Titles and Abstracts of MoCSSy Graduate Student Seminar Series
Titles and Abstracts of MoCSSy Graduate Student Seminar Series
Appointment scheduling of outpatient surgical services in operation room departments
Alireza Saremi, Ph.D. Candidate, Department of Mechatronics, SFU
November 26, 2010
In recent years, there has been an increase in number of outpatient cases and a corresponding decline in inpatient hospitalization within Canada. Availability of effective drug treatments and faster diagnosis (e.g., MRI and CT scans) have attributed to this increase by reducing the patients’ length of stay. Accordingly, many studies have been carried out to improve efficiency of procedures in order to address the increase in demand for outpatient services. Simulation, mathematical programming, and queuing theory have been used to address challenges of many problems such as scheduling and planning in the outpatient services. In this talk, we address the scheduling of outpatient cases in operation room departments.
Directionality of Criminal Vectors -Phase 2: A Simulation Model of an Offender's Journey to Crime
Natalia Iwanski, B.Sc. Student, Department of Mathematics, SFU
November 19, 2010
Crime Pattern Theory argues that offenders often commit crimes along routes that lead to a criminal attractor. In this project, members of both Mocssy and ICURS have developed a model which reconstructs the most plausible routes taken by offenders to attractors. Dijksta's shortest path algorithm and five years of police data are used to investigate if the simulation model supports the principles of Crime Patten Theory. The most frequently travelled roads are identified and compared with segments that contain crime locations. The impact of adding delays to a road network have also been investigated to see the degree to which offenders' routes change if obstructions are added to their paths. The aim of the model is to better understand criminal behaviour and to study the spatial distribution of crime in an urban landscape.
Food security and stunting in children unfer five in the developing world
Asim Farooq Raja, MPH Student, Faculty of Health Sciences, SFU
November 5th, 2010
Child malnutrition is a major consequence of food insecurity. Over the past decades, the concept of food security has evolved. The new broader concept focuses on three pillars: food availability, food access and food use and utilization. The presentation will be focused on my research study in which I used the broader concept of food security to identify the predictors of stunted growth among children in the developing world. Two models of stepwise regression were run with national prevalence of stunting in children under five as main outcome variable and eighteen selected food security indicators (form all the three pillars of food security) as potential independent variables. The first model included 93 countries with complete data for the selected food security variables. In the second model, the Expected Maximization method was used to impute the missing values for the countries lacking data for some of their indicators. The models were also cross-validated to assess their predictive power. The final discussion will also include the importance of addressing broader concept of food security.
Do we have all that we need for modelling physical activity in the built environment?
Danijela Gasevic, Ph.D. candidate, Dept. of Biomedical Physiology and Kinesiology, SFU
October 29, 2010
A joint work of students from the Chronic Disease System Modelling Laboratory (Dr. Diane Finegood) and Community Health Assessment and Management Facility (Dr. Scott Lear) resulted in a conceptual model of how agents move in the built environment. The built environment, a human-made or modified environment, has been shown to provide barriers or opportunities for safe and convenient engagement in physical activity. Current models do not commonly capture the complex interactions between the built environment and individuals’ physical activity behaviour . Recently, agent based modelling has been proposed as a new tool for testing hypotheses about the impact of changes in the built environment on physical activity. Our group developed a conceptual model of how agents move in their urban neighbourhood. The results of a conceptual modeling phase revealed data gaps in the area of the relationship between physical activity and the built environment. Data gaps will need to be addressed in order to develop a fully specified simulation model capable of predictive applications.
Recurrent Events Data: Introduction and Efficient Panel Designs
Elizabeth Juarez Colunga, Department of Statistics and Actuarial Science, SFU
October 22, 2010
Recurrent event data, which are the data generated from events that repeat over time, are encountered in many fields. Examples are, in criminology studies, the occurrence of crime, and in medical studies, the occurrences of asthma attacks and epileptic seizures. In recurrent event studies, sometimes partial information on the recurrence process, such as the number of events occurring in specific intervals, called panel data, provides nearly the same precision as full information based on data from continuous observation. In this talk I will introduce recurrent events analyses through examples, and then I will talk about conditions for efficient panel designs. I will compare the analysis of such panel data with the analysis of data recorded as times of recurrences, and articulate conditions for efficient panel designs, where the focus is on estimation of the treatment effect when adjusting for other covariates. I will also provide examples of efficient panel designs.
