• Publications
  • Influence
An Improved Convergence Analysis of Stochastic Variance-Reduced Policy Gradient
TLDR
We revisit the stochastic variance-reduced policy gradient (SVRPG) method proposed by Papini et al. (2018) for reinforcement learning. Expand
Sample Efficient Policy Gradient Methods with Recursive Variance Reduction
TLDR
We propose a novel policy gradient algorithm called SRVR-PG, which only requires $O(1/\epsilon^{3/2})$ episodes to find an $\epsilono$-approximate stationary point of the nonconcave performance function $J(\boldsymbol{\theta})$ and directly optimizes the policy by finding the optimal θ. Expand
Global Convergence of Langevin Dynamics Based Algorithms for Nonconvex Optimization
TLDR
We present a unified framework to analyze the global convergence of Langevin dynamics based algorithms for nonconvex finite-sum optimization with $n$ component functions. Expand
Stochastic Nested Variance Reduced Gradient Descent for Nonconvex Optimization
TLDR
We study finite-sum nonconvex optimization problems, where the objective function is an average of $n$ non Convex functions. Expand
Stochastic Nested Variance Reduction for Nonconvex Optimization
TLDR
We study finite-sum nonconvex optimization problems, where the objective function is an average of $n$ non Convex functions. Expand
A Finite Time Analysis of Two Time-Scale Actor Critic Methods
TLDR
We provide a non-asymptotic analysis for two time-scale actor-critic methods under non-i.i.d. setting. Expand
Stochastic Variance-Reduced Cubic Regularized Newton Method
TLDR
We propose a stochastic variance-reduced cubic regularized Newton method for non-convex optimization, which outperforms the state-of-the-art cubic regularization algorithms including subsampled cubic regularizations. Expand
Epidemic Model Guided Machine Learning for COVID-19 Forecasts in the United States
TLDR
We propose a new epidemic model (SuEIR) for forecasting the spread of COVID-19, including numbers of confirmed and fatality cases at national and state levels in the United States. Expand
Subsampled Stochastic Variance-Reduced Gradient Langevin Dynamics
TLDR
We propose a variant of stochastic gradient Langevin dynamics that replaces the full gradient in each epoch with a subsampled one, which improves the existing convergence rate by a factor of κn, when the sample size is large or the target accuracy requirement is moderate. Expand
Stochastic Variance-Reduced Cubic Regularized Newton Methods
We propose a stochastic variance-reduced cubic regularized Newton method (SVRC) for nonconvex optimization. At the core of our algorithm is a novel semi-stochastic gradient along with aExpand
...
1
2
3
4
...