Dynamic Programming and Optimal Control, Vol. II

By Dimitri P. Bertsekas. 1976

Subfields: Reinforcement Learning, Mathematical Optimization

How researchers see this paper

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

In Reinforcement Learning

In Mathematical Optimization

Citations in the corpus

Value Iteration Networks (2016, methodology): “, 1999), hierarchical macro-actions (Mann & Mannor, 2014), and state-aggregation (Taylor et al., 2009; Bertsekas, 2012) for reducing the number of iterations required for VI to converge, with the goal of reducing computation time.”
Consistent On-Line Off-Policy Evaluation (2017, methodology): “and is a γ(1−λ) 1−λγ -contraction (Bertsekas, 2012).”
Generalized Emphatic Temporal Difference Learning: Bias-Variance Analysis (2015, methodology): “Instead, an approximate value-function is sought us-ing various function-approximationtechniques (a.k.a. approximate dynamic-programming; Bertsekas 2012).”
Emphatic TD Bellman Operator is a Contraction (2015, methodology): “thevaluefunctionis often impossible. Instead, an approximate value-function is sought using various function-approximation techniques (Sutton & Barto 1998; a.k.a. approximate dynamic-programming; Bertsekas 2012). In this approach, the parameters of the value-functionapproximationare tuned using ”
Infinite-Horizon Policy-Gradient Estimation (2001, methodology): “, r(n)] (Bertsekas, 1995).”
Multi-Objective Markov Decision Processes for Data-Driven Decision Support (2016, background): “Because in the finite-horizon setting that the optimal π is in general non-stationary (Bertsekas, 2007), we define π to be a sequence of functions πt for t P t1, .”
Policy Gradients Beyond Expectations: Conditional Value-at-Risk (2014, background): “Assumption 6 is common in episodic MDPs (Bertsekas, 2012), and requires that all states have a positive probability of being visited.”
A Non-Parametric Approach to Dynamic Programming (2011, background): “Value functions are an essential concept for determining optimal policies in both optimal control [1] and reinforcement learning [2, 3].”
Risk-Sensitive and Robust Decision-Making: a CVaR Optimization Approach (2015, background): “We show a new result relating the CVaR objective in (3) to the worst-case expected discounted-cost in presence of worst-case perturbations of the MDP parameters, where the perturbations are budgeted according to the “number of things that can go wrong.””
Scaling Up Robust MDPs by Reinforcement Learning (2013, methodology): “Since the projection operator Π is known to be non-expansive in the d -weighted norm [2], we have the following corollary.”
Stability and control of a quadrocopter despite the complete loss of one, two, or three propellers (2014, methodology): “An LQR controller [23] is implemented in each case for the inner loop control.”