Emotion Depiction: More Expressive Game Characters using painterly rendering (NPR)
Hasti Seifi, MSc student, School of Interactive Arts and Technology (SIAT), SFU
October 15, 2010
Previous studies have shown that people associate specific colors to certain emotions. On the other hand, recently, some studies demonstrated the effect of painting techniques such as textural variations on guiding viewer’s gaze to specific regions of portrait paintings. Motivated by previous studies, a series of studies are conducted to explore the effect of color and painting techniques on users’ perception of an emotion. We hypothesize that the appropriate use of color and painting style leads to conveying emotions to viewer more effectively and provides more satisfactory emotional experience for the viewer/user.
Multiple Perspectives: A Contemporary Analysis of Renaissance Stereotypes in Painting
Denise A.M. Lang, M.A., Department of Art History, State University of New York at Buffalo
October 8, 2010
This topic will explore the historical misrepresentation of Renaissance Perspective as solely a mathematical tool, while analyzing the multi-dimensional rhetoric of Leon Battista Alberti’s treatise “On Painting.” Not only was this treatise the first academic doctrine to pursue the painted image, it was also a conscious attempt to re-shape the image of the artist into that of an interdisciplinary scholar; learned in everything from mathematics to metaphysics. In fact, a detailed inquiry into the subject of Renaissance philosophy uncovers numerous similarities to our contemporary era, and by interweaving specific theories of Jacques Derrida into Alberti’s treatise, both linear perspective and contemporary conceptions of Renaissance history can be re-traced for their intersections and overlaps.
Setting Staffing Requirements for an Emergency Department in the Event of Surge
Hengameh Vahabzadeh, M.Sc. student, Mechatronic Systems Engineering, SFU
September 24, 2010
Disasters and infectious disease outbreaks over the last several years have demonstrated the importance of emergency preparedness for large-scale events affecting many people. The ability to respond effectively to events producing a massive flow of patients requires surge capacity. In the event of surge timely access to an emergency provider is an important factor of quality for emergency departments (ED), though many EDs are facing a number of challenges which would lead to excessive waiting time and diversions. We employ queueing theory methods in combination with simulation optimization approach to address the challenging problem of determining the staffing requirements.
Modelling the Propagation of Air Pollutant from a Plume using Cellular Automata Approach
Abraham Amankwah, Ph.D. student, Physics Department, University of Ghana
September 17, 2010
Trajectory of air pollution and the amount of pollution is strongly dependent on the wind direction as well as the surface features of a region. Idrisi Taiga software was used to develop a Cellular Automata (CA) model for propagation of air pollutant from a plume. Cell space was kept similar to a valley completely surrounded by mountains with some roads leading into and out of the region. Four different cell states: polluted, short structures and open areas, tall structures, and mountains were used. Parallelism inherent in CA was achieved when simple transition rules were applied in the model. The windward side of the mountains was heavily polluted while the leeward side remained unpolluted. The air pollutant goes round the tall structures thus leaving them unpolluted- this occur with the assumption that there is no impact fallout of the air pollutant. This clearly indicates that the mountain and the tall structures can serve as an effective shield. However the short structures and open areas in the trajectory of the pollutant got polluted. This clearly indicates that these structures cannot be effective shields for air pollution. Roads in the trajectory of the pollutant were also heavily polluted whereas other roads were left unpolluted.
Reactions to resistance: The role of contextual factors in sex offending
Samantha Balemba - M.A. Student (Criminology, SFU)
April 9, 2010
Upon encountering victim resistance during the commission of a sexual assault, an offender makes a decision as to how to react to the resistance, via the use of varying degrees of coercive or noncoercive tactics. Exhaustive Chi-squared Automatic Interaction Detection (CHAID) analyses were performed on a sample of 369 criminal events to detect the significant situational and crime-specific factors, the pertinent interactions between these factors, and how these affect an offender’s reaction to victim resistance depending upon whether the victim is a child or an adult. Findings show that multiple interactions do occur and, moreover, factors affecting the offender’s reaction to victim resistance are very different depending on whether the victim is a child or an adult. Planning and preparation variables, such as the offender’s attack strategy, were found to be significant in the child victim model, whereas reactive variables, such as the type of victim resistance, were found to be significant in the adult victim model. The findings lend support to the social interactionist perspective and the importance of victim-offender dynamics.
Modelling the Spatio-temporal Complexities of Forest Disturbance by Mountain Pine Beetle
Liliana Perez - Ph.D. Candidate (Geography, SFU)
March 26, 2010
The mountain pine beetle, Dendroctonus ponderosae Hopkins, (MPB) is endemic to lodgepole pine, Pinus contorta, forests of western Canada. However, since the 1990s the province of British Columbia has experienced the most extensive outbreak with a record of over 13 million hectares of trees killed by the MPB, having a tremendous impact on the forest industry. Climatic changes such us hot dry summers and above normal winter temperatures, and abundance of mature lodgepole pine, resulting from decades of fire suppression, have been identified as contributors to the outbreak. The goal of this study is to learn about the spatial dynamics product of the beetles’ interaction with their environment by acknowledging that their behaviour is robust, flexible, adaptive, self-organized, intuitive, and scalable. The complexity of the MPB attack and the emerging patterns from their spatiotemporal dynamics are captured by this study, using an agent-based model (ABM) approach within a GIS frame.
