scholarly journals Comparing continual task learning in minds and machines

2018 ◽  
Vol 115 (44) ◽  
pp. E10313-E10322 ◽  
Author(s):  
Timo Flesch ◽  
Jan Balaguer ◽  
Ronald Dekker ◽  
Hamed Nili ◽  
Christopher Summerfield
Keyword(s):  
Human Error ◽  
Human Performance ◽  
Deep Neural Networks ◽  
Artificial Agents ◽  
Trial And Error ◽  
Stimulus Space ◽  
Cognitive Mechanisms ◽  
Error Training ◽  
Multiple Tasks ◽  
Continual Learning

Humans can learn to perform multiple tasks in succession over the lifespan (“continual” learning), whereas current machine learning systems fail. Here, we investigated the cognitive mechanisms that permit successful continual learning in humans and harnessed our behavioral findings for neural network design. Humans categorized naturalistic images of trees according to one of two orthogonal task rules that were learned by trial and error. Training regimes that focused on individual rules for prolonged periods (blocked training) improved human performance on a later test involving randomly interleaved rules, compared with control regimes that trained in an interleaved fashion. Analysis of human error patterns suggested that blocked training encouraged humans to form “factorized” representation that optimally segregated the tasks, especially for those individuals with a strong prior bias to represent the stimulus space in a well-structured way. By contrast, standard supervised deep neural networks trained on the same tasks suffered catastrophic forgetting under blocked training, due to representational interference in the deeper layers. However, augmenting deep networks with an unsupervised generative model that allowed it to first learn a good embedding of the stimulus space (similar to that observed in humans) reduced catastrophic forgetting under blocked training. Building artificial agents that first learn a model of the world may be one promising route to solving continual task performance in artificial intelligence research.

HUMAN FACTOR ANALYSIS OF CONTAINER VESSEL’S GROUNDING ACCIDENTS

2021 ◽  
Vol 159 (A1) ◽  
Author(s):  
U Yildirim ◽  
O Ugurlu ◽  
E Basar ◽  
E Yuksekyildiz
Keyword(s):  
Monte Carlo Simulation ◽  
Monte Carlo ◽  
Human Error ◽  
Human Performance ◽  
Human Factor ◽  
Study Data ◽  
Team Management ◽  
Contribution Rate ◽  
Physical Problems ◽  
Mental Problems

Investigation on maritime accidents is a very important tool in identifying human factor-related problems. This study examines the causes of accidents, in particular the reasons for the grounding of container ships. These are analysed and evaluation according to the contribution rate using the Monte Carlo simulation. The OpenFTA program is used to run the simulation. The study data are obtained from 46 accident reports from 1993 to 2011. The data were prepared by the International Maritime Organization (IMO) Global Integrated Shipping Information System (GISIS). The GISIS is one of the organizations that investigate reported accidents in an international framework and in national shipping companies. The Monte Carlo simulation determined a total of 23.96% human error mental problems, 26.04% physical problems, 38.58% voyage management errors, and 11.42% team management error causes. Consequently, 50% of the human error is attributable to human performance disorders, while 50% team failure has been found.

Deep Convolutional Neural Network for Object Classification

2021 ◽  
pp. 317-343
Author(s):  
Amira Ahmad Al-Sharkawy ◽  
Gehan A. Bahgat ◽  
Elsayed E. Hemayed ◽  
Samia Abdel-Razik Mashali
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Convolutional Neural Network ◽  
Computational Models ◽  
Human Performance ◽  
Deep Neural Networks ◽  
Object Classification ◽  
Classification Problem ◽  
Deep Convolutional Neural Network ◽  
Object Appearance

Object classification problem is essential in many applications nowadays. Human can easily classify objects in unconstrained environments easily. Classical classification techniques were far away from human performance. Thus, researchers try to mimic the human visual system till they reached the deep neural networks. This chapter gives a review and analysis in the field of the deep convolutional neural network usage in object classification under constrained and unconstrained environment. The chapter gives a brief review on the classical techniques of object classification and the development of bio-inspired computational models from neuroscience till the creation of deep neural networks. A review is given on the constrained environment issues: the hardware computing resources and memory, the object appearance and background, and the training and processing time. Datasets that are used to test the performance are analyzed according to the images environmental conditions, besides the dataset biasing is discussed.

