Keyphrases
Optimization Problem
100%
Hessian
96%
Second-order Methods
80%
Oracle Complexity
70%
Two-layer
60%
Finite-sum Optimization
53%
Symmetry Breaking
50%
Iteration Complexity
46%
Finite-sum Problem
40%
Nonconvex Stochastic Optimization
40%
Smooth Optimization
40%
Symmetry Structure
40%
ReLU Networks
40%
Second-order Information
33%
First-order Optimization Algorithms
30%
ReLU Neural Networks
30%
Stochastic Gradient Descent
30%
Tight
30%
Stationary Point
30%
Finite Sum
30%
Target Network
30%
Variance Reduction
30%
Squared Loss
30%
First-order Methods
26%
Optimization Algorithm
26%
Individual Functions
26%
Number of Iterations
26%
Equivariant Bifurcation Theory
26%
Gradient Method
26%
Hidden Neurons
20%
Mathematics
Finite Sum
80%
Stochastics
66%
Worst Case
58%
Neural Network
55%
Symmetry Breaking
50%
Convex Function
45%
Equivariant
32%
Gradient-Based Method
30%
Eigenvalue
30%
Critical Point
30%
Lower and upper bounds
24%
Convergence Analysis
20%
Objective Function
20%
Lagrangian
20%
Polynomial
20%
Variance Reduction
20%
Representation Theory
20%
Stationary Point
18%
Bifurcation Theory
15%
Quadratic Function
15%
Step Size
13%
Dimensional Problem
13%
Hidden Neuron
12%
Loss Function
12%
Positive Constant
12%
Approximates
10%
Descent Direction
10%
Linear Operator
10%
Convergence Rate
10%
Coincides
10%
Computer Science
Optimization Problem
70%
Individual Function
60%
Learning System
50%
Machine Learning
50%
Second Order Information
40%
Gradient Computation
40%
Algorithmic Approach
40%
Gradient Descent
40%
Efficient Algorithm
40%
Dimensional Problem
40%
Communication Complexity
20%
Objective Function
20%
Mathematical Convergence
20%
Experimental Result
20%
Computational Power
20%
Statistical Data
20%
Fundamental Limit
20%
Induced Problem
20%
Variance Reduction
20%
Convergence Rate
20%
Optimization Algorithm
14%
Convex Optimization
10%
Fast Algorithm
10%
Learning Problem
10%
