Deep reinforcement learning (DRL) is increasingly applied to de novo molecular design, but choices in data, rewards, and evaluation can yield uneven performance across disease areas and chemotypes. Despite this, there is no concise synthesis of how fairness is defined, measured, and tested in DRL-based drug discovery. In this rapid evidence review, we synthesize fairness definitions and metrics for DRL-driven molecule generation in healthcare. We focus on three questions: (i) how dataset composition and split strategies, especially scaffold versus random splits, affect evaluation and distribution shift; (ii) how reward design (e.g., QED, docking, toxicity, synthetic accessibility) can create or mitigate bias, with emphasis on cancer targets; and (iii) which measurable metrics best capture fairness. This includes parity across cancer versus non-cancer indications and across cancer subtypes. It also includes distributional balance in key physicochemical descriptors, scaffold/chemotype diversity, groupwise validity, toxicity, and synthetic accessibility. From 2017 onward, we searched major biomedical, computer science, and engineering literature databases and used arXiv for horizon scanning. Records were screened using PRISMA-style procedures and analyzed via content coding to link reported parity outcomes to dataset and reward choices. Our review provides a concise set of fairness definitions and metrics for DRL molecule generation. It offers practical guidance for reporting distribution parity and outcome parity. It also summarizes how dataset and reward choices relate to observed parity effects and identifies open gaps relevant to trustworthy, cancer-relevant DRL generation.

Algorithms that support screening, triage, and treatment decisions depend on training data drawn from patient populations. Limited access to patient-level records across institutions and jurisdictions can reduce representation and contribute to uneven model performance across populations. Canada's federated health system, where provinces and territories manage separate datasets and privacy regimes, limits multicenter medical AI research. We conducted a scoping review to map how Canadian researchers share patient-level data in multicenter medical AI collaborations. We searched PubMed, IEEE Xplore, ACM Digital Library, Scopus, and Web of Science from 2018 to February 2025 and implemented a human-in-the-loop large language model process to support screening and extraction, with reviewer validation. Among 3100 included studies, 160 reported multicenter patient-level data collection. Centralized pooling dominated this subset, with 95% of studies using centralized storage and 5% (n = 8) reporting decentralized approaches, including federated learning, sequential model transfer, and distributed feature sharing. Governance requirements were frequently described as multi-site and sequential, and 81.8% of multicenter collaborations reported parallel ethics approvals from three or more institutional review boards. Only one decentralized collaboration operated entirely within Canada. International partnerships comprised 80% of multicenter studies, and many cohorts included non-Canadian sites or non-Canadian data. Our findings support adoption of distributed model development protocols and interoperable governance that limit central pooling while enabling consistent training, validation, and reporting across sites, as only 1 of 160 multicenter studies reported a decentralized approach with Canadian patient data only.

Ultra-Wideband (UWB) indoor positioning systems suffer significant accuracy degradation in Non-Line-of-Sight (NLoS) conditions, where multipath distortions in the Channel Impulse Response (CIR) lead to biased range estimates. While deep learning (DL) models have shown potential in predicting such errors from CIR data, most prior studies focus on isolated architectures and static pipelines, without evaluating their broader integration into full positioning systems. This work presents a comparative study of three DL models—Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory networks (BiLSTM), and Transformer—for CIR-based range error prediction, integrated into both Weighted Least Squares (WLS) and Extended Kalman Filter (EKF) frameworks. In WLS, predicted errors are used as adaptive weights to suppress unreliable measurements; in EKF, they dynamically scale the measurement noise covariance for more accurate filtering. We evaluate all models across four realistic indoor environments using a public UWB dataset. Our results show that Transformer-EKF achieves the best performance, reducing mean positioning error to 0.59m in Environment 2 (severe NLoS/multipath) while maintaining real-time inference at 0.059 milliseconds per position. These findings establish a comprehensive benchmark for learning-based UWB positioning and demonstrate the value of fusing data-driven range correction with model-based tracking.

This paper provides an overview of our open (Software Engineering Practice and Education) SEPE 8K dataset which is made of 40 different 8K (8192 x 4320) video sequences and 40 variant 8K (8192 x 5464) images. The video sequences were captured at a framerate of 29.97 frames per second (FPS) and had been encoded into videos using AVC/H.264, HEVC/H.265, and AV1 codecs at resolutions from 8K to 480p. The images, video sequences, encoded videos, and various other statistics related to the media that make the dataset are stored online, published, and maintained on the repo on GitHub for non-commercial use. In this paper, the dataset components are described and analyzed using various methods. The proposed dataset is - as far as we know - the first to publish true 8K natural sequences; thus, it is important for the next level of applications dealing with multimedia such as video quality assessment, super-resolution, video coding, video compression, and many more.

Improvement of traffic signal control (TSC) efficiency has been found to lead to improved urban transportation and enhanced quality of life. Recently, the use of reinforcement learning (RL) in various areas of TSC has gained significant traction; thus, we conducted a systematic literature review as a systematic, comprehensive, and reproducible review to dissect all the existing research that applied RL in the network-level TSC domain, called as RL in NTSC or RL-NTSC for brevity. The review only targeted the network-level articles that tested the proposed methods in networks with two or more intersections. This review covers 160 peer-reviewed articles from 30 countries published from 1994 to March 2020. The goal of this study is to provide the research community with statistical and conceptual knowledge, summarize existence evidence, characterize RL applications in NTSC domains, explore all applied methods and major first events in the defined scope, and identify areas for further research based on the explored research problems in current research. We analyzed the extracted data from the included articles in the following seven categories: (i) publication and authors’ data, (ii) method identification and analysis, (iii) environment attributes and traffic simulation, (iv) application domains of RL-NTSC, (v) major first events of RL-NTSC and authors’ key statements, (vi) code availability, and (vii) evaluation. This paper provides a comprehensive view of the past 26 years of research on applying RL to NTSC. It also reveals the role of advancing deep learning methods in the revival of the research area, the rise of using non-commercial microscopic traffic simulators, a lack of interaction between traffic and transportation engineering practitioners and researchers, and a lack of proposal and creation of testbeds which can likely bring different communities together around common goals.

