AGENT-BASED APPROACH TO THE DESIGN OF RBF NETWORKS

This article proposes a novel approach to the radial basis function network (RBFN) design. Its main idea is to apply the agent-based population learning algorithm to the task of initialization and training RBFNs. The approach allows for an effective network initialization and estimation of its output weights. The initialization involves two stages, where in the first one initial clusters are produced using the similarity-based procedure and next, in the second stage, prototypes (centroids) from the thus-obtained clusters are selected. The agent-based population learning algorithm is used to select prototypes. In the proposed implementation of the algorithm, both tasks—RBFN initialization and RBFN training—are carried out by a team of agents executing various local search procedures and cooperating with a view to determine the solution to the RBFN design problem at hand. The performance of the RBFN constructed using the proposed agent-based approach is analyzed and evaluated. The proposed approach is also compared with different RBFN initialization and training procedures in the literature.     

Iterative augmentation of a medical training simulator: Effects of affective metacognitive scaffolding

Experiential training simulators are gaining increasing popularity for job-related training due to their potential to engage and motivate adult learners. They are designed to provide learning experiences that are directly connected to users' work environments and support self-regulated learning. Nevertheless, learners often fail to transfer the knowledge gained in the simulated environment to real-world contexts. The EU-funded ImREAL project aimed to bridge that gap by developing a suite of intelligent services designed to enhance existing training simulators. This paper presents work that was a subset of this research project, reporting the iterative development and evaluation of a scaffolding service, which was integrated into a simulator for training medical students to perform diagnostic interviews. The study comprises three evaluation phases, comparing the pure simulator to a first version with metacognitive scaffolding and then to a final version with affective metacognitive scaffolding and enriched user modelling. The scaffolding service provides the learner with metacognitive prompts; affective elements are realized by an integrated affect reporting tool and affective prompts. Using a mixed-method approach by analysing questionnaires (N = 106) and log-data (N = 426), the effects of the services were investigated with respect to real-world relevance, self-regulated learning support, learning experience, and integration. Despite some limitations, the outcomes of this study demonstrate the usefulness of affective metacognitive scaffolding in the context of experiential training simulators; significant post-simulation increases in perceived relevance of the simulator, reflective note-taking, overall motivation, and feeling of success could be identified. Perceived usability and flow of the simulation increased, whereas overall workload and frustration decreased. However, low response rates to specific functions of the simulation point to a need to further investigate how to raise users' awareness and understanding of the provided tools, to encourage interaction with the services, and to better convey the benefits of using them. Thus, future challenges concern not so much technological developments for personalizing learning experiences, but rather new ways to change user attitudes towards an open approach to learning systems that enables them to benefit from all offered features.     

A mobile learning system for scaffolding bird watching learning

Abstract This paper develops a mobile learning system for scaffolding students learning about bird-watching. The aim is to construct an outdoor mobile-learning activity using up-to-date wireless technology. The proposed Bird-Watching Learning (BWL) system is designed using a wireless mobile ad-hoc network. In the BWL system, each learner has a PDA (Personal Digital Assistant) with a Wi-Fi-based (IEEE 802.11b) wireless network card. The BWL system offers a mobile learning system which supports the students learning through scaffolding. The aim of a formative evaluation was twofold: to explore the possible roles and scaffolding aids that the mobile learning device offers for bird-watching activities and to investigate whether student learning benefited from the mobility, portability, and individualisation of the mobile learning device.     

Predicting false negative errors in digital breast tomosynthesis among radiology trainees using a computer vision-based approach

