人工神经网络及其应用

本文首首先介绍了人工神经网络的源泉一体神经网络的主要基本内容，然后简要论述了人工神经网络的主要内容，进程和问题，人工神经网的应用。     

Humans: the missing link in manufacturing simulation?

Computer based discrete event simulation (DES) is one of the most commonly used aids for the design of automotive manufacturing systems. However, DES tools represent machines in extensive detail, while only representing workers as simple resources. This presents a problem when modelling systems with a highly manual work content, such as an assembly line. This paper describes research at Cranfield University, in collaboration with the Ford Motor Company, founded on the assumption that human variation is the cause of a large percentage of the disparity between simulation predictions and real world performance. The research aims to improve the accuracy and reliability of simulation prediction by including models of human factors.     

Speed–accuracy tradeoff in Fitts’ law tasks—on the equivalency of actual and nominal pointing precision

Pointing tasks in human–computer interaction obey certain speed–accuracy tradeoff rules. In general, the more accurate the task to be accomplished, the longer it takes and vice versa. Fitts’ law models the speed–accuracy tradeoff effect in pointing as imposed by the task parameters, through Fitts’ index of difficulty (I_d) based on the ratio of the nominal movement distance and the size of the target. Operating with different speed or accuracy biases, performers may utilize more or less area than the target specifies, introducing another subjective layer of speed–accuracy tradeoff relative to the task specification. A conventional approach to overcome the impact of the subjective layer of speed–accuracy tradeoff is to use the a posteriori “effective” pointing precision W_e in lieu of the nominal target width W. Such an approach has lacked a theoretical or empirical foundation. This study investigates the nature and the relationship of the two layers of speed–accuracy tradeoff by systematically controlling both I_d and the index of target utilization I_u in a set of four experiments. Their results show that the impacts of the two layers of speed–accuracy tradeoff are not fundamentally equivalent. The use of W_e could indeed compensate for the difference in target utilization, but not completely. More logical Fitts’ law parameter estimates can be obtained by the W_e adjustment, although its use also lowers the correlation between pointing time and the index of difficulty. The study also shows the complex interaction effect between I_d and I_u, suggesting that a simple and complete model accommodating both layers of speed–accuracy tradeoff may not exist.     

人类HLA-DRB1基因型与民族和人群的神经网络无监督模式分类

HLA-DRB1等位基因主要参与人类的抗原免疫功能,对其进行比较研究,有助于追溯人类的进化迁移史,开发类群特异性药物.本研究以Matlab为平台,用自组织竞争网络(Self-Organizing Competitive Neural Network),对世界54个民族和人群、14个HLA-DRB1等位基因,进行了无监督模式分类.结果表明,各民族之间存在差异性,同民族的各人群之间有相似性.西伯利亚各人群,澳洲各土著人群,黑人各人群,南美印第安各人群,犹太族各人群,日本各人群,及欧美白人各人群有相对独立性:南美印第安人与西伯利亚人有高度相似性;中国民族在南方人群和北方人群间存在较大差异;中国汉族中,广东汉族地位特殊,与少数民族如拉祜族、瑶族关系密切.     

Rapid degradation of the mycotoxin patulin in man quantified by stable isotope dilution assays

The absorption and degradation of the mycotoxin patulin in man was quantified by using a recently developed stable isotope dilution assay. Application of this currently most sensitive method revealed a patulin content less than 200 ng l -1 in the blood serum of five consumers of apple juice. Likewise, no patulin was found in the serum of a volunteer, whose blood was drawn shortly after consumption of a juice containing a maximum tolerable amount of patulin. In further in vitro experiments, the degradation of patulin by reacting it with whole blood was investigated. After addition of 100 μg patulin to 9 ml blood, only 6.1% of the mycotoxin was detected after 2 min. It was concluded, therefore, that even high naturally occurring concentrations of patulin in foods are quickly degraded before reaching other tissues than the gastrointestinal tract.     

适用于实时协同编著系统的并发控制研究

实时协同编著系统是最受欢迎的实时ＣＳＣＷ应用系统之一，适用于实时协同编著系统的并发控制用于解决参与协同工作用户对共享对象操作时存在的冲突问题。本文对这种新的并发控制与适用于分布式数据库系统等的传统并发控制进行了比较，讨论了一些适用于实时协同编著系统并发控制方法的新评价参数，分析了传统并发控制方法应用到实时协同编著系统时所遇到的问题。最后，我们介绍了一些文献中已有原型系统所使用的并发控制方法。     

The Danish Twin Apartment Study – Part II: Mathematical modeling of the relative strength of sources of indoor air pollution