Cloud computing provisions scalable resources for high performance industrial applications. Cloud providers usually offer two types of usage plans: reserved and on-demand. Reserved plans offer cheaper resources for long-term contracts while on-demand plans are available for short or long periods but are more expensive. To satisfy incoming user demands with reasonable costs, cloud resources should be allocated efficiently. Most existing works focus on either cheaper solutions with reserved resources that may lead to under-provisioning or over-provisioning, or costly solutions with on-demand resources. Since inefficiency of allocating cloud resources can cause huge provisioning costs and fluctuation in cloud demand, resource allocation becomes a highly challenging problem. In this paper, we propose a hybrid method to allocate cloud resources according to the dynamic user demands. This method is developed as a two-phase algorithm that consists of reservation and dynamic provision phases. In this way, we minimize the total deployment cost by formulating each phase as an optimization problem while satisfying quality of service. Due to the uncertain nature of cloud demands, we develop a stochastic optimization approach by modeling user demands as random variables. Our algorithm is evaluated using different experiments and the results show its efficiency in dynamically allocating cloud resources.

We consider the problem of representing a multiple-criteria (i.e., multiple heterogeneous measurements) object by a single value that we can use to compare and rank different objects. An intrinsic characteristic of the multiple-criteria is their different nature (e.g., high quality and low price), and thus, the ranking process is vastly dependent on the decision-makers' preferences and viewpoints. The different criteria denote a severe problem to find an overall value to represent the trade-offs of an object.For example, is it possible to represent the different criteria of a car by a single number and utilize this number to rank different cars in single and multiple decision-makers settings? To answer this question, we extend our proposed method to calculate a weight vector of the Ordered Weighted Average (OWA) operator based on the Laplace distribution [1] and use it to illustrate how to rank a dataset of used cars, and we compare the results with six other OWA operators. In this paper, we prove the characteristics of the new operator and illustrate its benefits in single and multiple decision-making settings. Finally, to find out how well the new OWA operator can represent the information per object; we employ the values produced by the new OWA in a regression model to estimate the used car price.

In many software development projects, analysts are required to deal with systems' requirements from unfamiliar domains. Familiarity with the domain is necessary in order to get full leverage from interaction with stakeholders and for extracting relevant information from the existing project documents. Accurate and timely extraction and classification of requirements knowledge support analysts in this challenging scenario. Our approach is to mine real-time interaction records and project documents for the relevant phrasal units about the requirements related topics being discussed during elicitation. We propose to use both generative and discriminating methods. To extract the relevant terms, we leverage the flexibility and power of Weighted Finite State Transducers (WFSTs) in dynamic modelling of natural language processing tasks. We used an extended version of Support Vector Machines (SVMs) with variable-sized feature vectors to efficiently and dynamically extract and classify requirements-related knowledge from the existing documents. To evaluate the performance of our approach intuitively and quantitatively, we used edit distance and precision/recall metrics. We show in three case studies that the snippets extracted by our method are intuitively relevant and reasonably accurate. Furthermore, we found that statistical and linguistic parameters such as smoothing methods, and words contiguity and order features can impact the performance of both extraction and classification tasks.

Scenarios are commonly used for software requirements modeling. Scenarios describe how system components, users and the environment interact to complete the system functionality. However, several scenarios are needed to represent a complete system behavior and combining the scenarios may generate implied scenarios (IS) that are associated with some unexpected behavior. The unexpected behavior is commonly known as Emergent Behavior (EB), which is not evident in the requirements and design phase but may degrade the quality of service and/or cause irreparable damage during execution. Detecting and fixing EB/IS in the early phases can save on deployment cost while minimizing the run-time hazards. In this paper, we present a machine learning approach to model and identify the interactions between system components and verify which interactions are safe and which may lead to EB/IS. The experimental result shows that our approach can efficiently detect different types of EB/IS and applicable to large scale systems.

In the constantly evolving world of software development, switching back and forth between tasks has become the norm. While task switching often allows developers to perform tasks effectively and may increase creativity via the flexible pathway, there are also consequences to frequent task-switching. For high-momentum tasks like software development, "flow", the highly productive state of concentration, is paramount. Each switch distracts the developers' flow, requiring them to switch mental state and an additional immersion period to get back into the flow. However, the wasted time due to time fragmentation caused by task switching is largely invisible and unnoticed by developers and managers. We conducted a survey with 141 software developers to investigate their perceptions of differences between task switching and task interruption and to explore whether they perceive task switchings as disruptive as interruptions. We found that practitioners perceive considerable similarities between the disruptiveness of task switching (either planned or unplanned) and random interruptions. The high level of cognitive cost and low performance are the main consequences of task switching articulated by our respondents. Our findings broaden the understanding of flow change among software practitioners in terms of the characteristics and categories of disruptive switches as well as the consequences of interruptions caused by daily meetings.

This paper addresses the problem of long term location prediction for collision avoidance in Connected Vehicle (CV) environment where more information about the road and traffic data is available through vehicle-to-vehicle and vehicle-to-infrastructure communications. Gaussian Process Regression (GPR) is used to learn motion patterns from historical trajectory data collected with static sensors on the road. Trained models are then shared among the vehicles through connected vehicle cloud. A vehicle receives information, such as Global Positioning System coordinates, about nearby vehicles on the road using inter-vehicular communication. The collected data from vehicles together with GPR models received from infrastructure are then used to predict the future trajectories of vehicles in the scene. The contributions of this work are twofold. First, we propose the use of GPR in CV environment as a framework for long term location prediction. Second, we evaluate the effect of pre-analysis of training data via clustering in improving the trajectory pattern learning performance. Experiments using real-world traffic data collected in Los Angeles, California, US show that our proposed method improves prediction accuracy compared to the baseline kinematic models.