scholarly journals Towards Mixed-Initiative Human–Robot Interaction: Assessment of Discriminative Physiological and Behavioral Features for Performance Prediction

Sensors ◽  
2020 ◽  
Vol 20 (1) ◽  
pp. 296 ◽  
Author(s):  
Caroline P. C. Chanel ◽  
Raphaëlle N. Roy ◽  
Frédéric Dehais ◽  
Nicolas Drougard
Keyword(s):  
Human Performance ◽  
Cardiac Activity ◽  
Human Robot Interaction ◽  
Operational Performance ◽  
Artificial Agents ◽  
Robot Interaction ◽  
Input Devices ◽  
Level Of Automation ◽  
Mixed Initiative ◽  
And Behavior

The design of human–robot interactions is a key challenge to optimize operational performance. A promising approach is to consider mixed-initiative interactions in which the tasks and authority of each human and artificial agents are dynamically defined according to their current abilities. An important issue for the implementation of mixed-initiative systems is to monitor human performance to dynamically drive task allocation between human and artificial agents (i.e., robots). We, therefore, designed an experimental scenario involving missions whereby participants had to cooperate with a robot to fight fires while facing hazards. Two levels of robot automation (manual vs. autonomous) were randomly manipulated to assess their impact on the participants’ performance across missions. Cardiac activity, eye-tracking, and participants’ actions on the user interface were collected. The participants performed differently to an extent that we could identify high and low score mission groups that also exhibited different behavioral, cardiac and ocular patterns. More specifically, our findings indicated that the higher level of automation could be beneficial to low-scoring participants but detrimental to high-scoring ones, and vice versa. In addition, inter-subject single-trial classification results showed that the studied behavioral and physiological features were relevant to predict mission performance. The highest average balanced accuracy (74%) was reached using the features extracted from all input devices. These results suggest that an adaptive HRI driving system, that would aim at maximizing performance, would be capable of analyzing such physiological and behavior markers online to further change the level of automation when it is relevant for the mission purpose.

scholarly journals Machine Learning for the Diagnosis of Orthodontic Extractions: A Computational Analysis Using Ensemble Learning

2020 ◽  
Vol 7 (2) ◽  
pp. 55
Author(s):  
Yasir Suhail ◽  
Madhur Upadhyay ◽  
Aditya Chhibber ◽  
Kshitiz
Keyword(s):  
Machine Learning ◽  
Human Error ◽  
Ensemble Methods ◽  
Treatment Decision ◽  
Learning Models ◽  
Error Training ◽  
Treatment Plans ◽  
Suitable Treatment ◽  
Using Data ◽  
Machine Learning Models

Extraction of teeth is an important treatment decision in orthodontic practice. An expert system that is able to arrive at suitable treatment decisions can be valuable to clinicians for verifying treatment plans, minimizing human error, training orthodontists, and improving reliability. In this work, we train a number of machine learning models for this prediction task using data for 287 patients, evaluated independently by five different orthodontists. We demonstrate why ensemble methods are particularly suited for this task. We evaluate the performance of the machine learning models and interpret the training behavior. We show that the results for our model are close to the level of agreement between different orthodontists.

Apply Performance Evaluation Matrix on Investigating Human Error Events in the Main Control Room

2010 ◽  
Author(s):  
Tzu-Chung Yenn ◽  
Yung-Tsan Jou ◽  
Chiuhsiang Joe Lin ◽  
Wan-Shan Tsai ◽  
Tsung-Ling Hsieh
Keyword(s):  
Performance Evaluation ◽  
Situation Awareness ◽  
Power Plants ◽  
Human Error ◽  
Human Performance ◽  
Main Stream ◽  
Control Room ◽  
Human Errors ◽  
Wrong Decision ◽  
Evaluation Matrix

Digitalized nuclear instruments and control systems have become the main stream design for the main control room (MCR) of advanced nuclear power plants (NPPs) nowadays. Digital human-system interface (HSI) could improve human performance and, on the other hand, could reduce operators’ situation awareness as well. It might cause humans making wrong decision during an emergency unintentionally. Besides, digital HSI relies on computers to integrate system information automatically instead of human operation. It has changed the operator’s role from mainly relating operational activity to mainly relating monitoring. However, if operators omit or misjudge the information on the video display units or wide display panel, the error of omission and error of commission may occur. Therefore, how to avoid and prevent human errors has become a very imperative and important issue in the nuclear safety field. This study applies Performance Evaluation Matrix to explore the potential human errors problems of the MCR. The results show that the potential problems which would probably affect to the human performance of the MCR in advanced NPPs are multiple accidents, pressure level, number of operators, and other factors such as working environmental.

