm adapts hyperparameters to optimize unknown functions.
problem Misspecification of hyperparameters in BO leads to poor local optima.
method Adapts hyperparameters online to expand function class and converge to optimum.
result First provably no-regret BO algorithm with unknown hyperparameters.
SGLB boosts machine learning with Langevin diffusion for multimodal loss functions.
problem Dealing with multimodal loss functions in machine learning.
method Stochastic Gradient Langevin Boosting (SGLB) based on Langevin diffusion equation.
result SGLB guarantees global convergence for multimodal loss functions.
Inference problems in graphical models are often approximated by casting them as constrained optimization problems. Message passing algorithms, such as belief propagation, have previously been suggested as methods for solving these optimization problems. However, there are few convergence guarantees for such algorithms…
New algorithms improve likelihood of finding global optima in Bayesian inference.
problem Finding global optima in Bayesian inference is difficult due to nonconvexity.
method Developed two algorithms: consistent Laplace approximation (CLA) and consistent stochastic variational inference (CSVI).
result Both CSVI and CLA improve likelihood of obtaining global optima compared to standard methods.
We show that there are no spurious local minima in the non-convex factorized parametrization of low-rank matrix recovery from incoherent linear measurements. With noisy measurements we show all local minima are very close to a global optimum. Together with a curvature bound at saddle points, this yields a polynomial ti…
We propose a simple, scalable, and fast gradient descent algorithm to optimize a nonconvex objective for the rank minimization problem and a closely related family of semidefinite programs. With O(r3κ2nlogn) random measurements of a positive semidefinite n×n matrix of rank r and condition number κ…
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.
Ridge regression CV loss may have multiple local optima.
problem Can we globally optimize cross-validation loss in ridge regression?
method Analyzing quasiconvexity of CV loss in ridge regression.
result CV loss may fail to be quasiconvex and have multiple local optima.
This paper improves convergence guarantees for gradient clipping in deep learning.
problem Improving convergence guarantees for gradient clipping in deep learning models.
method Analyzes and provides precise convergence guarantees for arbitrary clipping thresholds.
result Shows tight convergence guarantees for clipped stochastic gradient descent.
Local search heuristics for non-convex optimizations are popular in applied machine learning. However, in general it is hard to guarantee that such algorithms even converge to a local minimum, due to the existence of complicated saddle point structures in high dimensions. Many functions have degenerate saddle points su…
The paper develops a method for analyzing tensor data with convergence and statistical guarantees.
problem Analysis of tensor data with convergence and statistical guarantees.
method Two-dimensional canonical correlation analysis extended to tensor-valued data using the higher-order power method.
result The method provides finite sample bounds and converges to the optimum under certain conditions.
Training neural networks is a challenging non-convex optimization problem, and backpropagation or gradient descent can get stuck in spurious local optima. We propose a novel algorithm based on tensor decomposition for guaranteed training of two-layer neural networks. We provide risk bounds for our proposed method, with…
The paper tackles finding stationary points in stochastic convex optimization problems.
problem Finding stationary points for stochastic convex optimization problems.
method The approach relies on dimension theory to decompose the graph of the subdifferential of a convex function, showing how stochastic sampling preserves 'pieces' of these graphs, and allowing effective application of proximal-point-like methods.
result The paper provides convergence guarantees for finding stationary points in stochastic convex optimization problems.
New method in Bayesian optimization finds optimal inputs knowing the optimal outputs.
problem Finding optimal inputs when the optimal outputs are known in advance.
method Transform Gaussian process surrogate using known optimum output; propose two acquisition functions.
result Our approaches give quantitatively better performance than standard BO methods.
Model shows screening for infectious disease is hard but Thompson sampling works well.
problem Optimal screening policy for infectious diseases is hard to find.
method Stochastic-control model with Thompson sampling for optimal performance.
result Thompson sampling provides optimal performance guarantees in screening for infectious diseases.
New algorithms reduce GCN computation complexity and improve convergence.
problem Reducing the computational complexity of GCNs by controlling the number of neighbors.
method Control variate based algorithms to sample an arbitrarily small neighbor size, proving convergence to a local optimum.
result Proved algorithms converge to a local optimum of GCN with a small neighbor size.
RWR converges to global optimum in certain settings.
problem Proving convergence of RWR to optimal policy.
method Iterative learning with return-weighted log-likelihood.
result RWR converges to global optimum under certain conditions.
New variational flows improve Monte Carlo and normalization tasks.
problem Intractable global optimum in expressive variational families.
method Constructing asymptotically exact variational flows from involutive MCMC kernels.
result Provable total variation convergence of new variational families.