Distributed software systems (DSS) are usually open-ended systems used in different domains such as robotics, energy, health, etc. Multi-agent system (MAS) are a sub-class of DSS.In DSS, maintaining consistency between the system iterations is a complex and expensive task that requires coping with requirements changes and systems upgrading.The interactions, complexity and decentralized communication between components of the DSS may emerge an unwanted behavior.An unwanted behavior, known as Emergent Behavior (EB) or Implied Scenario (IS), could lead to irreversible damages.Thus, detecting IS at an early stage of the system development is needed to decrease the cost of maintaining the system.This work focuses on verification of DSS that its requirements modeled using Message Sequence Chart (MSC).The system verification focuses on the detection of IS using two already proposed different approaches.This article presents the combination of the two approaches by improving the usability of the tool presented in the first approach and the catalogue presented in the second approach.This combination allows the detection of new implied scenarios not detected using the cited approaches separately.

The area of Traffic Management (TM) is characterized by uncertainty, complexity, and imprecision. The complexity of software systems in the TM domain which contributes to a more challenging Requirements Engineering (RE) job mainly stems from the diversity of stakeholders and complexity of requirements elicitation in this domain. This work brings an interactive solution for exploring functional and non-functional requirements of software-reliant systems in the area of traffic management. We prototyped the RETTA tool which leverages the wisdom of the crowd and combines it with machine learning approaches such as Natural Language Processing and Naïve Bayes to help with the requirements elicitation and classification task in the TM domain. This bridges the gap among stakeholders from both areas of software development and transportation engineering. The RETTA prototype is mainly designed for requirements engineers and software developers in the area of TM and can be used on Android-based devices.

This paper represents the measurement of the heat/air-conditioning energy consumption in a large open area using electronic technologies. This work is scoping the measurement and analyses of the energy-saving possibility for large open areas such as critical care units in hospitals that use central heat/AC systems. Zonal control is performed by closing the heat/AC Vents in the unoccupied sections of the open areas during the day. The heat/AC ON time is measured using a monitor circuit, which senses outdoor and indoor temperatures, according to the preferred temperature of the user. Factors that promote open-area zonal heat control systems include personal preferences, the area's dimensions and shape, distribution of Vents, the location of windows and doors, and direct sun exposure. Over a three-and-a-half-month test period in summer and fall, measurement results proved that the zonal control system of the open area could decrease average heating energy from 25.5% with all Vents opened to 18.3% with half of the Vents opened, with a difference of 7.2%, which amounts to 28.2% energy saving. The savings for air-conditioning are 13.57%. The saving varies depending on the temperature of the weather. According to our analysis, the zonal control has a negligible effect on the loading of the machine and on the comfort of the customers. These work results confirm the proposed idea of saving heat/AC energy by controlling the delivery of air on a zone-by-zone basis even in large open areas.

Cloud computing is a new era of computing that offers resources and services for Web applications. Selection of optimal cloud resources is the main goal in cloud resource allocation. Sometimes, customers pay more than required since cloud providers' pricing strategy is designed for the interest of the providers. Nonetheless, cloud customers are interested in selecting cloud resources to meet their quality of service (QoS) requirements. Thus, for the interest of both providers and customers, it is vital to balance the two conflicting objectives of deployment cost and QoS performance. In this paper, we present a cost-effective and runtime friendly algorithm that minimizes the deployment cost while meeting the QoS performance requirements. In other words, the algorithm offers an optimal choice, from customers' point of view, for deploying a Web application in cloud environment. The multi-objective optimization algorithm minimizes cost and maximizes QoS performance simultaneously. The proposed algorithm is verified by a series of experiments on different workload scenarios deployed in two distinct cloud providers. The results show that the proposed algorithm finds the optimal combination of cloud resources that provides a balanced trade-off between deployment cost and QoS performance in relatively low runtime.

Predicting Web service response time percentiles is often an important aspect of service level management exercises. Existing techniques can be very time consuming since they involve the manual construction of complex analytic or simulation models. To address this problem, we propose Prospective, a fully automated and data-driven approach for predicting Web service response time percentiles. Prospective relies on historical response time data collected from a Web service. Given a specification for workload expected at the Web service over a planning horizon, Prospective uses this historical data to offer predictions for response time percentiles of interest. At the core of Prospective is a lightweight simulator that uses collaborative filtering to estimate response time behaviour of the service based on behaviour observed historically. Results show that Prospective is able to predict various response time percentiles of interest with high accuracy for a wide variety of workloads.

Accurate effort estimation of software development projects plays a key role in project success.However, it is still a challenge activity to researchers and practitioners because of the nature of software products and dynamics in software industry and development environment.Artificial neural network (ANN) is as an effective method and has been widely used in various areas of software engineering.This paper proposes a new effort estimation method based on clustering and ANN ensembles.The contribution of the paper is twofold.First, the impact of clustering projects on the estimation accuracy is investigated.Second, the impact of using ANN ensembles instead of a single ANN is also investigated.The proposed method includes three phases called pre-processing, k-means clustering, and ANN ensembles effort estimation.The method starts with exploring the historical projects dataset.Afterward, k-means is used to cluster the projects.Finally, the proposed method as well as two other estimation methods (i.e. a single ANN and expert-based) were applied to the created clusters and results were compared using MMRE and PRED measures.The simulation results show that the proposed method significantly outperforms the two other estimation methods.

