Optimization Science

Convex Optimization Theory and Applications

Development and analysis of polynomial-time algorithms for convex optimization problems with theoretical guarantees and practical implementations.

RESEARCH GAP FRONTIERS

Stochastic Gradient Descent Convergence Analysis

Theoretical investigation of convergence rates and variance reduction techniques for stochastic gradient methods in large-scale learning.

RESEARCH GAP FRONTIERS

Mixed-Integer Programming Algorithms

Development of branch-and-cut methods and advanced cutting plane techniques for solving combinatorial optimization problems.

RESEARCH GAP FRONTIERS

Distributed Optimization over Networks

Design of decentralized algorithms for multi-agent systems with communication constraints and privacy preservation.

RESEARCH GAP FRONTIERS

Nonconvex Optimization Landscape Analysis

Characterization of loss landscapes and convergence behavior of first-order methods for nonconvex objectives.

RESEARCH GAP FRONTIERS

Federated Learning Optimization

Development of communication-efficient distributed algorithms for collaborative machine learning across heterogeneous data sources.

RESEARCH GAP FRONTIERS

Bilevel Optimization Methods

Algorithmic approaches for nested optimization problems with applications to hyperparameter tuning and meta-learning.

RESEARCH GAP FRONTIERS

Conic Programming and Interior Point Methods

Theory and implementation of polynomial-time interior point algorithms for semidefinite and second-order cone programs.

RESEARCH GAP FRONTIERS

Variance Reduction in Stochastic Optimization

Design of SVRG, SAGA, and related techniques to improve convergence rates of stochastic first-order methods.

Online Optimization and Regret Analysis

Development of algorithms for sequential decision-making with adaptive guarantees and logarithmic regret bounds.

Derivative-Free Optimization Algorithms

Methods for optimizing functions without gradient information including Bayesian optimization and trust-region approaches.

Proximal Methods and Operator Splitting

Theory and applications of ADMM, Douglas-Rachford, and forward-backward splitting algorithms for composite problems.

Quantum Optimization Algorithms

Development of quantum-inspired and quantum-enhanced optimization methods with potential speedup over classical algorithms.

Robust Optimization under Uncertainty

Optimization approaches that guarantee feasibility against worst-case uncertainty sets with tractable computational complexity.

Portfolio Optimization and Financial Mathematics

Development of optimization techniques for asset allocation, risk management, and derivative pricing under realistic constraints.

Combinatorial Optimization and Approximation

Design and analysis of approximation algorithms for NP-hard problems including traveling salesman and facility location.

Neural Network Training Optimization

Analysis of optimization dynamics in deep learning including loss landscape properties and implicit regularization effects.

Semidefinite Programming Applications

Use of semidefinite relaxations for solving combinatorial problems and applications in control theory and quantum information.

Momentum-Based Methods and Acceleration

Theory and variants of Nesterov acceleration, momentum methods, and polyak averaging for convex and nonconvex settings.

Second-Order Optimization Methods

Development of Newton, quasi-Newton, and natural gradient methods with improved convergence guarantees and scalability.

Supply Chain Optimization

Large-scale optimization of inventory management, routing, and procurement decisions in complex supply networks.

Power Systems and Smart Grid Optimization

Real-time optimization of electricity generation, distribution, and demand with renewable integration and stability constraints.

Machine Learning with Optimization

Integration of optimization theory into supervised and unsupervised learning including structured prediction and kernel methods.

Constraint Handling in Evolutionary Algorithms

Methods for incorporating constraints in genetic algorithms and swarm optimization while maintaining population diversity.

Multi-Objective Optimization Algorithms

Development of algorithms for finding Pareto-optimal solutions with applications in engineering design and resource allocation.

Inverse Optimization and Machine Teaching

Recovery of underlying objective functions from observed decisions and learning cost structures from behavioral data.

Decentralized Machine Learning at Edge

Optimization algorithms for collaborative learning on edge devices with bandwidth constraints and privacy requirements.