Abstract This study deals with the modeling of air pollution in apartments from laboratory measurements of source strengths, using formaldehyde and Total Volatile Organic Compounds (TVOCs) as model pollutants. The sources in two test apartments were grouped into two: building-related sources and occupant-related sources. The measured source strengths and ventilation rates were used for the prediction of concentrations expected in the apartments. These predictions were compared to measurements in the apartment over 12 months. The conclusions were that the model predictions based on emission rates measured in the laboratory can be used to predict the long-term concentration of the two model pollutants in the apartments. Considering the measured differences in ventilation between the apartments, an occupant emission rate of between 0.2 and 0.3 mg/h/kg body weight could be estimated. Based on previous suggested limits of acceptable exposures of humans to VOCs, an acceptable average emission rate of VOCs from building materials in general was estimated to be about 30 (μ/m²/h. The modeling showed that during the first 200 days, building materials dominated the emissions. After this, sources relating to the occupants dominated. On average about half of the VOC pollution originated from the building materials.     

Active symbols and internal models: Towards a cognitive connectionism

In the first section of the article, we examine some recent criticisms of the connectionist enterprise: first, that connectionist models are fundamentally behaviorist in nature (and, therefore, non-cognitive), and second that connectionist models are fundamentally associationist in nature (and, therefore, cognitively weak). We argue that, for a limited class of connectionist models (feed-forward, pattern-associator models), the first criticism is unavoidable. With respect to the second criticism, we propose that connectionist modelsare fundamentally associationist but that this is appropriate for building models of human cognition. However, we do accept the point that there are cognitive capacities for which any purely associative model cannot provide a satisfactory account. The implication that we draw from is this is not that associationist models and mechanisms should be scrapped, but rather that they should be enhanced.In the next section of the article, we identify a set of connectionist approaches which are characterized by active symbols — recurrent circuits which are the basis of knowledge representation. We claim that such approaches avoid criticisms of behaviorism and are, in principle, capable of supporting full cognition. In the final section of the article, we speculate at some length about what we believe would be the characteristics of a fully realized active symbol system. This includes both potential problems and possible solutions (for example, mechanisms needed to control activity in a complex recurrent network) as well as the promise of such systems (in particular, the emergence of knowledge structures which would constitute genuine internal models).     

Investigation of oil palm harvesting tools design and technique on work-related musculoskeletal disorders of the upper body

The oil palm industry is one of the important sectors in Malaysia. The growth and development of this industry shows that Malaysia is the world second-largest oil palm producers. However, in the fresh fruit bunch (FFB) harvesting process, the harvesters are exposed to many types of work-related musculoskeletal disorders (WMSDs). The FFB harvesters tend to develop WMSDs especially the shoulders and trunk. Hence, it is important to identify the exposure levels, awkward postures and the reaction forces of muscle activity based on the posture and movement of the harvesters when using pole, chisel and loading spike during the harvesting process. The objective of this study was to investigate the effect of the design of oil palm FFB harvesting tools on WMSDs of the upper body. Rapid Upper Limb Analysis (RULA) was used to investigate and assess the exposure level on the harvester body during the harvesting process. The assessment showed that the shoulders and trunk have high exposure level and undergo awkward posture. Human Musculoskeletal Model Analysis (HMMA) was used to identify the reaction force exerted on the muscle during the FFB harvesting process. In this study, 4 muscles were analysed including Triceps, Biceps, Erector Spinae and Psoas Major. The highest reaction force of 16.36 N was found on the left triceps when handling a loading spike. In conclusion, it is important to address the risks by reviewing all possible aspects that contribute to the WMSDs and interventions on the tool design, task and working shifts may be required.     

Improving human robot collaboration through Force/Torque based learning for object manipulation

Human–Robot Collaboration (HRC) is a term used to describe tasks in which robots and humans work together to achieve a goal. Unlike traditional industrial robots, collaborative robots need to be adaptive; able to alter their approach to better suit the situation and the needs of the human partner. As traditional programming techniques can struggle with the complexity required, an emerging approach is to learn a skill by observing human demonstration and imitating the motions; commonly known as Learning from Demonstration (LfD). In this work, we present a LfD methodology that combines an ensemble machine learning algorithm (i.e. Random Forest (RF)) with stochastic regression, using haptic information captured from human demonstration. The capabilities of the proposed method are evaluated using two collaborative tasks; co-manipulation of an object (where the human provides the guidance but the robot handles the objects weight) and collaborative assembly of simple interlocking parts. The proposed method is shown to be capable of imitation learning; interpreting human actions and producing equivalent robot motion across a diverse range of initial and final conditions. After verifying that ensemble machine learning can be utilised for real robotics problems, we propose a further extension utilising Weighted Random Forest (WRF) that attaches weights to each tree based on its performance. It is then shown that the WRF approach outperforms RF in HRC tasks.