Product development of an ergonomic keyboard

Abstract

The first truly ‘ergonomic’ keyboard for modern VDUs was shown to the public in April 1983 in Zurich. This split keyboard was the result of a research programme carried out by the Department of Hygiene and Work Physiology, ETH, Zurich. The ergonomic factors of the keyboard were convincing to the press, computer manufacturers and users. The steps from models of this keyboard to the final product and related problems of industrial design and engineering are explained. This will help in understanding the distinctive time lags between the results of scientific research and the availability of a product in the market–place.     

Counterfactual potency.

Counterfactual thoughts typically take the form of implied or explicit if–then statements. We propose that the multiplicative combination of “if likelihood” (the degree to which the antecedent condition of the counterfactual is perceived to be likely) and “then likelihood” (the perceived conditional likelihood of the outcome of the counterfactual, given the antecedent condition) determine the strength and impact of counterfactuals. This construct, termed counterfactual potency, is a reliable predictor of the degree of influence of counterfactual thinking upon judgments of regret, causation, and responsibility. Through 4 studies, we demonstrate the predictive power of this construct in a variety of contexts and show that it plays a causal role in determining the strength of the effects of counterfactual thought. Implications of counterfactual potency as a central factor of counterfactual influence are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

When debiasing backfires: Accessible content and accessibility experiences in debiasing hindsight.

Two studies demonstrated that attempts to debias hindsight by thinking about alternative outcomes may backfire and traced this to the influence of subjective accessibility experiences. Participants listed either few (2) or many (10) thoughts about how an event might have turned out otherwise. Listing many counterfactual thoughts was experienced as difficult and consistently increased the hindsight bias, presumably because the experienced difficulty suggested that there were not many ways in which the event might have turned out otherwise. No significant hindsight effects were obtained when participants listed only a few counterfactual thoughts, a task subjectively experienced as easy. The interplay of accessible content and subjective accessibility experiences in the hindsight bias is discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

利用可信反事实的不平衡数据过采样方法

针对传统过采样方法不能充分利用数据集信息的缺陷,提出一种基于反事实(counterfactual,CF)的不平衡数据过采样方法,并进一步对生成的少数类合成样本进行了\"可信\"清除.其核心思想是依据数据集原有实例特征值合成新样本,相比传统过采样的插值法,更能充分挖掘数据中的边界决策信息,从而为分类器提供更多的有用信息,提高分类性能.在9个来自KEEL与UCI的不平衡数据集、5种不同分类器(SVM、DT、Logistic、RF、AdaBoost)上与4种传统过采样方法(SMOTE、B1-SMOTE、B2-SMOTE、ADASYN)进行了大量对比实验,结果表明,所提方法具有更高的AUC值、F1值和G-mean值,可以更为有效地解决类不平衡问题.     

考虑因果约束的异常对象反事实解释

现有的异常检测方法大多关注算法的效率和精确度等, 而忽视了异常对象的可解释性. 反事实解释方法是当前可解释机器学习的研究热点之一, 旨在通过对研究对象的特征进行扰动, 进而生成反事实示例以解释模型的决策结果. 在实际应用中, 特征之间可能存在某种因果关系. 然而, 现有基于反事实的可解释方法大多关注如何生成更多样的反事实示例, 却忽视了特征之间的因果关系, 导致可能产生不合理的反事实解释. 为此, 提出了一种考虑因果约束的异常对象反事实解释算法IARC. 该方法在生成反事实解释时, 通过将特征间的因果性纳入目标函数来衡量该次扰动是否可行, 并通过改进后的遗传算法进行求解, 从而生成合理的反事实解释. 此外, 提出了一种新的度量指标, 用于衡量所生成反事实解释的矛盾程度. 同多个先进反事实解释方法在多个真实数据集上进行了对比实验和详细的案例可解释分析. 实验结果表明, 所提出的方法能够为异常对象生成具有强合理性的反事实解释.     