Diversity in Offending
Stacy Tzoumakis - Ph.D. Candidate (Criminology, SFU)
March 19, 2010
For the last forty years, criminologists have examined the idea that an offender specializes in a specific type of crime, or mainly commits one type of offense. Offending specialization has generally not been supported in the empirical literature. However, with the advent of new methodological techniques in the last decade, mixed support for offending specialization had been found. Specifically, techniques which examine the question at the individual level instead of at the aggregate level have found more evidence of specialization in some cases. This study uses the Diversity Index, a method which has been used in ecology and biology to determine species diversification (Agresti & Agresti, 1978), and has only recently begun to be applied in the social sciences. The sample used in this study involves 470 juvenile delinquents followed from age 12 until around age 40 from the city of Montreal, QC. Diversity indices are calculated and nonparametric analyses are completed. Theoretical and policy implications of the findings are discussed.
A system dynamics model of the British Columbia justice system
Alexa van der Waall - Research Associate (Complex Systems Modelling Group, SFU)
March 12, 2010
The Complex System Modelling Group has developed a high-level model for the minstries of Attorney General, and Public Safety and Solicitor General of BC. We show the general lay-out of the model, discuss choices that were made, and touch on several mathematical challenges.
Detection of Novel Sequence Insertions Using Paired-end Next-generation Sequencing Technology
Iman Hajirasouliha - Ph.D. Candidate (Computing Science, SFU)
March 5, 2010
We present a computational framework to discover the locus and content of novel sequence insertions using the NGS platforms. We test our methods with the high-coverage short-insert sequence library generated from the genome of a Yoruba individual (NA18507) sequenced by Illumina (Bentley et al. 2008). We validate the content of the predicted novel sequence insertions by using arrayCGH and sequenced fosmid clones generated from the genome of the same individual (Kidd et al. 2008). Our method can be used to characterize the DNA sequences missing from the reference assembly to obtain a better picture of the human genome diversity.
Modeling urban growth: An agent-based approach
Anthony Jjumba - Ph.D. student (Geography, SFU)
February 12, 2010
A city can be seen as a complex system in which a diversity of stakeholders with conflicting objectives and priorities interact and directly influence the process of urban growth. Their interactions, which are characterized by competition for space, economic benefits, environmental or sustainable goals, can be represented by an agent-based modeling approach. In this study, an agent-based model is developed in order to mimic the process of urban land-use change at a high spatial resolution of cadastral scale and by modelling the actions of the stakeholder ‘agents’ as key actors in the city’s development. The simulation outcomes provide several urban growth scenarios based on the implemented urban planning policies and the impact of relocated households. The model assists in better understanding and analysis of the stakeholders influences on urban land use patterns and also helps in the forecast of possible future urban growth.
Influence of Forested Areas On The Water Balance of Burns Bog
Yue-Ching Cheng - M.Sc. student (Biology, SFU)
February 5, 2010
The project is intended to determine and quantify the influence of forest encroachment on the water balance of a raised bog. The study takes place in Burns Bog in Delta, BC, approximately 20km southeast of Vancouver, BC. The water balance approach requires quantification of the differences in the evapotranspiration and interception losses between a forested and pristine bog site. Measurements of the water balance variables are made at both sites during the summer and fall of 2009. Using this data a simple water balance model will be developed for the study area.
Exploring Persian Rug Design Using a Dual Stage Genetic Algorithm Approach
Arefe Dalvandi - M.Sc. Candidate (Interactive Arts and Technology, SFU)
January 29, 2010
Looking at the art of Persian rug design as a computation creative design problem, with a vast domain space of possible design solutions that have aesthetic, cultural and historical considerations, we describe our dual stage genetic algorithm system for designing basic patterns of a specific type of Persian rugs. Our approach use hard and soft design rules that we have been gleaned from the passed down traditions of Shah Abbas Persian rug design. We break down the rug generation into two phases. In the first phase, the rug (a collection of connected spirals as a core structure) is generated exploiting the available genetic operators. An evaluation mechanism based on the most straightforward soft design rules ranks each generated genotype in the first phase. In the second phase, the highly ranked genotypes are presented to the user to select the most aesthetically acceptable rugs for the next evolution. We situate this process within general creative evolutionary design and report on early results
Crime Location Quotients: Measuring the Intensity of Crime in Small Areas of a City
Katie Wuschke - Ph.D. student (Criminology, SFU)
January 22, 2010
Crime patterns across the smaller areas that comprise cities are typically measured by event counts and resident population based event rates for aggregate categories such as "property crime" or "violent crime". Such measures can mask substantial differences in crime mix and crime risk across neighbourhoods. This study utilizes an alternative measurement approach - the crime location quotient - to highlight neighbourhood differences in crime patterns across a British Columbia city.
