Trust Region Methods

1,835 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Off-Policy Actor-Critic with Shared Experience Replay (2019, methodology): “b) a trust region scheme (Conn et al., 2000; Schulman et al., 2015; 2017) can mitigate bias and enable efficient learning in a strongly off-policy regime, where distinct agents share experience through a commonly shared replay module.”
Optimization Methods for Large-Scale Machine Learning (2016, methodology): “t used in the algorithm. 6.1.2 Dealing with Nonconvexity Hessian-free Newton methods are routinely applied for the solution of nonconvex problems. In such cases, it is common to employ a trust region [34] instead of a line search and to introduce an additional condition in Step5of Algorithm6.1: term”
Complexity and global rates of trust-region methods based on probabilistic models (2018, neutral): “Trust-region methods form a well established and understood class of methods for the minimization of a nonlinear (possibly nonsmooth) function subject or not to constraints on its variables (see the book [7] and the recent survey [28]).”
Gradient Primal-Dual Algorithm Converges to Second-Order Stationary Solutions for Nonconvex Distributed Optimization (2018, methodology): “have been focused on designing algorithms with convergence guarantees to local minimum points/ss2 for non-convex unconstrained problems. These include second-order methods such as trust region method [14], cubic regularized Newton’s method [15], and a hybrid of ﬁrst-order and second-order methods [1”