PurposeDigital breast tomosynthesis (DBT) can improve lesion visibility in comparison to mammography by eliminating breast tissue superimposition. While the benefits of DBT in breast cancer screening rely on well trained radiologists, the optimal training regimen in DBT is unknown. We propose a computer-aided educational system that individually selects the optimal training cases for each trainee. The first step towards this goal is to capture the individual weaknesses of each trainee. In this study, we present and evaluate a computer algorithm for this purpose with particular focus on false negative errors.MethodsWe developed an algorithm (a user model) that predicted the likelihood of a trainee missing an abnormal location. An individual model is applied for each trainee. The algorithm consists of three steps. First, the lesions on DBT images are segmented by a 3D active contour method with a level set algorithm. Then, 16 features are extracted automatically for the segmented lesions. Finally a multivariate logistic regression classifier predicts the likelihood of error based on the extracted features. The classifier is trained using the previous interpretation data of the trainee. We evaluated the individual predictive algorithms experimentally using data from a reader study in which 29 trainees and 3 expert breast radiologists read 60 DBT cases. Receiver operating characteristic (ROC) analysis, along with a repeated holdout approach, was used to evaluate the predictive performance of our algorithm.ResultsThe average area under the ROC curve (AUC) of the algorithms which predicted which lesions will be detected and which will be missed by a specific trainee was 0.627 (95% CI: 0.579–0.675). The average performance was statistically significantly better than chance (p<0.001). Under the status quo, training involves no specific strategy for case presentation, and this random behavior corresponds to AUC of 0.5. Therefore, the proposed algorithm may provide a significant improvement in distinguishing abnormal locations that will be detected by a trainee from those that will be missed.ConclusionsOur algorithm was able to distinguish abnormal locations that will be detected by a trainee from those that will be missed. This could be used to enrich the training set with cases that are likely to prompt error for the individual trainee while still maintaining a range of cases necessary for comprehensive education.     

An agent-based learning framework for modeling microbial growth

The overall idea of this paper is to study the intelligent behavior of microbes in a binary substrate environment with agent-based learning models. Study of microbial growth enables understanding of industrially relevant processes such as fermentation, biodegradation of pollutants, antibody production using hybridoma cells, etc. Artificial intelligence techniques such as genetic algorithms and agent-based learning methodologies have been used to study microbial growth. Specifically, the objective is to (1) qualitatively model the intelligent growth characteristics of the microbes using a minimal set of generic rules as against algebraic/differential mathematical relationships and (2) propose a suitable hypothesis that explains the origin of intelligence through learning in the microbes. A microbial cell has been modeled as a collection of agents characterized by a set of resources and an objective to survive and grow. The actions of the agents are governed by generic rules such as survival, growth and division as is common for any individual in a resource-limited competitive environment. The interaction of the agents with the environment and other fellow agents enables them to “learn” and “adapt” to the changes in the environment and thus defines the dynamics of the system. The origin of intelligence in the microbes has been studied by both a simple learning rule of imitation and rule discovery studies.     

Supporting synchronous collaborative learning: A generic, multi-dimensional model

Future CSCL technologies are described by the community as flexible, tailorable, negotiable, and appropriate for various collaborative settings, conditions and contexts. This paper describes the key design issues of a generic synchronous collaborative learning environment, called Omega+. In this approach, model-based genericity is applied to the four dimensions of collaborative learning: the situation, the interaction, the process, and the way of monitoring individual and group performance. These four aspects are explicitly specified in a set of models that serve as parameters for the generic environment. This opens the possibility of combining many structuring/scaffolding techniques that have been proposed in isolation in the CSCL literature. The paper also emphasizes the specificities and difficulties of evaluating a comprehensive generic support approach. Experimental evaluations conducted by system designers generally isolate the effects of a particular design feature on learning. This kind of evaluation can hardly demonstrate the usefulness of a generic model at the global level and the feasibility of system customization by non-specialist teachers. To address these difficulties, Omega+ is integrated into a larger collaborative web platform dedicated to CSCL practice, evaluation (by collecting anonymized logs), and dissemination (by supporting the technical and pedagogical development of teachers).     

Does the emotional design of scaffolds enhance learning and motivational outcomes in game-based learning?

Background

In recent years, the importance of emotions in learning has been increasingly recognized. Applying emotional design to induce positive emotions has been considered a means to enhance the instructional effectiveness of digital learning environments. However, only a few studies have examined the specific effects of emotional design in game-based learning.

Objectives

This quasi-experimental study utilized a value-added research approach to investigate whether emotional design applied to scaffolding in a game-based learning environment improves learning and motivational outcomes more than emotionally neutral scaffolding.

Methods

A total of 138 participants, mean age of 11.5 (SD = 0.73) participated in the study. A total of 68 participants played the base version of a fraction learning game (Number Trace), where scaffolding was provided with emotionally neutral mathematical notations, and 70 participants played the value-added version of the game using emotionally designed animated scaffolding agents. Pre-and post-tests were used to measure conceptual fraction knowledge and self-reported measures of situational interest and situational self-efficacy to evaluate motivational outcomes.