Gradient descent converges linearly for deep linear networks under specific conditions.
problem Speed of convergence in gradient descent for deep linear neural networks.
method Analysis of gradient descent training for deep linear neural networks minimizing ℓ2 loss. result Gradient descent converges linearly under specific conditions on layer dimensions, initialization, and initial loss.
OPFython simplifies Optimum-Path Forest for Python users.
problem Lack of diversity and complexity in conventional classification algorithms.
method Develops a Python-based Optimum-Path Forest framework.
result OPFython provides a more friendly and faster prototyping environment.
Neural networks solve Knapsack problems with provable guarantees.
problem Solving the Knapsack Problem efficiently and with guarantees.
method Recurrent neural networks (RNNs) with rectified linear units applied iteratively to each item.
result An RNN of depth four and width proportional to the profit of an optimum solution finds optimal solutions.
New method trains neural networks globally optimal with linear convergence.
problem Training neural networks to achieve global optimum is challenging and non-convex.
method Nonlinear Spectral Method for Generalized Polynomial Neural Networks.
result Achieves global optimal training with linear convergence rate.
The pathwise coordinate optimization is one of the most important computational frameworks for high dimensional convex and nonconvex sparse learning problems. It differs from the classical coordinate optimization algorithms in three salient features: {\it warm start initialization}, {\it active set updating}, and {\it …
Optimum-statistical collaboration improves black-box optimization efficiency.
problem Improving black-box optimization efficiency through better statistical collaboration.
method Introducing optimum-statistical collaboration framework for hierarchical bandits-based optimization.
result Demonstrated improved regret bounds and better performance in experiments.
We consider the problem of predicting an outcome variable using p covariates that are measured on n independent observations, in the setting in which flexible and interpretable fits are desirable. We propose the fused lasso additive model (FLAM), in which each additive function is estimated to be piecewise constant…
Bayesian optimisation is improved by incorporating expert prior through space warping.
problem Cold start phase in expensive function optimisation.
method Prior distribution warps the search space around high probability regions of function optimum.
result Improves optimisation performance through acquisition agnostic approach.
SCI-PI solves scale invariant problems efficiently.
problem Solving scale invariant problems in optimization.
method Introduces SCI-PI and proves its convergence.
result SCI-PI achieves local linear convergence.
AGGLIO optimizes non-convex functions with local convexity guarantees.
problem Optimizing non-convex functions with local convexity.
method Stage-wise, graduated optimization technique for locally convex functions.
result Global convergence to the global optimum for non-convex and locally convex objectives.
Bayesian Optimization with a Prior for the Optimum (BOPrO) improves efficiency and accuracy.
problem Bayesian Optimization's standard priors are not intuitive for domain experts.
method BOPrO injects expert knowledge into the optimization process using priors about the optimum.
result BOPrO is 6.67x faster than state-of-the-art methods and achieves new state-of-the-art performance.
LeJEPA learns latent variables from nonlinear observations.
problem Learning latent variables from nonlinear observations.
method Proves linear identifiability of Gaussian latent distributions.
result Gaussian distribution uniquely guarantees linear identifiability.
Efficiently approximates fairness-accuracy trade-offs for diverse datasets.
problem Inherent trade-off between fairness and accuracy in machine learning models.
method You-Only-Train-Once (YOTO) framework for computationally efficient approximation.
result Robust methodology for auditing model fairness with statistical guarantees.
Novel method for high-dimensional BO using CMA to define local regions.
problem Challenges in applying BO to high-dimensional optimization problems.
method CMA strategy to learn search distribution and define local regions.
result Our method outperforms existing techniques on various benchmarks.
Develops an SSBO algorithm for global optimization of expensive models.
problem Global optimization of expensive black-box models.
method Asynchronous hybrid-criterion with interval reduction.
result Improves global search ability and local search efficiency.
Optimizes data collection for ranking and selection problems.
problem Identifying the best system from multiple solutions with limited data.
method Sequential sampling algorithm with MPB estimator and kernel ridge regression.
result OSAR achieves optimal sampling ratios almost surely in the limit.
Paper introduces an asynchronous optimization algorithm for parameter servers.
problem Solving optimization problems with asynchrony and general regularizers.
method Asynchronous incremental aggregated gradient algorithm in a parameter server framework.
result Established linear convergence rate and step-size choices for strong convex data loss.
We consider the scenario where one observes an outcome variable and sets of features from multiple assays, all measured on the same set of samples. One approach that has been proposed for dealing with this type of data is ``sparse multiple canonical correlation analysis'' (sparse mCCA). All of the current sparse mCCA t…
The problem of finding overlapping communities in networks has gained much attention recently. Optimization-based approaches use non-negative matrix factorization (NMF) or variants, but the global optimum cannot be provably attained in general. Model-based approaches, such as the popular mixed-membership stochastic blo…