Search results for "Trial and error"
showing 10 items of 14 documents
CostNet: An End-to-End Framework for Goal-Directed Reinforcement Learning
2020
Reinforcement Learning (RL) is a general framework concerned with an agent that seeks to maximize rewards in an environment. The learning typically happens through trial and error using explorative methods, such as \(\epsilon \)-greedy. There are two approaches, model-based and model-free reinforcement learning, that show concrete results in several disciplines. Model-based RL learns a model of the environment for learning the policy while model-free approaches are fully explorative and exploitative without considering the underlying environment dynamics. Model-free RL works conceptually well in simulated environments, and empirical evidence suggests that trial and error lead to a near-opti…
Artificial organisms as tools for the development of psychological theory: Tolman's lesson
2007
In the 1930s and 1940s, Edward Tolman developed a psychological theory of spatial orientation in rats and humans. He expressed his theory as an automaton (the ‘‘schematic sowbug’’) or what today we would call an ‘‘artificial organism.’’ With the technology of the day, he could not implement his model. Nonetheless, he used it to develop empirical predictions which tested with animals in the laboratory. This way of proceeding was in line with scientific practice dating back to Galileo. The way psychologists use artificial organisms in their work today breaks with this tradition. Modern ‘‘artificial organisms’’ are constructed a posteriori, working from experimental or ethological observations…
Ecosistemas locales de aprendizaje ante la globalización tecnológica. Retos de los modelos educativos digitales pospandemia
2021
La educación frente a la covid-19 ha puesto en práctica una infinidad de respuestas, todas urgentes, tentativas, generadas por ensayo y por error, y todavía pendientes de validación. Para comprender mejor el impacto de estas respuestas, el presente artículo defiende, partiendo de la teoría de los ecosistemas de aprendizaje, la necesidad de investigar cómo han influido las tecnologías de ámbito global en la construcción de experiencias locales de aprendizaje. Además, pone de manifiesto la necesidad de estudiar qué tipo de apoyo tanto teórico como práctico se requiere para reconducir los posibles efectos nocivos de la introducción de la tecnología de emergencia. A tal efecto, se describen las…
Measuring of Geometrical Precision of Some parts Obtained by Asymmetric Incremental Forming Process After Trimming
2007
Asymmetric Incremental Forming exalts the advantages of Incremental Forming process since no dies are strictly necessary. In this way complex geometries may be manufactured with a very simple clamping equipment. On the other hand, this characteristic determines some intrinsic drawbacks which penalise its industrial suitability; first of all, the dimensional control of the manufactured part is a still open point for researchers all over the world. Several approaches have been already proposed in the last years to solve the problem, resulting only in partial solutions. At the same time, up to now, the numerical simulation did not supply significant aid to the designers, due to the problem com…
A systematic approach for fine-tuning of fuzzy controllers applied to WWTPs
2010
A systematic approach for fine-tuning fuzzy controllers has been developed and evaluated for an aeration control system implemented in a WWTP. The challenge with the application of fuzzy controllers to WWTPs is simply that they contain many parameters, which need to be adjusted for different WWTP applications. To this end, a methodology based on model simulations is used that employs three statistical methods: (i) Monte-Carlo procedure: to find proper initial conditions, (ii) Identifiability analysis: to find an identifiable parameter subset of the fuzzy controller and (iii) minimization algorithm: to fine-tune the identifiable parameter subset of the controller. Indeed, the initial locatio…
Agent's actions as a classification criteria for the state space in a learning from rewards system
2008
We focus in this paper on the problem of learning an autonomous agent's policy when the state space is very large and the set of actions available is comparatively short. To this end, we use a non-parametric decision rule (concretely, a nearest-neighbour strategy) in order to cluster the state space by means of the action that leads to a successful situation. Using an exploration strategy to avoid greedy behaviour, the agent builds clusters of positively-classified states through trial and error learning. In this paper, we implement a 3D synthetic agent which plays an 'avoid the asteroid' game that suits our assumptions. Using as the state space a feature vector space extracted from a visua…
Automatic EKF tuning for UAS path following in turbulent air
2018
By using two simultaneously working Extended Kalman Filters, a procedure is implemented in order to perform in a fully autonomous way the path following in turbulent air. To guarantee the robustness of the proposed algorithm, an automatic tuning procedure is proposed to determine optimal values of Process and Measurement Noise statistics. Such a procedure is based on both the characteristics of the disturbances and the desired flight path; in particular, a specific performance index is applied to tune filters. In this way control laws are adapted to the flight condition and these lead to an optimal path-following. This research represents an upload of previous papers. It allows eliminating …
Two Views for Understanding How TQM Fosters Learning and Value Innovation: Absorptive Capabilities and Action-Based Management
2014
In the last decade some frameworks have tried to explain how to devise strategies for innovation in value by determining the needs of customers and non-customers, also creating new industries in which competition becomes irrelevant (Hax, The delta model. Reinventing your business strategy. New York: Springer, 2010; Kim and Mauborgne, Blue ocean strategy. Boston: Harvard Business School Press, 2005; Madhok and Marques 2013). These reference frameworks are based on a common set of principles: Value is created through the relationship with the customer (Priem, Acad Manag Rev 23; 219–235, 2007; Vargo and Lusch 2008); Strategy is considered to be a continuous process of exploring new opportuniti…
Deep Drawing Process Design: A Multi Objective Optimization Approach
2009
In sheet metal forming most of the problems are multi objective problems, generally characterized by conflicting objectives. The definition of proper parameters aimed to prevent both wrinkles and fracture is a typical example of an optimization problem in sheet metal forming characterized by conflicting goals. What is more, nowadays, a great interest would be focused on the availability of a cluster of possible optimal solutions instead of a single one, particularly in an industrial environment. Thus, the design parameters calibration, accomplishing all the objectives, is difficult and sometimes unsuccessful. In order to overcome this drawback a multi-objectives optimization procedure based…
Multiobjective optimization of an ultrasonic transducer using NIMBUS
2005
The optimal design of an ultrasonic transducer is a multiobjective optimization problem since the final outcome needs to satisfy several conflicting criteria. Simulation tools are often used to avoid expensive and time-consuming experiments, but even simulations may be inefficient and lead to inadequate results if they are based only on trial and error. In this work, the interactive multiobjective optimization method NIMBUS is applied in designing a high-power ultrasonic transducer. The performance of the transducer is simulated with a finite element model, and three design goals are formulated as objective functions to be minimized. To find an appropriate compromise solution, additional pr…