Results and Conclusions

Our results indicate that the emotional design applied to scaffolds can improve the educational value of a game-based learning environment by enhancing players' situational interest and situational self-efficacy. However, although the intervention improved the participants' conceptual fraction knowledge, there was no significant difference between the scaffolding conditions in participants' learning outcomes.

Takeaways

The results suggest that emotional design can increase the educational impact of game-based learning by promoting the development of interest, as well as improving self-efficacy.     

A computer-aided myoelectric training system for young upper limb amputees

The techniques employed in most clinical centres for training amputees to use myoelectric prostheses are quite similar. Training aids employing meters, lights or modified toys as indicators of signal level are routinely used for control site selection and for ‘signals training’.It is clear from local experience and from the literature that the difficulty of maintaining motivation rather than any other criterion is the limiting factor determining the duration and effectiveness of signals training. This is especially true for the case of children in the age range of 5–14 years.In an attempt to provide a training aid for young amputees, which maintains user motivation, the Institute of Biomedical Engineering at UNB has developed a system based on ‘computer games’. The rationale for this project derives from the observation that computer games have a universal fascination for all ages, and may provide the stimulus and excitement necessary to motivate the trainee.A custom hardware board has been designed which interfaces with an IBM PC type machine, which allows both conventional joystick input and two channels of myoelectric signal (MES) to be used as control sources for video game software. Consequently, the system is suitable for both single-site and two-site signal training. This arrangement also permits a comparison to be made concerning the activity of the sound limb versus the prosthetic side.The video game software has been custom designed so that subject evaluation routines can be run in the background. In this way the therapist can keep track of user performance and have a permanent record of the training session. The games have also been designed so that the user must use both sound and prosthetic limbs which in turn promotes the development of ‘two-handed’ activities.     

Toward a real-time model-based training system

This article describes the development of a real-time model-based training system that provides adaptive “over-the-shoulder” (OTS) instructions to trainees as they learn to perform an Anti-Air Warfare Coordinator (AAWC) task. The long-term goal is to develop a system that will provide real-time instructional materials based on learners’ actions, so that eventually the initial set of instructions on a task can be strengthened, complemented, or overridden at different stages of training. The training system is based on the ACT-R architecture, which serves as the theoretical background for the cognitive model that monitors the learning process of the trainee. An experiment was designed to study the impact of OTS instructions on learning. Results showed that while OTS instructions facilitated short-term learning, (a) they took time away from the processing of current information, (b) their effects tended to decay rapidly in initial stages of training, and (c) their effects on training diminished when the OTS instructions were proceduralized in later stages of training. A cognitive model that learned from both the upfront and OTS instructions was created and provided good fits to the learning and performance data collected from human participants. Our results suggest that to fully capture the symbiotic performance between humans and intelligent training systems, it is important to closely monitor the learning process of the trainee so that instructional interventions can be delivered effectively at different stages of training. We proposed that such a flexible system can be developed based on an adaptive cognitive model that provides real-time predictions on learning and performance.     

An intelligent mobile vehicle navigator based on fuzzy logic andreinforcement learning

In this paper, an alternative training approach to the EEM-based training method is presented and a fuzzy reactive navigation architecture is described. The new training method is 270 times faster in learning speed; and is only 4% of the learning cost of the EEM method. It also has very reliable convergence of learning; very high number of learned rules (98.8%); and high adaptability. Using the rule base learned from the new method, the proposed fuzzy reactive navigator fuses the obstacle avoidance behaviour and goal seeking behaviour to determine its control actions, where adaptability is achieved with the aid of an environment evaluator. A comparison of this navigator using the rule bases obtained from the new training method and the EEM method, shows that the new navigator guarantees a solution and its solution is more acceptable.     

Virtual CEOs: A blended approach to digital gaming for enhancing higher order thinking and academic achievement among vocational high school students

The requirements of a contemporary workplace include the ability to think critically and creatively in order to solve problems and respond to changes in economic and social conditions. Unfortunately, vocational education often fails to prepare graduates for this environment due to limited resources, low student motivation, or the reliance upon outdated instructional strategies. The use of digital game-based learning (DGBL) for vocational education has been proposed, but has yet to be effectively implemented, particularly in terms of the promotion of higher order thinking skills (HOTS). Data from 68 eleventh grade vocational high school students were evaluated after a quasi-experimental, 27 week intervention. Pretest and posttest results were evaluated by MANCOVA and demonstrated that the experimental group (blended DGBL incorporating integrative HOTS activities) outperformed the comparison group (technology enhanced learning) in terms of creative thinking, critical thinking, problem solving, and academic achievement, with significant improvements on all four measures. While technology-enhanced learning was effective in promoting academic achievement and creative thinking, the DGBL condition was deemed most effective in providing an authentic context for developing employment-related skills and knowledge. Based on these results, a blended approach for DGBL, which incorporates instructor orchestration and scaffolding, provision of learning aids, and the use of collaborative learning, is recommended, particularly for vocational learners. This paper provides examples of a concrete model of DGBL instruction that was verified empirically as successful in significantly improving all three higher order thinking skills, including creative thinking, critical thinking, and problem solving.     