Manifold Optimization Techniques

Development of Riemannian optimization methods for problems constrained to curved spaces like rotation matrices.

Online Convex Optimization Games

Analysis of learning dynamics in multiagent settings with applications to mechanism design and algorithmic game theory.

Sparse Recovery and Compressed Sensing

Optimization methods for reconstructing sparse signals from underdetermined linear measurements and iterative thresholding algorithms.

Topology Optimization and Shape Design

Level-set and density methods for optimizing material distribution to achieve target structural properties.

Frank-Wolfe and Conditional Gradient Methods

Analysis and variants of projection-free optimization algorithms for structured constraint sets.

Bandit Optimization and Sequential Selection

Algorithms for balancing exploration and exploitation in multi-armed bandit problems with performance guarantees.

Stochastic Variational Inference

Scalable Bayesian inference through stochastic optimization of variational objectives on large datasets.

Resource Allocation in Cloud Computing

Optimization of virtual machine placement and load balancing for minimizing latency and power consumption.

Differentially Private Optimization

Development of optimization algorithms with differential privacy guarantees for protecting individual data in learning systems.

Optimal Control and Trajectory Optimization

Methods for computing optimal control inputs and trajectories satisfying dynamic constraints and performance objectives.

Graph-Based Optimization and Message Passing

Factor graph models and belief propagation algorithms for distributed inference and optimization on graphical structures.

Nonsmooth Optimization Theory

Algorithms and convergence analysis for minimizing nonsmooth and nondifferentiable objectives using subgradient methods.

Scheduling and Resource Optimization

Optimization algorithms for job scheduling, task allocation, and resource management in distributed systems.

Zeroth-Order Optimization Methods

Function-value only optimization algorithms including grid search, evolutionary strategies, and gradient-free methods.

Sum-of-Squares and Polynomial Optimization

Semidefinite programming relaxations for polynomial optimization using sum-of-squares certificates and hierarchy methods.

Dynamic Programming and Optimal Substructure

Algorithmic techniques for solving problems with optimal substructure including reinforcement learning value iteration.

Metaheuristics and Nature-Inspired Algorithms

Development and analysis of particle swarm, ant colony, simulated annealing, and tabu search algorithms for hard problems.

Learning Rates and Adaptive Methods

Theory and design of adaptive learning rate schedules including Adam, Adagrad, and RMSprop variants.

Geometric Deep Learning Optimization

Optimization techniques for neural networks on non-Euclidean domains including graphs and manifolds.

Auction Design and Mechanism Optimization

Development of truthful mechanisms and optimized auction formats for resource allocation and pricing.

Time-Series Forecasting Optimization

Optimization of deep learning models for sequential prediction with applications to demand and price forecasting.

Hyperparameter Optimization and AutoML

Automated methods for tuning machine learning model hyperparameters including Bayesian optimization and multi-fidelity approaches.

Game Theory and Equilibrium Computation

Algorithms for computing Nash equilibria and other solution concepts in simultaneous and sequential games.

Asynchronous Distributed Optimization Algorithms

Development and analysis of optimization methods for decentralized systems with delayed gradient information and asynchronous communication patterns.

Variational Inequality Problems and Solutions

Study of monotone operators and variational inequalities with applications to game theory, equilibrium problems, and machine learning.

Frank-Wolfe Variants and Extensions

Advanced conditional gradient methods including projection-free optimization, stochastic variants, and applications to large-scale problems.

Primal-Dual Optimization Methods

Theory and algorithms for solving structured convex problems using simultaneous updates of primal and dual variables.

Zeroth-Order Black-Box Optimization

Gradient-free optimization techniques using only function evaluations for problems with unknown or unavailable derivatives.

Federated Multi-Task Learning Optimization

Optimization algorithms for collaborative learning across heterogeneous devices with personalized model objectives and communication constraints.

Compositional Optimization and Hierarchical Problems

Methods for optimizing composite objective functions and nested optimization problems arising in machine learning and control.