Transit in Transforming Neighbourhoods - A Look at the Perception of Crime in an Urban Environment
Monique Guterres - M.A. student (Criminology, SFU)
January 15, 2010
Past research has confirmed that certain features in the urban landscape such as graffiti and litter generate fear of crime. A mixed method survey will be conducted in Vancouver’s trendy Yaletown-Granville district, with a prevalence of bars and nightclubs in the area. The survey is intended to be compared with other community surveys conducted in British Columbia’s Metro Vancouver area to develop a community policing strategy designated to reduce fear of crime. Investigating the impact(s) of the perception of crime with ecological psychology may provide insight towards how Computational Criminology techniques could be used to measure nuisance crimes, or disorder.
Infrasonic Cardiac Signals
Kouhyar Tavakolian - Ph.D. Candidate (Engineering Science, SFU)
November 27, 2009
Infrasonic cardiac signals are a category of cardiac signals that have their main components in the infrasonic range (less than 20 Hz) and reflect the mechanical function of the heart as a pump. During the past century, extensive research has been conducted on interpretation of these signals in terms of their relationship to cardiovascular dynamics and their possible application in cardiac abnormality diagnostics. Signals such as ballistocardiogram (BCG), seismocardiogram (SCG), apexcardiogram (ACG) and radar seismocardiogram reflect the displacement, velocity, or acceleration of the body in response to the heart beating. In this talk a comparative analysis of these signals is presented based on the most recent research in the field.
Social Influences amoung Drug Users and Mean Field Approximation of Cellular Automata
Afsaneh Bakhtiari - M.Sc. Candidate (Mathematics, SFU)
November 20, 2009
Street-involved youth have a propensity for street-level illicit drug use and trade. There are approximately one million street-involved youth in the United States and about 150,000 in Canada. The need for a proper intervention and follow-up strategy seems remarkably clear. Mathematical modeling is a potentially valuable tool for developing such a strategy.
A stochastic cellular automata (CA) model of the influences among current and potential illegal drug users and traffickers is presented. Simulation and mean field analysis are used to study the phase diagram of the CA model. The mean field approximation (MFA) of the model corresponds to a compartmental model which is defined as a system of ordinary differential equations. The phase plane of the mean field and possible bifurcations of the system are explored. MFA typically provides a good picture of a CA model near a bifurcation. The model allows us to compare the potential effectiveness of different types of responses to the drug epidemic. Both indirect and direct strategies are found effective on their own, but combined strategies proved to be most effective.
Patterns of crime in the surrounds of drinking establishments: a small place analysis in Vancouver
Andrew Reid - M.A. Candidate (Criminology, SFU)
November 13, 2009
Human behaviour is known to exhibit clear spatial and temporal patterns. Environmental criminology uses this assertion as a basis for exploring the relationship between crime and place. Urban areas provide a particularly rich environment for ecological studies of crime since complex street networks, divisions of land use, activity patterns, as well as a host of other environmental phenomena are known to contribute to patterns of social behaviour. Using GIS we combine police calls for service and licensed liquor location datasets for Vancouver, BC. In our analysis we investigate the spatial and temporal patterns of crime and their relationship to drinking establishments. This paper forms the baseline for continued work on the nature of liquor service-related crime problems in the built environment one year before the arrival of the Winter Olympics in 2010. The results of our study contribute to the growing body of literature in the field of environmental criminology and more specifically, help to build a greater understanding of the spatial distribution of crime in urban environments.
Estimating the prevalence of dysmenorrhea in Iranian women and investigate associated risk factors
Mahkam Tavallaee - MPH student (Health Sciences, SFU)
October 30, 2009
Objectives: To estimate the prevalence of dysmenorrhea in Iranian women and investigate associated risk factors.
Methods: In a cross-sectional study in Tehran in 2007, Iran, 381 women (81% response rate, age 16-56 years) were selected through a stratified random sample of 22 different districts and completed a questionnaire about dysmenorrhea. Descriptive statistics, spearman rank correlation statistic, and ordinal logistic regression models were used. Confounding and effect-modification were explored for each association.
