scholarly journals Robot Fast Adaptation to Changes in Human Engagement During Simulated Dynamic Social Interaction With Active Exploration in Parameterized Reinforcement Learning

2018 ◽  
Vol 10 (4) ◽  
pp. 881-893 ◽  
Author(s):  
Mehdi Khamassi ◽  
George Velentzas ◽  
Theodore Tsitsimis ◽  
Costas Tzafestas
Author(s):  
Riccardo Poiani ◽  
Andrea Tirinzoni ◽  
Marcello Restelli

Many real-world domains are subject to a structured non-stationarity which affects the agent's goals and the environmental dynamics. Meta-reinforcement learning (RL) has been shown successful for training agents that quickly adapt to related tasks. However, most of the existing meta-RL algorithms for non-stationary domains either make strong assumptions on the task generation process or require sampling from it at training time. In this paper, we propose a novel algorithm (TRIO) that optimizes for the future by explicitly tracking the task evolution through time. At training time, TRIO learns a variational module to quickly identify latent parameters from experience samples. This module is learned jointly with an optimal exploration policy that takes task uncertainty into account. At test time, TRIO tracks the evolution of the latent parameters online, hence reducing the uncertainty over future tasks and obtaining fast adaptation through the meta-learned policy. Unlike most existing methods, TRIO does not assume Markovian task-evolution processes, it does not require information about the non-stationarity at training time, and it captures complex changes undergoing in the environment. We evaluate our algorithm on different simulated problems and show it outperforms competitive baselines.


2021 ◽  
pp. 287-358
Author(s):  
Michael A. Arbib

This chapter approaches aesthetics anew by considering empathy and Einfühlung, “feeling ourselves into” a work of art or architecture. The key neuroscience is the discovery of mirror neurons in monkeys that inspired the discovery of mirror systems in humans. Unsupervised, supervised, and reinforcement learning, each based on a different rule for synaptic plasticity, are presented as background for a computational model of how mirror neuron wiring is learned. Mirror neurons may serve social interaction, but they also self-monitor in acquiring new behaviors. This is exemplified in modeling how adaptive sequences of behavior may be mastered through learning the desirability and executability of actions. Such opportunistic scheduling complements the role of scripts. Empathy is linked to mirror systems but also depends on systems beyond the mirror. Returning to Einfühlung, we explore how a motor component may enrich our aesthetic appreciation by recognizing the actions and emotions of protagonists in a representational painting, or by gaining some feeling for the actions of the artist, sculptor, or architect in creating the work. Finally, case studies are sampled, including those in neuroaesthetics seeking neural correlates for aesthetic appreciation, that contribute to a tool kit for assessing the experience of buildings to enrich future design.


2020 ◽  
Vol 88 ◽  
pp. 103948 ◽  
Author(s):  
Leor M. Hackel ◽  
Peter Mende-Siedlecki ◽  
David M. Amodio

Sign in / Sign up

Export Citation Format

Share Document