Adversarial Robustness in Optimization Training

Optimization techniques for training machine learning models resilient to adversarial perturbations and worst-case perturbations.

Augmented Lagrangian and Penalty Methods

Constraint handling techniques combining penalty functions with Lagrangian approaches for constrained optimization problems.

Optimization in Reproducing Kernel Hilbert Spaces

Theoretical and algorithmic foundations for optimization over infinite-dimensional spaces with kernel methods and functional analysis.

Byzantine-Resilient Distributed Optimization

Optimization algorithms tolerating faulty or malicious agents in distributed systems through robust aggregation mechanisms.

Stochastic Approximation and Robbins-Monro Methods

Foundational theory of iterative stochastic algorithms with applications to root-finding and parameter estimation problems.

Optimization Over Simplices and Polytopes

Specialized algorithms for problems constrained to probability simplices and other convex polytopes with combinatorial structure.

Operator Splitting and Alternating Direction Methods

Decomposition-based approaches for solving large-scale problems through alternating minimization of coupled objectives.

Stochastic Mirror Descent and Mirror Descent Methods

Generalized gradient descent variants using Bregman divergences for non-Euclidean geometries and structured constraints.

Optimization with Time-Varying Networks

Distributed optimization over time-varying communication graphs with applications to mobile networks and dynamic systems.

Trust Region Methods and Local Convergence

Advanced nonlinear programming techniques using quadratic approximations within restricted regions for local and global convergence.

Optimization in Hyperbolic Spaces and Riemannian Geometry

Optimization algorithms adapted to non-Euclidean geometries including hyperbolic geometry with applications to graph embeddings.

Coordinate Descent and Block Coordinate Methods

Optimization through sequential or randomized updates of subsets of variables with theoretical guarantees and convergence analysis.

Optimization for Imbalanced Data Classification

Algorithmic approaches to training machine learning models on datasets with severe class imbalance through specialized optimization objectives.

Optimization Under Partial Information and Feedback

Algorithms leveraging limited, corrupted, or delayed feedback for optimization in incomplete information settings.

Submodular Optimization and Greedy Algorithms

Theory and algorithms for optimizing submodular functions with applications to feature selection and combinatorial problems.

Optimization for Generative Adversarial Networks

Specialized optimization techniques addressing convergence challenges and training instabilities in adversarial min-max problems.

Natural Gradient Descent and Information Geometry

Optimization using Fisher information geometry for improved convergence in probabilistic models and information-theoretic settings.

Optimization for Sparse Neural Networks

Algorithms for training and optimizing pruned or sparsified neural networks with structured sparsity constraints.

Optimization with Orthogonality Constraints

Specialized methods for problems with orthogonal matrix constraints arising in principal component analysis and matrix factorization.

Catalyst Acceleration and Universal Methods

General acceleration frameworks that enhance convergence rates of arbitrary first-order methods without problem-dependent tuning.

Optimization for Reinforcement Learning Control

Policy gradient, actor-critic, and value-based optimization methods for sequential decision making and optimal control.

Majorization-Minimization and Expectation Maximization

Iterative optimization techniques using surrogate functions and probabilistic lower bounds for non-convex problems.

Optimization for Matrix Completion and Recovery

Algorithms for recovering low-rank matrices from incomplete observations using convex and non-convex optimization approaches.

Parallel and GPU-Accelerated Optimization Algorithms

Optimization methods designed for modern parallel computing architectures including GPU and specialized hardware acceleration.

Optimization with Curvature Information and Hessian Methods

Second and higher-order methods utilizing curvature information for improved convergence in smooth and structured problems.

Optimization for Vision Transformers and Attention

Specialized optimization techniques addressing training dynamics of transformer architectures and self-attention mechanisms.

Optimization with Inexact Gradients and Errors

Robust optimization algorithms tolerating biased or noisy gradient estimates with convergence guarantees under model errors.