This paper measures the heat energy consumption in a typical Canadian house taking into account a commonly used forced air gas furnace. The aim of this work is to measure and analyze the possible energy savings for homes if zonal heat control system is used. A control circuit is used to sense the indoor and outdoor temperature and turns the furnace ON or OFF according to the preferred temperature set by the householders. Zonal heat control is performed by closing the heating vents in the unoccupied areas of the house during the nighttime. Over one-month test period in winter, the measurement results proved that the zonal control system could save the heating energy up to 36% depending on the weather temperature. The study results support the assumption that, more saving in energy consumption can be reached by controlling the delivery of heat on a zone-by-zone basis.

Cloud computing provides a reliable and cost- effective setting for deploying large-scale web applications. However, choosing and configuring an appropriate cloud Infrastructure-as-a-Service (IaaS), e.g., the appropriate database and computing instances and acceptable service rates, is a daunting task. The task is also challenging when trying to optimize the IaaS for conflicting objectives such as performance and cost. Furthermore, due to lack of understanding of the pricing model and the cloud IaaS, a cloud consumer may pay more than necessary or may not fully utilize the purchased resources. For this reason, we propose an algorithm that suggests the most cost-effective configuration meeting the QoS requirements and budget constraint. In contrast to existing cost optimization proposals, our proposed algorithm maps the minimum requirements of the to- be-deployed web application to deployment costs according to the price model set by cloud providers. The algorithm also considers QoS requirements for different resource types in the cloud, namely, database servers, computing servers, storage, and service rate. The proposed algorithm is evaluated by a series of experiments on a web application with seven different workload scenarios. The experimental results show the effectiveness of the proposed algorithm in achieving a solution with the minimum deployment cost for each scenario while satisfying all customer's requirements.

Web services have increasingly begun to rely on public cloud platforms. The virtualization technologies employed by public clouds can however trigger contention between virtual machines (VMs) for shared physical machine (PM) resources thereby leading to performance problems for the Web service. Past studies have exploited PM level performance metrics such as Clock Cycles Per Instruction to detect such platform induced performance interference. Unfortunately, public cloud customers do not have access to such metrics. They can typically only access VM-level metrics and application level metrics such as transaction response times and such metrics alone are often not useful for detecting inter-VM contention. This poses a difficult challenge to Web service operators for detecting and managing platform induced performance interference issues inside the cloud. We propose a machine learning based interference detection technique to address this problem. The technique applies collaborative filtering to predict whether a given transaction being processed by a Web service is suffering adversely from interference. The results can then be used by a management controller to trigger remedial actions, e.g., reporting problems to the system manager or switching cloud providers. Results using a realistic Web benchmark show that the approach is effective. The most effective variant of our approach is able to detect about 96% of performance interference events with almost no false alarms.

Fatigue induced vehicle accidents have seen an increase in the last few decades.Fatigue monitoring using noninvasive and real time image processing and computer vision techniques have shown great promise and are an active research area.To that extent, in the proposed work a blink detection algorithm is proposed that serves as a visual cue that may be correlated to the state of fatigue of the driver.Using a complimentary but independent approach, shape analysis and histogram analysis are carried out in parallel to perform the blink detection task.Close to real time performance and a high level of accuracy in controlled settings show great promise of such approach in enhancing the monitoring of the driver's blinking patterns.One of the main constraints of using such algorithm in a real world setting is the minimized processing time required to allow for sufficient driver response time.In this work implementation of the algorithm is described using optimization techniques to meet such latency requirements.The validation of the algorithm was carried out by visual inspection of the video sequences in terms of precision and accuracy.The presented blink detection algorithm has a precision rate of 84% and an accuracy rate of 69% obtained through using 12 sequences of different duration videos in varying lighting conditions using a small sample of participants.

Web services have increasingly begun to rely on public cloud platforms. The virtualization technologies employed by public clouds can, however, trigger contention between virtual machines (VMs) for shared physical machine resources, thereby leading to performance problems for Web services. Past studies have exploited physical-machine-level performance metrics such as clock cycles per instruction to detect such platform-induced performance interference. Unfortunately, public cloud customers do not have access to such metrics. They can only typically access VM-level metrics and application-level metrics such as transaction response times, and such metrics alone are often not useful for detecting inter-VM contention. This poses a difficult challenge to Web service operators for detecting and mitigating platform-induced performance interference issues inside the cloud. We propose a machine-learning-based interference detection technique to address this problem. The technique applies collaborative filtering to predict whether a given transaction being processed by a Web service is adversely suffering from interference. The results can be then used by a management controller to trigger remedial actions, e.g., reporting problems to the system manager or switching cloud providers. Results using a realistic Web benchmark show that the approach is effective. The most effective variant of our approach is able to detect about 96% of performance interference events with almost no false alarms. Furthermore, we show that a load redistribution technique that exploits the information from our detection technique is able to more effectively mitigate the interference than techniques that are interference agnostic.

In this research we utilize PARAMICS traffic micro-simulation software to study the impact of deploying Connected Vehicles (CV) in Deerfoot trail, Calgary, Alberta. We have implemented a V2V (Vehicle-to-Vehicle) Assisted V2I (Vehicle-to-Infrastructure) system for PARAMICS. It uses Dedicated Short Range Communication (DSRC) protocol to acquire traffic data, calculate and compare important traffic safety and mobility parameters and their impacts on CV by testing five scenarios differentiated by the percentage of 0% to 40% market penetration of CVs. Despite of previous studies which focused on upstream traffic, in this study we demonstrate effect of considering DSRC, re-routing guidance and advisory speed for upstream and downstream traffic. The study demonstrated that the CV technology can enhance traffic safety and mobility in freeways, if the percentage of CVs is significant (e.g. 30–40%) and the CV technology is accompanied by advisory speed reflected on Variable Message Signs (VMS) on both upstream and downstream of the incident location using DSRC range. In other words, equipping freeways with VMS, to use V2I communication, complements the CV technology, improves CV efficiency and leads to higher safety and mobility enhancement in freeways.

