基于平均报酬模型的强化学习算法研究

对于有吸收目标状态的循环任务,比较合理的方法是采用基于平均报酬模型的强化学习.平均报酬模型强化学习具有收敛速度快、鲁棒性强等优点.本文介绍了平均报酬模型强化学习的3个主要算法:R学习、H学习和LC学习,并给出了平均报酬模型强化学习的主要应用及研究方向.     

渗透心理辅导 强化高校思想政治工作

从广泛的教育内涵来说,心理辅导是思想政治教育的一个重要组成部分。本文在阐述了心理辅导与思想政治工作的关系基础之上,提出在渗透心理辅导的教育功能时,应把握好心理的时空特征、对象特征、问题特征和发展特征,以强化思想政治工作的效果。     

Defining delayed consequences as reinforcers: some do, some don't, and nothing changes

Results of a survey sent to members of the editorial boards of five behavior-analytic journals in 1990 indicated that there was no consensus among respondents with respect to whether delayed events can function as reinforcers (Schlinger, Blakely, Fillhard, & Poling, 1991). Since that time, several studies with nonhuman animals have demonstrated that events delayed by at least several seconds can increase the probability of occurrence of the responses that produce them. To determine the relative number of contemporary behavior analysts who nonetheless believe that reinforcers must immediately follow the responses that produce them, the present study replicated the procedures used by Schlinger et al., with similar results. That is, respondents did not agree with respect to whether delayed events could be reinforcers.     

扩能运输对朔黄铁路桥梁的影响及维管对策研究

随着朔黄铁路扩能运输的快速推进,单元列车轴重从21 t逐步提升到25 t,年运量即将突破2亿t,荷载与疲劳效应的增加对既有铁路桥梁特别是中小跨径桥梁的安全运营要求越来越高。分析重载运输对桥梁的主要影响因素,采取桥梁检查、安全状态评估、快速维修加固、加强维修质量验收相结合等系列应对措施,提出桥梁科学维管的对策,为建立桥梁安全运营保障体系奠定基础。     

Occupational Safety:

After a feedback condition, assigned (foreman-selected) and participative (worker-selected) goal-setting were examined to determine how they might augment the safety performance of 150 workers in 17 rooms of a paper mill. A statistical comparison of performance under assigned goal-setting, with that under baseline conditions of feedback-alone demonstrated a significant increase in the percentage of nonhazardous conditions but not of safe practices. Goal-setting by workers did not produce significant increases for either practices or conditions, and rates of injuries remained at the same reduced level as during feedback-alone. A visual analysis of a graphic plot of repeated performance measures suggested that trends established during feedback alone could account, in part, for the improvements. Possibly ceiling effects and the stated opinions of workers about various attributes of goal-setting may have been related to the outcomes.     

A People-Based Approach to OBM

Abstract

This rejoinder discusses some common points raised in the commentaries of this special issue of JOBM, including (1) different definitions of positive and negative, (2) use of the terms reinforcement and establishing operation, and (3) concerns over the abandonment of behavior analysis for mentalistic research. In addition, a people-based approach to psychology is introduced and advocated as a leading-edge approach to understanding and improving organizational behavior.     

基于深度强化学习的救护车动态重定位调度研究

救护车是挽救患者生命的重要医疗资源,合理调配有限的救护车资源可以降低呼叫响应时间,提高医疗服务水平.本文面向救护车动态重定位调度问题,提出了一种基于强化学习的调度策略结构.为解决传统强化学习所面临的高维状态空间的挑战,本文基于深度Q值网络(DQN)方法,提出了一种考虑多种调度交互因子的算法RedCon-DQN,以在给定环境状态下得到最优的重定位调度策略.在此基础上,本文还提出了急救网络弹性概念,以评估各站点对全局救护优化目标的影响力.最后,基于南京市2016年~2017年的实际救护车呼叫及响应数据,构造了环境交互模拟器.在模拟器中通过大规模数据实验,验证了模型得到的调度策略相比已有方法的优越性,并分析了不同时段下调度策略的有效性及其特点.     

Comments on Michael (1993): Establishing Operations

The present comments concern Michael''s concept of motivative variables, and the implications of that concept for our understanding of the nature of reinforcement as well as the extinction of responses maintained through positive and negative reinforcement. We note that both extinction and altering motivative variables decrease responding, but they do so differently. The former does so by discontinuing the response-reinforcer relation. The latter does so by altering the motivation to respond. We emphasize that we shouldn''t conclude we have extinguished a response just because we have performed some operation that results in decreased responding. The difference is especially important for an understanding of how we might reduce maladaptive avoidance responses, such as found in phobias or obsessive-compulsive disorders.     

The Several Meanings of “Positive”

Abstract

Wiegand and Geller propose that the salient role of positive reinforcement in behavior analysis should enable a melding of behavior analysis with developments and concepts that have appeared under the banner of “positive psychology.” However, as is true of many words, the term positive has more than one meaning, and the positive of positive reinforcement is not the same as the positive of “positive psychology.” The latter is parasitic upon the vernacular, as “nice” or “desirable,” whereas the former is analogous to the algebraic “add” as when an action produces the appearance (as contrasted with the removal) of some event. The distinct meanings become clear with recognition that addictive and criminal behavior often are maintained through positive reinforcement, and that negative reinforcement of behavior often is benign and beneficial to the persons involved. In addition, most of the phenomena identified with positive psychology that Wiegand and Geller propose to embrace entail more subtle and complex combinations of behavioral principles than these authors acknowledge. Wiegand and Geller also propose to accommodate vernacular assumptions in ways that separate their approach from its conceptual base; this risks impairing the effectiveness of their work whether or not its marketability would be improved.     

小城镇建设:实践、困境与建设思路

小城镇建设是中国城镇化体系中重要的环节,对于中国城市发展与农村发展均具有重要意义。中国小城镇建设已经持续多年,取得积极的成果,但也存在很多问题。例如,小城镇在产业和人口集聚、体制机制等方面存在发展难题。本文通过对山东、浙江小城镇建设实践的对比分析,从小城镇的定位与建设内容等方面,对小城镇建设思路进行梳理,认为小城镇需要在基础设施建设、体制机制改革、经济建设等三方面共同努力,才能促进小城镇的健康发展。