Markov Chain Monte Carlo Optimization Sampling

Optimization through probabilistic sampling methods including simulated annealing and Gibbs sampling variants.

Optimization for Causal Inference and Discovery

Optimization approaches for learning causal relationships and directed acyclic graphs from observational data.

Optimization in Presence of Computational Noise

Algorithms robust to rounding errors, finite precision arithmetic, and hardware-induced noise in optimization.

Optimization for Continual and Lifelong Learning

Methods for sequential learning across multiple tasks while mitigating catastrophic forgetting through specialized optimization objectives.

Optimization Using Sketching and Dimensionality Reduction

Large-scale optimization leveraging random projections and sketch techniques to reduce memory and computational requirements.

Optimization for Graph Neural Network Training

Specialized optimization algorithms addressing challenges in training neural networks on graph-structured data with node dependencies.

Optimization Under Distribution Shift and Domain Adaptation

Optimization techniques for training models robust to distribution changes between training and deployment environments.

Optimization for Kernel Methods and Kernel Learning

Algorithms for optimizing over kernel matrices and learning kernel parameters in support vector machines and kernel methods.

Coordinate-Free and Geometric Optimization Methods

Optimization approaches independent of coordinate systems using differential geometry and coordinate-invariant formulations.

Optimization for Tensor Networks and Decomposition

Algorithms for optimizing tensor decompositions and tensor networks with applications to multi-way data analysis.

Optimization with Chaotic Dynamics and Escape Saddles

Theoretical study of how optimization dynamics escape saddle points and navigate loss landscapes using chaos and bifurcation theory.

Optimization for Label Noise and Weak Supervision

Methods for training models under noisy labels and weak supervision signals through robust loss formulations and curriculum learning.

Optimization in Wasserstein Spaces and Optimal Transport

Optimization over probability measures using optimal transport distances and Wasserstein geometry for generative models.

Optimization with Subsampled Hessian Information

Second-order methods using sketched or sampled Hessian matrices for computational efficiency in large-scale optimization.

Optimization for Graph Cuts and Image Segmentation

Specialized optimization algorithms for minimum cut problems and energy minimization in computer vision applications.

Optimization Under Memory and Communication Constraints

Algorithms optimized for bandwidth-limited and memory-constrained environments including mobile and edge computing scenarios.

Asynchronous Distributed Optimization Convergence

Studies convergence guarantees and communication-efficient algorithms for asynchronous optimization across heterogeneous distributed systems.

Composite Optimization and Proximal Splitting

Develops algorithms for minimizing composite functions combining smooth and nonsmooth terms using advanced proximal techniques.

Saddle Point Optimization Dynamics

Analyzes convergence behavior and acceleration strategies for min-max problems arising in adversarial training and game theory.

Stochastic Variance-Reduced Mirror Descent

Investigates mirror descent methods with variance reduction for optimization over non-Euclidean geometries.

Distributed First-Order Methods with Delays

Develops theoretical frameworks for gradient-based optimization with communication delays and asynchronous updates in networks.

Nonconvex Federated Optimization Privacy

Analyzes convergence and privacy-utility tradeoffs in federated learning with nonconvex objectives and differential privacy constraints.

Coordinate Descent for Large-Scale Learning

Develops randomized and accelerated coordinate descent methods for solving massive machine learning and statistical problems.

Variance-Aware Adaptive Learning Rates

Designs adaptive optimization methods that exploit variance information to achieve improved convergence guarantees.

Submodular Optimization and Greedy Algorithms

Studies approximation algorithms and hardness results for maximization of submodular functions with cardinality and matroid constraints.

Variational Inequality Methods and Applications

Develops algorithms for monotone variational inequalities with applications to min-max problems and equilibrium computation.

Projection-Free Conditional Gradient Methods

Extends Frank-Wolfe algorithms with advanced step-size strategies and variance reduction for structured constraints.

Zeroth-Order Bandit Feedback Optimization

Analyzes optimization algorithms using only noisy function evaluations without gradient information in adversarial settings.