A co-creation blended KM model for cultivating critical-thinking skills

Both critical thinking (CT) and knowledge management (KM) skills are necessary elements for a university student’s success. Therefore, this study developed a co-creation blended KM model to cultivate university students’ CT skills and to explore the underlying mechanisms for achieving success. Thirty-one university students participated in this study. Findings from the 17-week training program suggest that scaffolding university students through knowledge sharing, internalization, and co-creation processes in a blended KM environment can effectively enhance their CT skills. Moreover, the attribute–treatment interaction (ATI) analysis suggests that judicial thinking style which relates to a deep learning approach may facilitate KM and help improve CT skills. Notably, the complex underlying mechanisms and paths of influence found in this study attest to the highly dynamic nature of the proposed KM processes.     

An integrated logistics support system for training crew medical officers in advanced cardiac life support management.

The purpose of this paper is to present an approach for the development of an integrated logistics support system for training space flight crew medical officers in advanced cardiac life support (ACLS). The most practical approach to deal with this problem has been determined to be the training of one or more of the crew members on the mission as crew medical officers (CMO's). If this approach is taken, there are a number of problems that must be addressed. The basic approach presented in this paper is to develop a performance quality index for ACLS tasks and use this index to monitor and control the performance of crew members throughout their tenure as CMO's. The control tool of the system is based on an integrated learning and forgetting model used to forecast CMO performance level at a given point during flight training. The model represents an aid for trainers in determining a training regime and maintaining the performance standards. A performance evaluator and trainer is also developed to help in the establishment of trainee performance level during training or retraining. All of these tools were evaluated using either expert opinion questionnaires or experimental results. In conclusion, the results presented provide the tools required for an integrated logistics support system for training ACLS personnel.     

Architecture layers and engineering approach for agent-based system

It is necessary to support user-centric service provision paradigm in distributed, dynamic and complex computing environment. Software agent technology is considered as one of the technologies suitable to adopt such computing environment. Many researchers have emphasized on agent-based system development, but, many agent-based systems are designed and constructed in ad hoc. In particular, they do not enough consider system organization and performance aspects. More systematic engineering approach of agent-based system is required. We propose the layered architecture and engineering approach for agent-based system design. We devise the layers necessary to design agent-based system, and methods to engineer each layer. Also we show that the devised approach can be used to design agent-based system and analyze system features. The layered architecture and engineering approach of agent-based system proposed in this paper support that engineer designs efficient agent-based system.     

A comparison of approaches to learning task selection in the training of complex cognitive skills

This paper presents a comparison of learning task selection approaches that have been used throughout the last three decades in the training of complex cognitive skills. In general, a development from static part-task selection to dynamic whole-task selection can be noticed. The four approaches of static part-task approaches, static whole-task approaches, dynamic part-task approaches, and dynamic whole-task approaches are identified and compared in terms of their flexibility and adaptability to the needs of the individual trainee during training. The comparison shows that dynamic whole-task approaches are the most flexible and adaptive. For each approach it is discussed to what complex cognitive skills they might be useful training methods.     

Reusing back-propagation artificial neural networks for land cover classification in tropical savannahs

This paper focuses on a method to overcome some of the disadvantages that are related with the use of artificial neural networks (ANNs) as supervised classifiers. The proposed method aims at speeding up network learning, improving classification accuracies and reducing variability on classification performance due to random weight initialization. This can be realized by transferring implicit knowledge from a previously learned source task to a new target task using the proposed algorithm, Discriminality Based Transfer (DBT). The presented approach is compared with conventional network training and a literal transfer method in a 13-class tropical savannah classification experiment using Landsat Thematic Mapper (TM) data. Knowledge was extracted from a network trained on the Kara experimental site in Togo. This information was used to classify the Savanes-L'Oti area which differs in terms of geographical position, image acquisition date, climatological condition and land cover. It was possible to speed up network learning 5.2, 4.3 and 1.8 times using, respectively, 5-, 10- and 20-pixels-per-class training sets. Larger training sets showed less speed improvement. After applying DBT, average classification accuracies were not significantly different from accuracies obtained after training random initialized networks, although DBT tended to show better performance on smaller training sets. It was possible to explain differences in individual class accuracies by analysing Bhattacharyya (BH) distances calculated between all Kara and Savanes-L'Oti classes. Finally, variability on classification performance decreased significantly when training with 5-, 10- and 20-pixels-per-class training sets after DBT application.     

