Learning One-Hidden-Layer Neural Networks with Landscape Design
Explore optimization techniques for one-hidden-layer neural networks, focusing on landscape design and new objective functions to overcome traditional training challenges.
Mixed Integer Optimization in High Dimensions
Explore advanced techniques for solving high-dimensional mixed integer optimization problems, focusing on cutting-edge algorithms and theoretical foundations.
The Approximation Ratio of the k-Opt Heuristic for Euclidean TSP
Explore the k-Opt heuristic's approximation ratio for Euclidean TSP, improving upper bounds and providing new lower bounds for k ≥ 3, applicable to various p-norms.
LP-Norm Multiway Cut
Explores graph partitioning with lp-norm multiway cut, unifying min-sum and min-max variants. Discusses hardness, integrality gap, and approximation algorithm for this generalized problem.
Dual Principal Component Pursuit
Explore a novel approach to learning subspaces from corrupted data, tolerating high outlier ratios and dimensions. Discover algorithms and theoretical guarantees for this non-convex optimization problem.
Gradient Descent and Stochastic Gradient Descent
Explore gradient descent and stochastic gradient descent in deep neural networks, covering convergence rates, learning rate effects, and challenges in optimization algorithms.
The Traveling Salesman Problem - When Good Enough Beats Perfect
Explore heuristic approaches and optimization techniques for the Traveling Salesman Problem, including nearest neighbor, greedy, Christofides, local search, simulated annealing, and ant colony optimization.
Introduction to Optimization Algorithms to Compress Neural Networks
Explore optimization techniques for compressing neural networks, including pruning and quantization, to enable deployment on resource-constrained devices like smartphones.
Estimating Human Motion: Past, Present, and Future
Explore the evolution of human motion estimation techniques, from early computer vision approaches to modern 3D body modeling, with insights on challenges and future directions.
Inexact Accelerated High-order Proximal-point Methods in Convex Programming
New framework for accelerated methods in convex programming using approximations of high-order proximal points. Presents second-order method with O(k^(-4)) convergence rate, surpassing traditional complexity limits.
Interacting Particle Systems for Expectation Maximization in Latent Variable Models
Explore a novel interacting particle system for EM optimization in latent variable models, featuring Langevin diffusion analysis and non-asymptotic concentration bounds for maximum likelihood estimation.
Towards a Theory of Non-Commutative Optimization - Geodesic First and Second Order Methods for Moment Maps
Explore geodesic methods for non-commutative optimization, focusing on moment maps and their applications in mathematical theory and algorithms.
IPM and Nested Dissection for Planar and Separable LPs
Exploring efficient LP solvers using nested dissection in IPM framework for planar min-cost flow, multicommodity flow, low treewidth LPs, and separable LPs. Focuses on optimization techniques and algorithm design.
Fast and Accurate Deep Neural Networks Training
Exploring LARS and LAMB optimizers for scaling deep learning on supercomputers, enabling faster training and higher accuracy in AI applications across various industries.
Deep Learning IV - Advanced Topics in Neural Networks
Explore advanced deep learning concepts with NYU professor Joan Bruna, covering cutting-edge techniques and applications in neural networks and machine learning.
Deep Learning III - Advanced Topics in Neural Networks
Explore advanced deep learning concepts with NYU professor Joan Bruna, covering cutting-edge techniques and applications in neural networks and machine learning.
Bandits II - Advanced Algorithms and Applications
Explore advanced bandit algorithms and their applications in decision-making under uncertainty with Kevin Jamieson from the University of Washington.
Continuous-Time Quantum Walks for MAX-CUT Optimization
Explore quantum walks for MAX-CUT optimization, linking Hamiltonians to thermalization. Discover novel insights into unitary dynamics for combinatorial problems using multi-stage walks and Floquet systems.
Memory as a Lens to Understand Efficient Learning and Optimization
Exploring memory's role in optimization, discussing convergence rates of memory-limited algorithms vs. computationally expensive methods, and examining new problem structures to improve gradient descent variants.
Generative Techniques for Music AI
Explore diverse techniques for AI music generation, from symbolic AI to deep learning, covering optimization algorithms, complex systems, and statistical methods. Gain insights into key approaches and their applications.
Global Optimization in Wolfram Language: Solving Nonlinear and Nonconvex Problems
Explore advanced global optimization techniques in Wolfram Language for solving complex nonlinear and mixed-integer problems with real-world examples.