Results: The prevalence of no, mild, moderate, and severe menstrual pain was 10%, 41%, 28%, and 22% respectively. Older age and high intake of fruits and vegetables were protective factors for menstrual pain while women with family history of dysmenorrhea, higher stress and depression tended to have more severe pain. BMI, parity, smoking, and physical activity were not significantly associated with dysmenorrhea.
Conclusion: Menstrual pain is a common complaint in Iranian women. The inverse association between fruit and vegetable intake and dysmenorrhea, and reduction of stress and depression need to be further explored and considered in terms of recommendation to reduce dysmenorrhea.
Exploring Trends, Key Variables, and Feedback Loops in the Obesity System Map
Tommy Merth - M.Sc. Candidate (Biomedical Physiology and Kinesiology, Chronic Disease Systems Modelling Lab, SFU)
October 16, 2009
Obesity rates for all age groups have drastically risen over the past several decades and now occur in epidemic proportions. A complex system of physiological, social, and environmental factors is to blame for this increase. Recently, the UK’s Foresight Programme created the Obesity System Map to represent this complex system of variables. These variables were arranged into 8 clusters on the map (Food Production, Food Consumption, Social Psychology, Individual Psychology, Individual Physical Activity, Physical Activity Environment, Physiology, and Engine).
To explore the details of the map, social network analysis software was utilized. A reduced version of the map was produced. Variables were sorted according to their number of incoming and outgoing connections, and key variables were identified. Finally, both positive and negative feedback loops were explored.
Findings from this project provide new insight into key variables, clusters, and feedback loops present in the obesity system. They may help inform future solutions oriented discussions on obesity, and reveal areas or connections which deserve more consideration.
The Criminal Careers of Chronic Offenders in Vancouver, British Columbia
Chris M. Giles - Ph.D. Candidate (Criminology, SFU)
October 9, 2009
One of the most pivotal findings in the criminal career paradigm is that a small group of offenders are responsible for the majority of offences in a given time period. This subset of the offending population is the focus of a substantial amount of empirical and theoretical study and debate. One of the short-comings of past research is that it is based on general population samples or general inmate samples. The outcome of these sampling parameters is that a large number of chronic offenders are not included. This research attempts to fill this gap by analysing the offending careers of 152 retrospectively identified chronic offenders from the Chronic Offending Program at the Vancouver Police Department. This study, using convictions and negative police contacts, examines the commonly reported criminal career parameters of participation, age of onset, incidence and age/crime with respect to this chronic sample. In addition, this study analyses the often debated constructs of lambda, defined in criminal career research as offending frequency of active offenders, and specialization for the sample of chronic offenders. Contrary to past research, this study finds a substantial difference between lambda estimates that incorporate time at risk and estimates based on frequency alone. It is the contention of the author that these constructs are ideally suited to study high frequency offenders as they constitute a large body of data required for analysis of these two constructs.
Recent Advances in Privacy Preserving Data Publishing
Rhonda Chaytor - Ph.D. Candidate (Computing Science, SFU)
October 2, 2009
In today's electronic society, collecting and selling information is a big business. It seems like everywhere you turn, someone is asking for your personal information. In this talk, we discuss who has your personal information and the problem of protecting individual privacy, while at the same time, extracting useful knowledge from collections of personal information that may benefit society as a whole. We describe the origins of privacy preserving data publishing and introduce recent work on randomization approaches. We provide arguments why this new field of research is worthwhile studying and we suggest several new research directions. In particular, we emphasize the lack of work on finding randomization operators that maximize utility in the existing literature.
Analysis of complex networks
Philippe J. Giabbanelli - M.Sc. (Computing Science, SFU)
September 25, 2009
Numerous situations can be modeled as networks, such as social networks (in which people are connected by social interactions), road networks (cities connected via highways), protein networks, etc. Analysing a network relies on knowing the properties that we want to measure, their consequences, and being time efficient by using the right software. This talk is a gentle introduction to network analysis; our practical demonstration will use Visone on a recent example from the health sciences. No specific knowledge is required. Additional talks on complex networks (models&simulations) can be planned depending on the audience's interest.
Modeling Salmon Return Rates Using Compositional Data Methods
Carolyn Huston - Ph.D Candidate (Statistics and Actuarial Science, SFU)
September 18, 2009
This research proposes a unique hierarchy to a multivariate conditional autoregressive (MVCAR) model. While statistical methods for MVCAR models are well developed, we extend these tools so that they can be employed when the response vector is compositional, i.e. has components that sum to one. This allows us to harness the advantages of an MVCAR model when the response variables of interest are relative measures, rather than absolute measures.