Based on Human Behavior Process of Human Error Defensive Management Research for NPP

2017 ◽  
Author(s):  
Shen Yang ◽  
Geng Bo ◽  
Li Dan
Keyword(s):  
Power Plant ◽  
Nuclear Power Plant ◽  
Control Chart ◽  
Human Behavior ◽  
Nuclear Power ◽  
Human Error ◽  
Human Performance ◽  
Human Factor ◽  
Error Management ◽  
Root Cause

According to the research of nuclear power plant human error management, it is found that the traditional human error management are mainly based on the result of human behavior, the event as the point cut of management, there are some drawbacks. In this paper, based on the concept of the human performance management, establish the defensive human error management model, the innovation point is human behavior as the point cut, to reduce the human errors and accomplish a nip in the bud. Based on the model, on the one hand, combined with observation and coach card, to strengthen the human behavior standards expected while acquiring structured behavior data from the nuclear power plant production process; on the other hand, combined with root cause analysis method, obtained structured behavior data from the human factor event, thus forming a human behavior database that show the human performance state picture. According to the data of human behavior, by taking quantitative trending analysis method, the P control chart of observation item and the C control chart of human factor event is set up by Shewhart control chart, to achieve real-time monitoring of the process and result of behavior. At the same time, development Key Performance Indicators timely detection of the worsening trend of human behavior and organizational management. For the human behavior deviation and management issues, carry out the root cause analysis, to take appropriate corrective action or management improvement measures, so as to realize the defense of human error, reduce human factor event probability and improve the performance level of nuclear power plant.

APPLYING WHAT WE KNOW ABOUT HUMAN ERROR: FROM THEORY INTO PRACTICE

2004 ◽  
Vol 44 (1) ◽  
pp. 885
Author(s):  
E. Grey ◽  
P. Wilkinson
Keyword(s):  
Human Error ◽  
Petroleum Industry ◽  
Health And Safety ◽  
Management Systems ◽  
Process Systems ◽  
Error Training ◽  
Accident Causation ◽  
Applied Approach ◽  
Improve Health

Human error is often said to be at the heart of the majority of incidents and the developing discipline of human factors a way of understanding how these errors occur. There is little debate about this. But do we practise what we preach and are we reaping the benefits of applying the insights? Anecdotal evidence suggests not. Human error is too often interpreted as people being reckless, careless or just ignorant in discharging their duties. This so-called careless worker approach was the unstated assumption behind early moves to improve health and safety. It could be argued in the petroleum industry that we have adopted a more sophisticated approach, emphasising the importance of the engineering integrity of process systems and the role of formal management systems. However, there remains a need to better integrate what we know about human and organisational error. Reason’s (1997) organisational accident model has had a profound effect on how accidents are viewed and how we can learn from them. The clarity with which the model is presented does not, however, necessarily translate directly into ease of application. The model is a description of accident causation, but does not provide a method for making assessments about organisational resilience in its own right. As such, individuals wanting to use the model need to be well trained if benefits are to be realised. This paper describes a practical and applied approach to human error training based on principles of adult learning that is designed to tap into trainees’ existing knowledge and experiences.