Although management of Road Safety has been an area of concern over the past several decades the following behavior-based problem areas still exist: unsafe driving behaviors and high-risk drivers. A Behavior-Based Safety Management System (BBSMS) can help address these areas by introducing concepts of Activators, Behaviors and their Consequences. The focus of BBSMS is on improving and changing behavior rather than dealing with the consequences of bad behavior. This paper explores the application of Utilization Z-scores and Reliability Demonstration Chart, a reliability engineering technique, to help analyze driver behavior. The driver behavior is represented by In Vehicle Monitoring System (IVMS) data collected over several years. The events recorded and monitored by the IVMS include over speeding, over revving, harsh acceleration, harsh braking and seat belt disconnects while driving. The techniques provide an easy and effective way for drivers and their managers to monitor driver risk profiles by classifying and identifying drivers with high risks - drivers with a higher probability of generating IVMS events. The consequences of the unsafe behavior can then be identified and activators can be modified in order to reduce risk. By positively influencing the behavior the consequences can be better managed to reduce the risks associated with the Safety Management System (SMS).

The verification of Multiagent Systems (MAS) and Distributed Software Systems (DSS) has taken a special attention due to the growing demand of having DSS in recent years. The distributed functionality and lack of having a central control in MAS and DSS may cause to emerge new behaviors in the execution time. This unexpected behavior which was not seen in the requirements is known as emergent behavior and may cause irreparable damages. Detection of these emergent behaviors is more valuable and cost effective in the early phases compared to detecting them after the deployment. In this paper we propose a new technique for the detection of a specific type of emergent behavior in the design phase. We take the advantage of social network visualization in this method. The novelty and direct advantage of this technique is presenting the exact point and cause of emergent behavior.

Social networks are ever-growing systems by inheritance. The increase in the number nodes in these systems often brings forth the need to add additional functionalities. However due to the distributed nature of social networks, system growth can be a challenging task. Therefore scalability of the system is of vital importance in the design of social networks. This research attempts to establish a comprehensive framework for analysis and validation of requirements and design documents for software systems. In previous work, we applied this framework to analyze the requirements of a social network of agents; expressed using scenario-based specifications. Scenarios are appealing because of their expressive power and simplicity. Moreover due to the clear and concise notation of scenarios, they can be used to analyze the system requirements for general validity, lack of deadlock, and existence of emergent behavior. In this paper a methodology to analyze the scalability of social networks is presented. This methodology is devised to indicate whether or not the new requirements of the system are consistent with the current requirements in place. A larger prototype of a social network of MSA for semantic search is utilized to illustrate the developed methodology.

Microscopic examination of a properly prepared blood smear is valuable in complete blood count (CBC) and differential blood count (DBC). A hematopathologist may spend enormous time manually inspecting the good working area (GWA) of the blood smear under a light microscope system to perform CBC or DBC. In this paper we focus on automatic localization of the GWA by classifying microscopic images of blood smears using different machine learning algorithms into three areas: Clumped, Good, and Sparse. The features used are the statistical and texture features. This approach yields a good localization of GWA in images acquired by a low cost light microscope system, scanned under magnifying power of x100 oil-immersed objective. Our experiment using images with resolution (3488×2616 pixels) of Giemsa-stained blood smears shows that the proposed method has an accuracy of 82% for the localizing the GWA and 79.73% for all areas in a validation set of 301 images.

Chronic lymphocytic leukemia (CLL) is the most common type of blood cancer in Canadian adults. The relative 5-year survival rates for CLL in Canada is decreasing. CLL cell morphology maybe similar to normal lymphocytes and require a hematopathologist examination for diagnosis. There are a low number of related works on image analysis in CLL. This paper focuses on lymphocyte color cell segmentation using Support Vector Machine (SVM) and k-means clustering algorithms. The algorithm overcomes the occlusion problem when lymphocytes are tightly bound to the surrounding Red Blood Cells. Over and under-segmentation problems are significantly reduced. In this paper we used 440 lymphocyte images (normal and CLL), in which 140 images are used for segmentation accuracy measurement and 12 images for SVM training. The algorithm obtained 98.43% maximum accuracy for nucleus segmentation, and 98.69% for cell segmentation. The cytoplasm region can be extracted by 99.85% maximum accuracy with simple mask subtraction.

A major challenge in design of distributed software systems is predicting and avoiding unexpected behaviors at the run time. Detecting those behaviors after the system is implemented can be very costly and detecting them during design and implementation stages is a cost effective alternative. Therefore, model-based verification at early design stages is an important step in designing distributed systems. Most of the existing verification techniques analyze system behaviors by going from specifications to state machines that model individual components' behaviors. Although those methods are shown to be effective in detecting unexpected behaviors for each component, they fail to detect the unexpected behaviors that occur at the system level. There exist a few ad-hoc methods to combine components' behavior into system level behavior. In this paper, we devise a method that considers interactions among components, and propose an algorithm to combine the behavior models of interacting components. The proposed algorithm can be used to perform automated system-level verification. A case study is developed to validate the efficiency of the proposed algorithm in detecting the implied scenarios for distributed system.