A common problem when response values are compositional occurs when there is a large number of observed zeros in a component category. Such low abundance categories are difficult to incorporate into existing models. This is because most models rely on some form of logarithmic transformation of the data so that estimation occurs on all real numbers, and parameters of interest are then back transformed to between zero and one. These transformations tend to either perform poorly at the zero boundary, or not at all (Billheimer et al., Env. and Ecol. Stat., 1997; Grunwald et al., J. R. Statist. Soc. B, 1993).
A second modeling concern arises because the most natural distributions for compositions, the multinomial and the Dirichlet, impose a rigid negative covariance structure between the different components (Aitchison, J. R. Statist. Soc. B, 1982; Mosiman, Biometrika, 1962). Our model addresses both the concern of handling zero observed counts and can estimate positive correlations between model parameters for the different component categories.
This research was motivated by data collected by the Pacific Salmon Commission to help manage the mixed stock commercial fishery in the Fraser River. In addition to maintaining an overall population large enough to sustain a commercial fishery, managers also want to preserve the genetically distinct stock groupings that pass through the river (components). Each stock grouping represents a set of rivers where salmon stocks are genetically distinct, and it is desirable to maintain viable populations in all of these habitats.
GIS-based spatial accessibility and location analysis modelling techniques for examining access to health services
Jon Cinnamon - MSc (Geography, SFU)
August 5, 2009
Abstract: GIS modelling and spatial analysis techniques are increasingly used to examine a wide variety of public health issues. In this presentation, I will describe some GIS-based methods that are appropriate for examining the spatial accessibility of health services, and for determining appropriate locations for new services. These techniques are applied to the provision of palliative care in rural British Columbia. Results highlight the areas of the province that have unfavourable access to these services, and the most suitable communities for new service locations.
A Graphical Model of Crime Vectors
Suraj Kumar Singh - MSc (Applied Mathematics, Indian Institute of Technology Kanpur)
July 29, 2009
We graphically model the directions offenders take to commit their crimes. This direction can be represented by the vector, called crime vector, towards crime location from offender's home location. The model is to be based on 5 years of real police data. But because of confidentiality associated with it we first work with a hypothetic data created randomly. Using this data we create a graphical model of crime activity vectors to analyze the crime movement patterns. Later we introduce a crime attractor location to see the behavior of crime vectors around it by inclusion of a parameter namely crime angle, the angle between the crime activity vector and the vector towards the crime attractor location from offender's home location. We also see the effect of distance on crime patterns including the parameter namely crime distance which is the distance between offender's home and crime location.
How Computational Geometry can solve all your problems
Laurens Bakker - MSc candidiate (Computing Science, SFU)
July 22, 2009
Let me make an audacious statement: all problems and data can be modelled using computational geometry, and (here is the catch) in many cases this may even be meaningful, useful or informative. I will show how sorting stuff, gift wrapping, minimum risk robot path planning and deciding on a location for a mobile telephony tower can all be solved using exactly the same method. You will hear about the Delaunay Triangulation, the Convex Hull, various Voronoi Diagrams, and projections as we go along. Time permitting, I will also venture into some new territory: calculating a single Steiner point in O(n log n) with restrictions or the Convex Hull in 2 dimensions in less than O(n log n) time on a real computer.
A system dynamics model of body weight regulation and obesity
Ozge Karanfil - PhD candidiate (Biomedical Physiology and Kinesiology, SFU)
July 15, 2009
The growing interest in body weight regulation has culminated in the growth of simulation models that are employed as a tool to investigate this complex system, and as a means for evaluating hypotheses concerning the induction and maintenance of obesity. The purpose of this modeling study is to develop a dynamic representation of our body weight regulatory system in normal and obese states, and to examine the interactions between the body composition and food intake regulation to see their effects on body weight maintenance.
An information model for visualizing critical infrastructure interdependencies
Coco Jiang - MSc candidiate (Interactive Arts and Technology, SFU)
July 8, 2009
Critical infrastructures such as electricity, water, telecommunication, transportation and medical services are complex networks of interconnected and interdependent components. Identifying, understanding and managing the interdependencies between critical infrastructures is a new challenge of emergency management. I2C JIIRP is a multidisciplinary project that studies new simulation approaches and techniques to analyze critical infrastructure interdependencies. This research is part of the visualization research of the I2C JIIRP research project.
In this research, we propose an information model - a high level design of visual interface - to the I2sim prototype simulator. We intend to explore ways to visualize interdependencies, capture missing knowledge, support shared understanding and assist decision-making between emergency and infrastructure managers during the response phase and short-term recovery phase of emergency management. To demonstrate the information model, we walk through the I2sim 5-cell test case with 2006 UBC campus power shortage scenario to showcase the problems and solutions. At the end, we further discuss how the information model might support emergency management work from three user perspectives: planning, analysis and reaction.