Catalytic Gradient Methods Acceleration

Studies acceleration techniques using auxiliary variables and momentum for improving convergence rates of gradient methods.

Distributed Nonconvex Optimization Landscape

Characterizes stationary points and escape saddle point behavior in distributed nonconvex optimization algorithms.

Primal-Dual Algorithm Design and Analysis

Develops primal-dual methods for saddle point problems with applications to constrained optimization and learning.

Structured Sparsity and Group Regularization

Optimizes objectives with group sparsity patterns using proximal gradient and coordinate descent approaches.

Optimization Under Limited Feedback Information

Designs algorithms for sequential decision making with limited or delayed gradient and loss feedback.

Decentralized Consensus Optimization Networks

Develops gossip and consensus-based algorithms for solving optimization problems without central coordination.

Smoothing Techniques for Nonsmooth Problems

Studies smoothing approximations and Moreau envelopes to apply smooth optimization techniques to nonsmooth objectives.

Variance Reduction with Importance Sampling

Combines importance sampling with variance-reduced stochastic methods to accelerate convergence.

Extragradient Methods Monotone Operators

Analyzes extragradient and reflected gradient methods for monotone operators with modern acceleration techniques.

Sparse Optimization and Compressed Sensing

Designs recovery algorithms for sparse signals using iterative thresholding and greedy optimization methods.

Optimization on Riemannian Manifolds

Develops first and second-order methods for optimization problems constrained to smooth manifolds.

Gradient Compression Quantization Communication

Designs communication-efficient distributed optimization with gradient compression and quantization techniques.

Dual Decomposition and Subgradient Methods

Develops decomposition algorithms for large-scale structured problems using Lagrangian duality and subgradient methods.

Mirror Descent and Bregman Divergences

Studies mirror descent algorithms with Bregman divergences for optimization over convex sets.

Inexact Optimization Error Analysis

Analyzes convergence when solving subproblems inexactly in Newton and second-order methods.

Stochastic Heavy Ball and Polyak Momentum

Investigates momentum-based methods for convex and nonconvex problems with theoretical guarantees.

Gossip Algorithms and Consensus

Designs randomized gossip protocols for distributed averaging and optimization over time-varying networks.

Sketching Methods Dimensionality Reduction

Uses random sketching and sampling to reduce problem dimension while maintaining solution quality in optimization.

Proximal Policy Optimization Reinforcement

Develops trust-region and clipped optimization methods for stable reinforcement learning policy updates.

Dual Averaging and Online Prediction

Studies dual averaging methods for online convex optimization with application to sequential prediction.

Optimization with Coupled Constraints

Develops algorithms for optimization with constraints coupling multiple variables across distributed agents.

Fast Saddle Point Escape Methods

Designs second-order methods to efficiently escape saddle points in nonconvex optimization problems.

Stochastic Proximal Gradient Variants

Analyzes variants of stochastic proximal gradient methods including SAG, SAGA, and variance-reduced approaches.

Optimization with Switching Constraints

Studies optimization problems with discrete switching decisions and continuous control variables.

Nesterov Acceleration Beyond Convexity

Extends Nesterov acceleration techniques to nonconvex and structured optimization settings.

Optimization for Tensor Decomposition

Develops algorithms for low-rank tensor approximation using gradient descent and alternating optimization.

Forward-Backward Splitting Algorithms

Analyzes forward-backward splitting methods for composite optimization with smooth and nonsmooth components.

Optimization Under Concept Drift

Designs adaptive optimization algorithms for online learning with time-varying loss functions and data distributions.

Cubic Regularization Newton Methods

Studies cubic regularized Newton methods for finding approximate second-order stationary points efficiently.

Optimization for Matrix Factorization

Develops alternating and gradient-based methods for low-rank matrix factorization in recommendation systems.

Trust Region Methods Nonconvex

Analyzes trust region algorithms for nonconvex optimization with convergence to second-order stationary points.