Road related deaths and injuries continue to be one of the highest incidents recorded in organizations. Road Safety has become a major concern worldwide. Therefore the United Nations has introduced a new movement: UN Decade of Action for Road Safety 2011-2020, which aims to reduce road deaths and injuries worldwide. An effective Safety Management System (SMS) can help in reducing risk of incidents, injuries and fatalities. The National Safety Council defines the three performance areas to benchmark SMS to be Leadership - Management, Technical - Operational and Cultural - Behavioral. This paper proposes a systematic way of finding relationships between Technical - Operational factors, Cultural - Behavioral and Safety Management Systems through the use of Neural networks-fuzzy. A sample SMS is simulated using critical factors (environmental and road conditions). Moreover, neural networks are used to predict the next outcome given historical information of various parameters such as road and weather conditions. Fuzzy logic is used to fuzzily the membership functions. The model helps us understand the effects of factors such as snow, rain and Mean Temperature as well as the events reported by In Vehicle Monitoring System (IVMS) on the number of incidents recorded and the "Road Safety Score". To illustrate the methodology in this paper, the neural networks-fuzzy model is fed with environmental factors to see how they affect the overall number of incidents recorded on a daily basis in the City of Calgary.

Multi agent systems with autonomous interaction, negotiation and learning capabilities can efficiently model social behavior of individuals participating in a social network. A central problem in a social network is to identify the nodes that actively participate in the expansion of the net both physically and functionally. Several metrics have already been proposed to identify those hot spots. The algorithms to identify hot spots are either heuristic based or computationally expensive. In this paper we use an agent model of the social net and propose a method that can identify the neutral nodes, i.e. the nodes that can never be considered as hot spot nodes given the network topology and rules of negotiation among nodes. Therefore these nodes can be eliminated from the net. A direct advantage of this method is reducing the computational complexity for the configuration and identification of hot spots. Through a case study we have shown that the proposed method can lead to 33% reduction of computation regarding the number of agent types in the example.

One of the most important clinical examination tests is the blood test. In a clinical laboratory, counting different blood cells is important. Manual microscopic inspection is time-consuming and requires technical knowledge. Therefore, automatic medical diagnosis systems are required to help physicians to diagnose diseases in a fast and yet efficient way. Cell automatic classification has larger interest especially for clinics and laboratories; the most important step in automatic classification success is segmentation. This paper shows an efficient technique for automatic blood cell nuclei segmentation. This technique is relying on enhancing and filtering the gray scale image contrast. False objects are removed utilizing minimum segment size. 365 blood images were used to examine this segmentation technique. Quantitative analysis of the proposed segmentation technique on the blood image set gives 80.6% accuracy. In comparison to other techniques the proposed segmentation technique performance was found to be superior. The five normal white blood cells types were used for evaluation to compare isolated performance. Eosinophil was found to have the lowest segmentation accuracy which is 71.0% and Monocyte was the highest one with 85.9%. The blood images dataset and the source code are published on MATLAB file exchange website for comparison and re-production.

Distributed knowledge has attracted more and more attention as a way to improve knowledge sharing across the world using the Internet. This paradigm enables many systems to interact with each other and share their knowledge while keeping their own ontology. Several researchers have worked on this topic with different strategies but they all argue that the main issue is to make sure that the other systems understand the concepts of its domain correctly. In order to be sure that they understand each other, systems use concept learning to learn the meaning of concepts they communicate with. In this paper, we try to overcome this complexity by suggesting a system that enables agents to learn new concepts from several different agents at the same time and each agent has its own ontology. We use social networks paradigm to communicate between agents to enhance the accuracy of learning process.

Blood testing is one of the most important clinical examinations. Counting different blood cells is a significant process in a clinical laboratory. Manual microscopic evaluation is compulsory in case there is suspicious abnormality in the blood sample. Yet, the manual inspection is time-consuming and requires adequate technical knowledge. Therefore, automatic medical diagnosis systems are necessary to help physicians to diagnose diseases in a fast and nonetheless competent way. Cell automatic classification has wider interest especially for clinics and laboratories. Segmentation is the most important step for automatic classification success. This paper represents an efficient technique for automatic blood cell nuclei segmentation. This technique is relying on enhancing the color of the target object, nucleus, and filtering the image. Small objects are eliminated employing morphological operations. A set of 365 blood images was used to quantitatively evaluate this segmentation technique. Assessment of the proposed technique on the blood image set gives 85.4% accuracy. In comparison to other published technique that was implemented and executed on the same dataset, the proposed segmentation technique performance was found to be superior. A differential segmentation performance evaluation was performed on the five normal white blood cell types to compare isolated performance. Eosin Phil was found to have the highest segmentation accuracy with 90.1%. Lymphocyte and Basophil have the lowest accuracy with 78.3% and 78.6% respectively. The blood images dataset and the source code are published on MATLAB file exchange website for comparison and re-production.

In design of distributed systems with specification languages such as message sequence charts (MSC), communication between different component (agent) types or instances of them are defined. There are a number of methods to verify the design using scenarios of inter-component communication. Those methods usually ignore the intra-component communication, i.e. communication between components of the same type. However in large scale systems, such as e-commerce systems, there are several components of one type that may communicate with each other and this may violate some regulatory policies defined in the design. On the other hand, there are declarative policies in system design that need to be integrated in the implemented system. In this paper a method that takes a topology of the system and regulatory policies as its inputs and detects the components having emergent behavior at its output is proposed. This method is defined to reveal the components that may violate the policies in the design phase by defining message types and extracting a version of MSCs called modified MSCs (MMSCs). Then by clustering and analyzing the send messages in the communications of different components the violating components are detected. By applying this method, all instances of components can be examined for policy violation in the implemented system. The method is explained along with a case study of a realistic online auction system and it is shown how this method can detect the components with emergent behaviors.