Prefer to Push - why women want to birth naturally
Brenda Kent - MPH student (Health Sciences, SFU)
June 24, 2009
It is well recognized that natural childbirth, without unnecessary medical interventions, is the safest and healthiest way to give birth. This study focuses on the reasons why some women choose to have a natural birth experience despite the dominant cultural influences discouraging such a decision. In this qualitative study, Brenda is conducting interviews with maternity care providers and pregnant women to answer the questions: “what are the factors that influence attitudes and beliefs about labour and childbirth among pregnant women?”, and “what motivates pregnant women to seek a natural childbirth experience?” She had just finished her data collection and will share some emerging themes and an overall perspective of the risks of medicalizing childbirth.
CourBC: A Gateway to Understanding Criminal Court Processes in British Columbia
Andrew Reid - M.A. Candidate (Criminology, SFU)
June 17, 2009
Criminal court is an institution within the Criminal Justice System that plays an important role in the administration of justice in Canada. The role of criminal courts is to resolve disputes between individuals and the state that involve matters of criminal law. The processes involved in resolution, however, may be very complex and consequently may be difficult to understand. Obtaining useful data is critical to obtaining a better understanding of the court system. CourBC is a new database of criminal court appearance results for the province of British Columbia. Containing detailed information from all criminal courts in the province, this database may provide a gateway to important research on case variation in different courts, case complexity, processing steps in different types of cases and analysis of sentencing outcomes for guilty outcomes.
From Heuristics to Complexity: Integrating Artificial Intelligence and Agent-based Modelling for Simulating Patterns and Processes of Forest Cover Change
Chris Bone - Ph.D. Candidate (Geography, SFU)
June 10, 2009
Forest cover change is a complex spatially dynamic phenomenon involving the interaction of numerous processes leading to emerging forest patterns over time. This is especially true when considering forestry operations that attempt to harvest trees for maximizing short-term profits while contending with natural disturbances, fluctuating economies, and the need to conserve long-term ecosystem functions. Conventional computer models assist harvesting activities by generating forest cover strategies that satisfy both economic and ecological objectives. However, such models ignore the dynamic forces that govern the harvesting process, and as such produce strategies that can be in direct conflict with emerging patterns. The purpose of this dissertation is to enhance existing modeling approaches by bridging complex systems theory and artificial intelligence in order to incorporate spatial and temporal complexities of forest harvesting. Specifically, this dissertation introduces a novel approach for integrating geographic information systems (GIS), agent-based modeling (ABM) and reinforcement learning (RL) for developing intelligent agents that can represent stakeholder behaviours and their influence on forest cover change. Agents embedded with RL algorithms possess learning mechanisms that allow them to gain knowledge from their experiences in a dynamic forest environment as represented by GIS digital data structures. Agents learn where and when harvesting activities should take place in the forest in order to satisfy different and at times conflicting objectives that exist at varying spatial scales. Model results provide forest management with suitable harvesting strategies that satisfy conflicting objectives, information regarding the relationship between stakeholder interactions and emerging forest cover patterns, and the ability to evaluate the tradeoffs between different harvesting objectives. The developed approach is implemented in the context of forest management in British Columbia using datasets representing forest cover in the Chilliwack Forest District.
Software (What's It Good For?)
Piper Jackson - Ph.D. Candidate (Computing Science, SFU)
June 3, 2009
Why have software as part of a research project, and how can the process of developing research software be improved? A proposed taxonomy of the roles software can play in a research project will be presented. I hope to have some input and discussion on the characteristics related to each role. Several projects from the Software Technology Lab, including Mastermind, Genius and NADIF, will be shown as examples.
Modelling congestion in the intensive care unit of acute care hospitals
Pouya Bastini- M.Sc. Candidate (Mathematics, SFU)
May 20, 2009
Over the years, the growing population in British Columbia has lead to escalating wait-times and overcrowding in hospital Emergency Departments (EDs). This effect can be traced to the insufficient number of beds in specific units of the hospital, such as the Intensive Care Unit (ICU) or the Medical Unit (MU). To enhance the level of access to care, a successful prediction of the bed requirements is needed. This is achieved by having an adequate model of the patient flows to and between the different compartments of the hospital. Focusing only on the stream of emergency patients, we developed a queueing network model of the interaction between the ICU and the MU, which is believed to be causing a major proportion of the congestion in the ED. Through approximate analytical methods and simulation, we determined the sufficient bed counts in each of these two units so as to guarantee certain access standards.