Leveraging Graph Cut’s Energy Function for Context Aware Facial Recognition in Indoor Environments

Context-aware facial recognition regards the recognition of faces in association with their respective environments. This concept is useful for the domestic robot which interacts with humans when performing specific functions in indoor environments. Deep learning models have been relevant in solving facial and place recognition challenges; however, they require the procurement of training images for optimal performance. Pre-trained models have also been offered to reduce training time significantly. Regardless, for classification tasks, custom data must be acquired to ensure that learning models are developed from other pre-trained models. This paper proposes a place recognition model that is inspired by the graph cut energy function, which is specifically designed for image segmentation. Common objects in the considered environment are identified and thereafter they are passed over to a graph cut inspired model for indoor environment classification. Additionally, faces in the considered environment are extracted and recognised. Finally, the developed model can recognise a face together with its environment. The strength of the proposed model lies in its ability to classify indoor environments without the usual training process(es). This approach differs from what is obtained in traditional deep learning models. The classification capability of the developed model was compared to state-of-the-art models and exhibited promising outcomes.     

Identification of metaphors for virtual environment training systems

The objective of this effort was to develop potential metaphors for assisting wayfinding and navigation in current virtual environment (VE) training systems. Although VE purports a number of advantages over traditional, full-scale simulator training devices (deployability, footprint, cost, maintainability, scalability, networking), little design guidance exists beyond individual instantiations with specific platforms. A review of metaphors commonly incorporated into human—computer interactive systems indicated that existing metaphors have largely been used as orientation aids, mainly in the form of guided navigational assistance, with some position guidance. Advanced metaphor design concepts were identified that would not only provide trainees with a useful orienting framework but also enhance visual access and help differentiate an environment. The effectiveness of these concepts to aid navigation and wayfinding in VEs must be empirically validated.     

Identification of metaphors for virtual environment training systems

The objective of this effort was to develop potential metaphors for assisting wayfinding and navigation in current virtual environment (VE) training systems. Although VE purports a number of advantages over traditional, full-scale simulator training devices (deployability, footprint, cost, maintainability, scalability, networking), little design guidance exists beyond individual instantiations with specific platforms. A review of metaphors commonly incorporated into human-computer interactive systems indicated that existing metaphors have largely been used as orientation aids, mainly in the form of guided navigational assistance, with some position guidance. Advanced metaphor design concepts were identified that would not only provide trainees with a useful orienting framework but also enhance visual access and help differentiate an environment. The effectiveness of these concepts to aid navigation and wayfinding in VEs must be empirically validated.     

Intelligent trainee behavior assessment system for medical training employing video analysis

This paper addresses the problem of assessing a trainee’s performance during a simulated delivery training by employing automatic analysis of a video camera signal. We aim at providing objective statistics reflecting the trainee’s behavior, so that the instructor is able to give valuable suggestions after the training. The basic idea is to analyze the moving and location parameters of the trainee, on which the behavior of the trainee can be judged and also compared. Our system consists of three major steps. In the first step, we label specific pixels with a given color, based on a Gaussian model. In the second step, the mean shift (MS) algorithm is employed to find the densest region of a color, where the center of that region indicates the center of a medical cap worn by a trainee. To accelerate the convergence of the MS algorithm, we propose to combine the distribution sampling and on-line mode updating based on the pyramid sampling technique. In the last step, we assume that the cap’s position represents the position of a trainee. Therefore, several statistics, such as the moving trajectory and the total movement of each trainee, can be calculated. These statistics associated with the domain knowledge, help us to determine trainees’ teamwork. Our system also enables an interactive way for instructors to choose the interested individual trainee, and then provides more results of him. Experimental evaluations using real delivery training videos demonstrate the effectiveness of the proposed work.¹