Blood tests are of the most important and often requested clinical examinations. Manual microscopic assessment is a must do when a blood sample is suspicious of abnormality. This manual process is tedious, time consuming and subjective. Automating microscopic blood classification is desirable to help the pathologists to speed-up and enhance the results accuracy. Segmentation is the first and most important step in automatic blood cell classification. In this paper, we present an effective technique for automatic blood cell nuclei segmentation. The technique is based on gray scale contrast enhancement and filtering. Minimum segment size is implemented to remove false objects. The technique is tested on 365 blood images. The segmentation performance is quantitatively evaluated on the test set to be 79.7%. This performance is high compared to other published algorithm executed on the same dataset. Evaluation is done on each of the five normal white blood cell types to compare separate performance. The lowest segmentation accuracy is for Eosinophil with 69.3% and the highest is Monocyte with 86.3%. The MATLAB source code and the blood images dataset are published on MATLAB file exchange website for comparison and re-production.

Network filtering has become an important security issue worldwide. Network filters are designed and put in place to enforce restrictions for a variety of different motives, such as political, social, economical or merely security reasons. Although network filters can be applied to different networks, their main use is for the Internet. However, as is the case with most network security measures, many network filters are bypassed by users and thus are not completely adequate to perform their tasks. This paper approaches the network filtering concepts from a software engineering perspective. The general purpose of this approach is to utilize automated methodologies to analyze the correctness of the requirements of the filtering mechanisms, and to reduce their vulnerability. In order to achieve this, requirements are expressed using scenario-based specifications. The resulting scenarios are then analyzed for unwanted behavior using automated methodology. To demonstrate the effectiveness of this approach, it is applied to the case study of a real-life Internet-filtering system.

Emergent behavior in distributed systems is a central problem that may lead to unexpected behaviors and major faults. Emergent behaviors are usually categorized into three groups: emergent behaviors occur due to scenarios incompleteness; emergent behaviors as a result of violation of a system wide policy; and emergent behaviors as a result of synthesis of behavior models. In this paper, a technique for addressing the latter group is proposed. The technique prevents from overgeneralization in the behavior model synthesis. Overgeneralization happens as the result of behavior model synthesis and depends on the assumptions of the process. In addition, the proposed technique addresses the issue of the existing ad-hoc methodologies by providing an automated algorithm. This algorithm can be used by a syntax checker to automatically detect and correct the emergent behaviors in the scenarios. The proposed algorithm is validated using a case study of a fleet management system.

Emergent behavior is an important issue in distributed systems' design. Detecting and removing emergent behavior during the design phase will lead to huge savings in deployment costs of such systems. An effective approach for the design of distributed systems is to describe system requirements using scenarios. A scenario, commonly known as a message sequence chart or a sequence diagram, is a temporal sequence of messages sent between system components. However, scenario-based specifications are prone to subtle deficiencies with respect to analysis and validation known as incompleteness and partial description. In this research, a method for detecting emergent behavior of scenario-based specification is proposed. The method is demonstrated and verified using a mine-sweeping robot as an example. Furthermore it has been demonstrated in this paper that scenario-based specifications can be used in agile software development and that the proposed methodologies in this research can be utilized effectively in agile approaches.

The Web of information has grown to millions of independently evolved decentralized information repositories. Decentralization of the web has advantages such as no single point of failure and improved scalability. Decentralization introduces challenges such as ontological, communication and negotiation complexity. This has given rise to research to enhance the infrastructure of the Web by adding semantic to the search systems. In this research we view semantic search as an enabling technique for the general Knowledge Management (KM) solutions. We argue that, semantic integration, semantic search and agent technology are fundamental components of an efficient KM solution. This research aims to deliver a proof-of-concept for semantic search. A prototype agent-based semantic search system supported by ontological concept learning and contents annotation is developed. In this prototype, software agents, deploy ontologies to organize contents in their corresponding repositories; improve their own search capability by finding relevant peers and learn new concepts from each other; conduct search on behalf of and deliver customized results to the users; and encapsulate complexity of search and concept learning process from the users. A unique feature of this system is that the semantic search agents form a social network. We use Hidden Markov Model (HMM) to calculate the tie strengths between agents and their corresponding ontologies. The query will be forwarded to those agents with stronger ties and relevant documents are returned. We have shown that this will improve the search quality. In this paper, we illustrate the factors that affect the strength of the ties and how these factors can be used by HMM to calculate the overall tie strength.

Lack of central control makes the design of distributed software systems a challenging task because of possible unwanted behavior at runtime, commonly known as emergent behavior. Developing methodologies to detect emergent behavior prior to the implementation stage of the system can lead to huge savings in time and cost. However manual review of requirements and design documents for real-life systems is inefficient and error prone; thus automation of analysis methodologies is considered greatly beneficial. This paper proposes the utilization of an ontology-based approach to analyze system requirements expressed by a set of message sequence charts (MSC). This methodology involves building a domain-specific ontology of the system, and examines the requirements based on this ontology. The advantages of this approach in comparison with other methodologies are its consistency and increased level of automation. The effectiveness of this approach is explained using a case study of an IntelliDrive system.

Distributed systems are employed in countless applications such as information systems, robotics, etc. Lack of central control makes the design of such systems a challenging task because of possible unwanted behavior at runtime, commonly known as emergent behavior. Developing a methodology to detect emergent behavior in the pre-implementation stages of the software development life-cycle of distributed systems can potentially lead to huge savings in time and cost. Moreover, due to the typical large size of the modern distributed systems, automating the detection methodology is considered greatly beneficial. An effective and efficient approach for the design of distributed systems is to describe system requirements using scenarios. A scenario, commonly known as a message sequence chart (MSC), is a temporal sequence of messages sent between system components. However, scenario-based specifications may contain subtle deficiencies with respect to analysis and validation known as incompleteness and partial description. In this research, a tool to automatically detect emergent behavior of scenario-based specification of distributed systems is developed and demonstrated using a robotics example.