Optimization with Imperfect Oracles

Studies optimization algorithms when gradient and function evaluations contain systematic or stochastic errors.

Quasi-Newton Methods and Secant Updates

Investigates BFGS, L-BFGS, and other secant methods for large-scale unconstrained and constrained optimization.

Optimization for Operator Equations

Develops iterative methods for solving operator equations and fixed-point problems in optimization.

Distributed Gradient Tracking Methods

Designs distributed algorithms using gradient tracking for consensus optimization over networks.

Optimization with Random Projections

Uses random projections to reduce dimensionality while solving optimization problems with performance guarantees.

Stochastic Optimization Generalization Bounds

Analyzes generalization error and sample complexity of stochastic optimization for statistical learning.

Optimization for Spectral Methods

Develops optimization algorithms for computing eigenvalues, eigenvectors, and singular value decompositions.

Composite Optimization with Structured Sparsity

Develops efficient algorithms for minimizing composite functions subject to structured sparsity constraints in high-dimensional settings.

Stochastic Variational Optimization Inequalities

Analyzes convergence rates and optimality conditions for stochastic optimization problems formulated as variational inequalities.

Byzantine-Robust Federated Optimization Methods

Designs and analyzes distributed optimization algorithms resilient to adversarial or faulty agents in federated settings.

Operator Splitting for Large-Scale Problems

Develops scalable operator splitting techniques for decomposing large-scale optimization problems into manageable subproblems.

Saddle Point Escape and Local Geometry

Analyzes how optimization algorithms navigate nonconvex landscapes by escaping saddle points and exploiting local geometric structure.

Federated Optimization with Personalization

Develops algorithms for federated learning that balance global model convergence with personalized local objectives.

Parametric Uncertainty in Robust Control

Optimizes control policies under parametric uncertainty constraints in dynamical systems using robust optimization frameworks.

Tensor Decomposition and Low-Rank Optimization

Develops optimization techniques for tensor decomposition problems with applications to multi-dimensional data analysis.

Conditional Value-at-Risk Optimization

Studies optimization under risk measures focusing on conditional value-at-risk minimization in financial and robust settings.

Sketching-Based Optimization for Massive Data

Develops memory-efficient optimization algorithms using random sketching techniques for datasets exceeding available memory.

Primal-Dual Methods with Adaptive Metrics

Analyzes convergence of primal-dual algorithms that adaptively adjust distance metrics based on problem geometry.

Lifted Representations for Combinatorial Problems

Studies optimization by lifting discrete problems into higher-dimensional spaces where relaxations provide tighter bounds.

Gradient Compression and Quantization Effects

Analyzes convergence guarantees when optimization algorithms use quantized or compressed gradient information.

Matrix Completion with Side Information

Develops optimization algorithms for matrix completion problems enhanced with auxiliary structural or contextual information.

Dual Decomposition for Network Problems

Studies dual decomposition techniques for solving large-scale network optimization problems in distributed fashion.

Stochastic Coordinate Descent Variance

Analyzes variance reduction and acceleration techniques for coordinate descent methods in stochastic optimization.

Nonconvex-Concave Minimax Optimization

Studies convergence to stationary points in nonconvex-concave minimax optimization relevant to adversarial training.

Constrained Optimization over Riemannian Manifolds

Develops optimization algorithms for constrained problems on Riemannian manifolds with geometric convergence analysis.

Accelerated Methods for Ill-Conditioned Problems

Designs acceleration techniques that account for worst-case conditioning and provide dimension-independent convergence rates.

Multi-Time-Scale Stochastic Approximation

Analyzes convergence of coupled stochastic approximation algorithms with different learning rate scales.

Optimization with Sample-Dependent Constraints

Develops algorithms for problems where feasible regions depend on empirical data with concentration guarantees.

Cooperative Multi-Agent Optimization Games

Studies distributed optimization algorithms for cooperative agents with local objectives and network communication constraints.