A Framework to Assess Human Performance in Normal and Emergency Situations

2019 ◽  
Vol 6 (1) ◽  
Author(s):  
Salman Ahmed ◽  
H. Onan Demirel
Keyword(s):  
Computer Aided Design ◽  
Nuclear Power ◽  
Human Error ◽  
Human Performance ◽  
Plant Management ◽  
Early Design ◽  
Emergency Situations ◽  
Design Deficiencies ◽  
Emergency Scenarios ◽  
Digital Human

Abstract Human error is one of the primary reasons for accidents in complex industries like aviation, nuclear power plant management, and health care. Physical and cognitive workload, flawed information processing, and poor decision making are some of the reasons that make humans vulnerable to error and lead to failures and accidents. In many accidents and failures, oftentimes, vulnerabilities that are embedded in the system, in the form of design deficiencies and poor human factors, lead to latent or catastrophic failures, but the last link is a human operator who gets blamed or worse, injured. This paper introduces an early design human performance assessment framework to identify what type of digital prototyping methodologies are appropriate to detect the deviation of the operator's performance due to an emergency condition. Fire in a civilian aircraft cockpit was introduced as a performance shaping factor (PSF). Ergonomics performance was evaluated using two prototyping strategies: (1) a computational prototyping framework includes digital human modeling (DHM) and computer-aided design; and (2) a novel mixed prototyping framework includes motion capture, DHM, and virtual reality. Results showed that the mixed prototyping framework can simulate emergency scenarios with increased realism and also has the potential to incorporate subjective aspects of ergonomics outcomes, overcoming the underlying lack of design knowledge in conventional early design methodologies.

Development of Accidental Scenarios Involving Human Errors for Risk Assessment in Restricted Waters

2020 ◽  
Author(s):  
Danilo Taverna Martins Pereira de Abreu ◽  
Marcos Coelho Maturana ◽  
Marcelo Ramos Martins
Keyword(s):  
Human Error ◽  
Human Performance ◽  
Typical Feature ◽  
Performance Model ◽  
Traffic Density ◽  
Technical Failure ◽  
Event Tree ◽  
Analysis Model ◽  
Event Tree Analysis ◽  
Tree Analysis

Abstract The navigation in restricted waters imposes several challenges when compared to open sea navigation. Smaller dimensions, higher traffic density and the dynamics of obstacles such as sandbanks are examples of contributors to the difficulty. Due to these aspects, local experienced maritime pilots go onboard in order to support the ship’s crew with their skills and specific regional knowledge. Despite these efforts, several accidents still occur around the world. In order to contribute to a better understanding of the events composing accidental sequences, this paper presents a hybrid modelling specific for restricted waters. The main techniques used are the fault tree analysis and event tree analysis. The former provides a framework to investigate the causes, while the latter allows modelling the sequence of actions necessary to avoid an accident. The models are quantified using statistical data available in the literature and a prospective human performance model developed by the Technique for Early Consideration of Human Reliability (TECHR). The results include combined estimates of human error probabilities and technical failure probabilities, which can be used to inform the causation factor for a waterway risk analysis model. In other words, given that the ship encounters a potential accidental scenario while navigating, the proposed models allow computing the failure probability that of the evasive actions sequence. The novelty of this work resides on the possibility of explicitly considering dynamicity and recovery actions when computing the causation factor, what is not a typical feature of similar works. The results obtained were compared with several results available in the literature and have been shown to be compatible.

scholarly journals An Experiment on the Use of Genetic Algorithms for Topology Selection in Deep Learning

2019 ◽  
Vol 2019 ◽  
pp. 1-12 ◽  
Author(s):  
Fernando Mattioli ◽  
Daniel Caetano ◽  
Alexandre Cardoso ◽  
Eduardo Naves ◽  
Edgard Lamounier
Keyword(s):  
Genetic Algorithms ◽  
Deep Learning ◽  
Deep Neural Networks ◽  
State Of The Art ◽  
Complex Task ◽  
Trial And Error ◽  
Complex Solution ◽  
Computation Algorithms ◽  
Computational Resources ◽  
Topology Selection

The choice of a good topology for a deep neural network is a complex task, essential for any deep learning project. This task normally demands knowledge from previous experience, as the higher amount of required computational resources makes trial and error approaches prohibitive. Evolutionary computation algorithms have shown success in many domains, by guiding the exploration of complex solution spaces in the direction of the best solutions, with minimal human intervention. In this sense, this work presents the use of genetic algorithms in deep neural networks topology selection. The evaluated algorithms were able to find competitive topologies while spending less computational resources when compared to state-of-the-art methods.

Sign in / Sign up

Export Citation Format

Share Document