The Strategy pattern decouples algorithms from the class that uses them allowing the algorithms to vary independently. It does not, however, allow the algorithms to have different parameters. The parameterized strategy pattern presented in this paper addresses the case when the algorithms have different sets of parameters, and when the user is allowed to see and modify these parameters for each concrete algorithm before its execution. This is accomplished by introducing special parameter classes which encapsulate algorithms parameters. The abstract algorithm class is completely decoupled from parameters letting each concrete algorithm class to create its own list of parameter instances which mirrors its parameters.

There is an increasing demand for the multi-agent systems (MAS) in the automation of manufacturing systems. However, similar to other distributed systems, autonomous agents' interaction in the automated manufacturing systems (AMS) can potentially lead to runtime behavioral failures including deadlock. Deadlocks can cause major financial consequences by negatively affecting the production cost and time. Therefore, a multi-agent manufacturing system should be monitored against the unwanted emergent behaviors such as deadlocks. In this paper, we propose a monitoring technique for deadlock detection in multi-agent manufacturing system based on the MAS design models. In this technique, the MAS is instrumented with a dedicated communication protocol to use the potential deadlock information derived from the design models to propagate deadlock detection query messages. The technique is able to reduce the message communication overhead among the agents by limiting the number of agents that the deadlock detection query messages should be initiated to.

We present an extension to the definition of a concept in an ontology that allows an agent to simultaneously communicate with a group of agents that might have different understandings of some concepts. We also provide a way to learn such non-unanimo

Safe realizability of Message Sequence Charts (MSCs) specifications is a measure of whether or not there exists a distributed implementation of the specification such that it is deadlock free and shows exactly the behaviours specified in the specification. There are also some works that given a specification, can answer whether it is safely realizable or not. However, while these works are restricted by certain assumptions such as synchronous message passing in the system, they also cannot answer why given two specifications, one is safely realizable and the other is not. In this paper, we present a property of MSC specifications that explains implementation problems for them. Using this result, we show how we can effectively correct a specification to avoid implementation problems such as deadlocks and implied scenarios.

The increasing demand for multi-agent systems (MAS) in the software industry has led to development of several agent oriented software engineering (AOSE) methodologies. The autonomous agents' interaction in a dynamic software environment can potentially lead to runtime behavioral failures such as deadlock. Therefore, the MAS environment should be tested and monitored against the unwanted emergent behaviors. The AOSE methodologies usually do not cover monitoring and testing. On the other hand model-based software development practices such as the Unified Modeling Languages (UML) are commonly used in practice and are equipped with a rich set of model based testing and monitoring tools. In this paper, we propose a conversion tool to help MAS engineers use UML-based monitoring and testing tools to test and monitor MAS design and analysis artifacts created by multi-agent software engineering (MaSE) as one of the most powerful and famous AOSE methodologies.

Intelligent software measurement system (ISMS) has been developed to generate a software measurement plan towards a user's initial business goal. The ISMS is based on adapting two methodologies: (1) the normalized 10-step goal-driven software measurement process within the goal/question/metrics (GQM) paradigm; and (2) the software measurement knowledge base building based on the ISMS ontology. In order to take full advantages of these two methodologies, ISMS is designed to be a multiagent system (MAS). In this paper, the goal-driven software measurement process, the design and implementation of the ISMS knowledge base and the design and implementation of the MAS for ISMS are discussed

The paper presents a reliability driven decision support approach to study the effects of defect repository patterns on software release decisions as the systems evolve with continuously changing requirements. The proposed approach called 6C evaluates the suitability of existing reliability models in guiding release decisions and provides different solution alternatives with respect to some project specific parameters for making such decisions. The case study conducted with the approach reveals the impact of project and domain specific uncertainty factors such as risk, testing effort and target reliability on time to market decisions for the software release when the underlying assumptions made by existing reliability models are violated.

Argues that multi-agent system design can be formalized by borrowing from artificial intelligence and software engineering concepts and techniques, such as ontologies, organization, decomposition and synthesis. We particularly focus on ontology sharing of software agents, define agencies as organizations of agents, propose a method to conceptualize the ontology of the domain using a multi-layered bipartite graph called a symbol structure (SS), and propose a method to extract organizational information from the SS of interacting agents. This information is then used in multi-agent system design and development.

The paper reports on results of a study aiming at establishing a fundamental basis for automating design of any kind of software. Considering the final object, an automatic design learning human designer has been taken. An excellent software organization with high maturity has been taken as the expert, and the hierarchical work process is the knowledge model. For detailing, the major operations are made by hierarchical detailing. Namely, a software design may be reduced hierarchically to various design rules, which are parent and children relationship of a human concept created during expansion of a piece of design to more detailed form. To know the inside of human mental operations generating a design rule, a more detailed study was made and it was found that a design rule may be further reduced hierarchically to some fundamental human mental operations, called micro design rules. They represent basic operations during a design. After reporting them, a discussion on further lower level structure used during them, which are dictionary type definitions is made.

Major approaches to automatic fault diagnosis of industrial plants are either subjective or objective. Subjective approaches imitate and synthesize the way that human experts diagnose faults. Objective approaches automate a portion of a diagnostic task which, due to cognitive limitation, humans do not handle efficiently. Currently available subjective fault diagnosis techniques suffer from certain drawbacks, such as: lack of knowledge for modeling and reasoning with the required levels of detail; inefficiency in utilization of sensory data; and inadequacy in learning experienced schemas. A subjective approach to fault diagnosis, using qualitative modeling and reasoning within the multiple view of the system, is introduced. The focus is on automation of the cognitive skills of human experts, which include utilizing conceptual models to detect inherent redundancy in system behavior qualitative reasoning to predict future states, and information selection to avoid computation overload.>