NETWORKED NONLINEAR DECISION-MAKING: OPPORTUNISM, EXPLANATIONS, AND LEARNING ECHO-CHAMBERS

Abstract

In this proposal, our goal is to research and develop a rigorous mathematical framework for learning networked nonlinear decision-making with the specific focus to understand, define, and capture the concepts of networked intentions, emergence and opportunism. Our approach is to develop a new model of decision-making and opinion formation called a networked Double Transition Model (DTM) which is built from a base computational cognitive model called the DTM. Networked DTMs will be applied to understand nonlinear decision-making among interacting decision-makers with an emphasis on sequences of interacting decisions. Through DTMs, we will develop learning algorithms based on inverse reinforcement learning for networked DTMs as well as define a framework for explanations to provide insights into the nature of networked intentions and how decisions are made for both opportunistic and emergent situations. At its simplest, from new results through our prior AFOSR grant, we demonstrate that it is possible to provide explanations for Markov Decision Problems (MDPs), after MDPs have been learned through techniques such as inverse reinforcement learning. The induced reward function can be mathematically analyzed to allow us to build a reward structure that describes what drives a decision-maker, their decisions+processes, and eventually their intentions. The extension to networked decisions brings a new challenge for explanation as we now have the added interactions. Yet, this also opens up the potential to model conditions such as "echo chambers" and the impact of "fake information" as well as concepts like novelty in a rigorous fashion. We will carefully examine and define the relationship between these concepts and networked emergence and opportunism.

Document Details

Document Type
DoD Grant Award
Publication Date
Aug 12, 2021
Source ID
FA95502010032

Entities

People

  • Eugene Santos

Organizations

  • Air Force Office of Scientific Research
  • Board of Trustees of Dartmouth College
  • United States Air Force

Tags

Fields of Study

  • Computer science

Readers

  • Neural Network Machine Learning.
  • Systems Analysis and Design
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms