COACH: Learning continuous actions from COrrective Advice Communicated by Humans

Author(s):  
Carlos Celemin ◽  
Javier Ruiz-del-Solar
2021 ◽  
Vol 11 (1) ◽  
pp. 19-35
Author(s):  
Nicolae Pintilie ◽  

This paper aims to create an image of progress towards circular economy registered by European Union countries through specific indicators. In this way, this paper is based on the study and analysis of the 13 indicators, grouped on 4 pillars: Production and consumption, Waste management, Secondary raw materials, Competitiveness and innovation. After the presentation of the methodology, the paper develops an analysis in time and space of the selected indicators, then an analysis of the countries with their grouping on clusters, creating a map of them and highlighting the current situation of circular economy in the European Union. Moreover, the paper also presents the evolution of the countries regarding circular economy, which has a big importance taking into account that in the European Union the preoccupations for this concept is higher from one period to another. Among the most interesting results are: (1) a massive concentration of countries with problems for Waste management pillar; (2) Europe is one of the regions with the largest contribution in terms of circular economy, but the concept is developing differently from one country to another; (3) The scoreboard evolution is particularly useful in revealing the continuous actions adopted by countries in order to facilitate the conversion to circular economy. Finally, the paper presents possible limits of the research, but also future directions of its development.


Author(s):  
Simon Walters ◽  
Andy Rogers ◽  
Anthony R.H. Oldham

2020 ◽  
pp. 1-12
Author(s):  
HANFENG LI ◽  
ZHEN RONG

We study the independence density for finite families of finite tuples of sets for continuous actions of discrete groups on compact metrizable spaces. We use it to show that actions with positive naive entropy are Li–Yorke chaotic and untame. In particular, distal actions have zero naive entropy. This answers a question of Lewis Bowen.


2013 ◽  
Vol 25 (6) ◽  
pp. 1512-1547 ◽  
Author(s):  
Tingting Zhao ◽  
Hirotaka Hachiya ◽  
Voot Tangkaratt ◽  
Jun Morimoto ◽  
Masashi Sugiyama

The policy gradient approach is a flexible and powerful reinforcement learning method particularly for problems with continuous actions such as robot control. A common challenge is how to reduce the variance of policy gradient estimates for reliable policy updates. In this letter, we combine the following three ideas and give a highly effective policy gradient method: (1) policy gradients with parameter-based exploration, a recently proposed policy search method with low variance of gradient estimates; (2) an importance sampling technique, which allows us to reuse previously gathered data in a consistent way; and (3) an optimal baseline, which minimizes the variance of gradient estimates with their unbiasedness being maintained. For the proposed method, we give a theoretical analysis of the variance of gradient estimates and show its usefulness through extensive experiments.


Sign in / Sign up

Export Citation Format

Share Document