Stochastic Recursive Inclusions in Two Timescales with Nonadditive Iterate-Dependent Markov Noise

Yaji, Vinayaka G.; Bhatnagar, Shalabh

doi:10.1287/moor.2019.1037

Cited by 12 publications

(4 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following the pioneering work of Benaim, Hofbauer and Sorin [4,5], which, among other things, establish a counterpart of Theorem 2.1 for this case, this iteration has been extensively studied in literature, some of it motivated by applications to reinforcement learning [7,13,14,22,23,27,28,29]. The aforementioned extension of Theorem 2.1 is as follows.…”

Section: Stochastic Approximation: Preliminariesmentioning

confidence: 96%

Remarks on Differential Inclusion limits of Stochastic Approximation

Borkar¹,

Shah²

2023

Preprint

View full text Add to dashboard Cite

show abstract

Section: Stochastic Approximation: Preliminariesmentioning

confidence: 96%

Remarks on Differential Inclusion limits of Stochastic Approximation

Borkar¹,

Shah²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…This allows us to illustrate the behaviour of the Diff-DAC architecture with little effort. More general conditions can be found in, for example, Tadic (2004), Ramaswamy and Bhatnagar (2017), Yaji and Bhatnagar (2020).…”

Section: Convergence Analysismentioning

confidence: 99%

Fully distributed actor-critic architecture for multitask deep reinforcement learning

Macua¹,

Davies²,

Tukiainen³

et al. 2021

The Knowledge Engineering Review

View full text Add to dashboard Cite

We propose a fully distributed actor-critic architecture, named diffusion-distributed-actor-critic Diff-DAC, with application to multitask reinforcement learning (MRL). During the learning process, agents communicate their value and policy parameters to their neighbours, diffusing the information across a network of agents with no need for a central station. Each agent can only access data from its local task, but aims to learn a common policy that performs well for the whole set of tasks. The architecture is scalable, since the computational and communication cost per agent depends on the number of neighbours rather than the overall number of agents. We derive Diff-DAC from duality theory and provide novel insights into the actor-critic framework, showing that it is actually an instance of the dual-ascent method. We prove almost sure convergence of Diff-DAC to a common policy under general assumptions that hold even for deep neural network approximations. For more restrictive assumptions, we also prove that this common policy is a stationary point of an approximation of the original problem. Numerical results on multitask extensions of common continuous control benchmarks demonstrate that Diff-DAC stabilises learning and has a regularising effect that induces higher performance and better generalisation properties than previous architectures.

show abstract

“…We use the Markov model to build a panel data model to deeply explore the specific impact mechanisms and effects of disaster risk levels in different regions on the development of agricultural insurance. e characteristic of the Markov process is that the future state is only related to the present and has nothing to do with history [27]. at is to say, the state of the system at time t + 1 is only related to the state it was in at time t and is not affected by the state it was in before time t [28].…”

Section: Agricultural Insurance Risk Management Based On Markov Modelmentioning

confidence: 99%

Application of Markov Model-Based IoT in Agricultural Insurance and Risk Management

Chen

Jiang

2021

Mobile Information Systems

View full text Add to dashboard Cite

As the foundation of the national economy, agriculture is a high-risk, weak industry. Affected by many factors, agricultural production is subject to catastrophe risks from time to time. Agricultural production is mainly faced with two major threats, natural disaster risk and market risk. As an effective risk management tool, the production and promotion of agricultural insurance have played an essential role in guaranteeing the development of the agricultural industry in some developed countries and major agricultural countries in the world. This article combines the Internet of Things and Markov model for agricultural insurance risk management. First, we combine the structure of the Internet of Things and select relevant statistical data. Then, we build a panel data system, starting from two perspectives in different regions and analyze agricultural insurance’s current development and characteristics at each stage. In addition, we use the Markov model to build a panel data model to explore the specific impact mechanisms deeply. We also study the effects of disaster risk levels in different regions on the development of agricultural insurance. After simulation verification, we believe that this model can effectively promote the balanced regional development of agricultural insurance.

show abstract

Stochastic Recursive Inclusions in Two Timescales with Nonadditive Iterate-Dependent Markov Noise

Cited by 12 publications

References 36 publications

Remarks on Differential Inclusion limits of Stochastic Approximation

Remarks on Differential Inclusion limits of Stochastic Approximation

Fully distributed actor-critic architecture for multitask deep reinforcement learning

Application of Markov Model-Based IoT in Agricultural Insurance and Risk Management

Contact Info

Product

Resources

About