反事实增强的对抗学习序列推荐

最近,强化学习技术在序列推荐系统取得成功,它能从用户长期反馈信号中学习有效的推荐策略.然而,模型的激励函数设计面临区分度过低的难题.这限制了模型学习不同用户反馈信号间的价值差异的能力,并导致推荐策略总是次优的.现有工作主要通过调节衰减因子来保证激励函数区分度,但它依赖专家先验知识缺乏理论基础.为了更合理地设计激励函数和提高其区分度,本文依据因果论来分析推荐系统,并提出一种基于反事实区分度增强的序列推荐算法CAL4Rec.首先,所提出方法用结构因果图描述序列推荐过程,并创造性地用因果图定义了因果可鉴别的价值激励区分度.其次,该方法用反事实生成对抗的自监督学习过程优化推荐策略网络,以学习用户的真实倾向.在一系列序列推荐基准数据集上,对CAL4Rec开展了广泛对比和消融实验,实验结果表明CAL4Rec的提升对多种网络实现结构有效(平均2.34%).     

因果推断三种分析框架及其应用综述

介绍因果推断所涉及的基本概念及其三种分析框架：反事实框架、潜在结果模型和结构因果模型。首先,从反事实框架介绍因果效应的发端;然后,从基于反事实的两个因果推断分析框架：潜在结果模型和结构因果模型,来分别阐述两个分析框架所涉及的关键理论和应用方法。其中,潜在结果模型使用数学和可计算的语言对因果理论进行阐述,是一种将假设、命题和结论清晰化表达的计算模型,其在原因和结果变量已知的前提下定量分析原因变量对结果变量的因果效应,并对缺失的潜在结果进行补齐,使观察性研究的效果接近试验性研究。结构因果模型则是一种基于图论的因果推断方法,它将事件分为观察、干预和反事实三个层级,并通过do运算将干预和反事实层级的因果关系都降维成可以通过统计学手段解决的问题。最后,探讨了现今多领域内因果推断的应用场景,并总结了三种分析框架的异同点。     

'It's Not My Fault-But Only I Can Change It": Counterfactual and Prefactual Thoughts of Managers.

In testing for the self-serving bias in performance evaluation, the authors propose that comparing managers' counterfactual and prefactual thoughts about subordinates' performance is more conclusive than the attributional approach and also offers practical advantages. In a study with 120 managers, a 4-way interaction of subordinate performance, temporal perspective, direction, and reference confirmed the predicted pattern. Managers' thoughts about how a weak performance could have been enhanced had external references, but thoughts about how such a performance could be enhanced in the future focused on the leader. This asymmetry was only observed for weak performance. Results are discussed with regard to biases in leaders' performance evaluations and to how counter- and prefactual thoughts could be used for leadership research and practice. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

How Surveillance Begets Perceptions of Dishonesty: The Case of the Counterfactual Sinner.

Three studies support the hypothesis that observers' impressions of actors reflect not only what actors do but also what they can easily be imagined doing. Participants in Studies 1 and 2 observed a 10-year-old boy take a math test in a context in which the incentive to cheat and the constraints against cheating varied. When the incentive to cheat was high but the likelihood of getting caught was also high, observers perceived a target who resisted the temptation to cheat as less honest than the average boy. This effect was not found when the incentive to cheat was low, which suggests that its occurrence under high temptation resulted from observers in that condition generating the counterfactual thought that the target would have cheated had the likelihood of detection been low. Study 3 further supported the link between spontaneous counterfactual thought and inferences of dishonesty. The implications of the counterfactual correspondence bias are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

From what might have been to what must have been: Counterfactual thinking creates meaning.

Four experiments explored whether 2 uniquely human characteristics—counterfactual thinking (imagining alternatives to the past) and the fundamental drive to create meaning in life—are causally related. Rather than implying a random quality to life, the authors hypothesized and found that counterfactual thinking heightens the meaningfulness of key life experiences. Reflecting on alternative pathways to pivotal turning points even produced greater meaning than directly reflecting on the meaning of the event itself. Fate perceptions (“it was meant to be”) and benefit-finding (recognition of positive consequences) were identified as independent causal links between counterfactual thinking and the construction of meaning. Through counterfactual reflection, the upsides to reality are identified, a belief in fate emerges, and ultimately more meaning is derived from important life events. (PsycINFO Database Record (c) 2010 APA, all rights reserved)