Variance-Reduced Policy Gradient Methods

Develops variance reduction techniques for policy gradient reinforcement learning with convergence guarantees.

Cutting Plane Methods for Structured Problems

Studies cutting plane and subgradient methods for structured convex optimization with application-specific oracles.

Time-Varying Network Optimization Algorithms

Analyzes convergence of distributed optimization algorithms over networks with time-varying topology and delays.

Variational Inference with Implicit Models

Develops optimization techniques for variational inference with implicit density models avoiding explicit likelihood computation.

Gradient Tracking and Distributed Control

Studies gradient tracking algorithms enabling exact consensus in distributed optimization without knowing true gradients.

Hyperparameter-Free Adaptive Algorithms

Develops optimization algorithms that achieve near-optimal convergence without requiring problem-dependent hyperparameter tuning.

Lifted Convex Relaxations Combinatorics

Studies hierarchies of convex relaxations for combinatorial optimization with computational complexity tradeoffs.

Optimization Under Long-Range Dependencies

Develops algorithms for optimization problems with long-range dependencies in sequential or structured data.

Gaussian Process Bandit Optimization

Studies Bayesian optimization using Gaussian process models for sequential decision making under uncertainty.

Nonsmooth Nonconvex Optimization Landscapes

Analyzes convergence and complexity of algorithms for nonsmooth nonconvex problems lacking differentiability.

Optimization with Exogenous Information Structure

Studies optimization problems where agents have asymmetric information with strategic or cooperative dynamics.

Partial Relaxation and Implicit Gradients

Develops optimization techniques using partial relaxations and implicit function theorem for implicit differentiation.

Convex Geometry and Optimization Complexity

Studies how geometric properties of convex sets determine fundamental complexity bounds for optimization algorithms.

Optimization with Latency and Communication Costs

Analyzes optimization algorithms accounting for network latency and communication overhead in distributed settings.

Hypergradient Optimization and Meta-Learning

Develops techniques for optimizing optimization hyperparameters through hypergradient methods and bilevel formulations.

Optimal Transport and Distributionally Robust

Studies optimization with optimal transport distance constraints for distributionally robust and uncertainty-aware objectives.

Coordinate-Wise Variance Reduction Acceleration

Analyzes acceleration through coordinate selection strategies combined with variance reduction in large-scale optimization.

Evolutionary Optimization with Recombination

Studies theoretical foundations and performance analysis of evolutionary algorithms using crossover and recombination operators.

Optimization-Based Physics-Informed Neural Networks

Develops optimization techniques for physics-informed neural networks balancing data fitting and physical constraint satisfaction.

Personalized Federated Risk Minimization

Studies algorithms for federated optimization where clients minimize personalized objectives with global collaboration.

Incremental Aggregated Gradient Methods

Analyzes incremental aggregated gradient algorithms processing samples in batches with diminishing communication.

Asynchronous Distributed Optimization with Delays

Research on convergence guarantees and algorithmic design for distributed optimization systems with heterogeneous communication delays and asynchronous updates across networked agents.

Constrained Markov Decision Process Optimization

Studies optimization algorithms for constrained MDPs with convergence guarantees to statistically optimal policies.

Zeroth-Order Federated Learning in Heterogeneous Settings

Development of gradient-free federated optimization algorithms that handle non-IID data distributions and computational heterogeneity across edge devices without full gradient information.

Warm-Starting and Transfer in Optimization

Develops techniques leveraging prior solutions or related problems to accelerate optimization convergence.

Operator Splitting for Large-Scale Machine Vision

Application of proximal splitting and alternating direction methods to solve large-scale image processing and computer vision problems with structured nonconvex regularization.

Stochastic Optimization Under Distribution Shift

Analysis and design of optimization algorithms that maintain convergence and generalization guarantees when training data distributions change over time or across domains.

Submodular Maximization and Approximations

Studies optimization algorithms for submodular function maximization with worst-case approximation ratio guarantees.