Modelling Long Term Care in Hampshire, UK
Shivam Desai - Ph.D. Candidate (University of Southampton, UK)
May 13, 2009
The "ageing population" presents many significant challenges for health and social care services at both a national and local level, one of which is to meet the demand for long-term care. The population of people aged over sixty five will continue to grow for some time as the baby boom generation ages. There are other challenges associated with an ageing population which will have an impact on the provision of long-term care. Increased life expectancy is one, but there has also been a reduction in the size of the working population, which has led to less money being raised by taxation, an important source of funding for long-term care, and a reduction in the size of the workforce. These issues will be investigated in further detail using approaches from the discipline of Gerontology. The aim is then to use System Dynamics to model the demand for long-term care in the county of Hampshire, England amongst those aged sixty-five years and over.
A key part of the modernisation of Hampshire County Council has been the establishment of a new contact centre. The research will address such questions as:
- How can Discrete Event Simulation be used to assist HCC to plan the development of the new Call Centre?
- How could a detailed tactical model for the Call Centre benefit from the additional use of a long-term system dynamics demographic model for population change?
Data for the discrete event simulation model will be provided by the local authority. The long-term SD demographic model will use be populated from a variety of data sources, including the local authority, local area statistics and nationally representative data. Key features of the model include an attempt to model dynamic changes in the older population as well as the inclusion of health statistics.
This project takes a cross-disciplinary approach combining Operational Research modelling with Gerontology, Demography and Social Policy. The research is funded jointly by the Economic and Social Research Council and Engineering and Physical Sciences Research Council.
Criminal Activity Vectors: Finding the attractors of offenders based on their direction of offending
Richard Frank - Ph.D. Candidate (Computing Science, SFU)
May 6, 2009
Human Figure Segmentation Using Clothing Texture Cues in a Max-Margin Framework
Mani Ranjbar - Ph.D. Candidate (Computing Science, SFU)
March 20, 2009
Solving challenging image segmentation problems requires the use of all available information. We examine the class-specific problem of automatically segmenting human figures. This work focuses on what can be accomplished when high-resolution images of human figures are available. We present a pair of textural cues, describing fine scale material textures and coarse scale clothing fold textures. While powerful, these bottom-up cues alone are insufficient for segmenting human figures. We combine them with top-down knowledge of body shape and generic bottom-up segmentation cues. We phrase the learning of this model as a structured output learning problem, and exploit the Structural Support Vector Machine to learn appropriate weights for these different cues. Experimental results quantitatively demonstrate the effectiveness of the proposed cues on a challenging dataset of images.
Mathematical and societal challenges to modelling home and community care
Warren Hare - Ph.D. (Mathematics, SFU)
March 13, 2009
In British Columbia (BC) Home and Community Care (HCC) services provide a range of health care and support services for British Columbians with palliative or rehabilitative health care needs. Given that almost 80% of HCC clients are senior citizens (aged 65 and older), and the most recent population projections suggest that the senior population of BC will double in the next twenty years, the BC Ministry of Health is interested in predictive modelling of HCC service loads in BC.
In this talk we examine some mathematical and societal challenges that arise in developing models to explore the future of HCC in BC. In particular, we address the issue of how health care usage and age are not easily related, and the issue of non-publicly funded HCC options. We discuss the development of a deterministic multi-state Markov model of the HCC system that was recently developed for the BC Ministry of Health and is currently being used as a ?rst step to developing a strategic direction plan for BCs HCC sector.
Predictive Modelling of Clandestine Graves from Civil War in Spain
Derek Congram - Ph.D. Candidate (Archaeology, SFU)
March 6, 2009
The talk will be about my PhD research, which involves the search for non-combatants who disappeared (were detained and never seen again) during the 1936-1939 Spanish Civil War. I will provide a brief background on the war and the challenge of searching for missing persons, followed by a proposed methodology for modelling the location of clandestine grave sites. As I have little background on modelling, I will specifically be seeking methodological advice from the group.
Taking out the trash: Evaluating options for long-term disposal of radioactive waste
Laurens Bakker - M.Sc. Candidate (Computing Science, SFU)
February 27, 2009
Nuclear power plants, nuclear research facilities and Uranium mines, but also hospitals and the military (nuclear submarines, decommissioning of nuclear arms, etc.) all produce significant amounts of radioactive waste. Storage seems to be the only option, but even then the question is: where?
This talk will set out the requirements for a radioactive waste repository, and touch upon some of the scientific and political debates in setting these requirements. The bulk of this talk will be based on a paper we published at RepoSafe'07: "Prioritizing R&D for seven radioactive waste disposal options — an independent, interactive approach".
Estimate of CVD following different reductions in sodium intake
Azadeh Alimadad - Ph.D. Candidate (Statistics and Health Sciences, SFU)
February 20, 2009
Cardiovascular disease (CVD) is the leading cause of death in adult Canadian. In 2002, CVD accounted for 74,530 Canadian deaths. According to Statistics Canada (1990-1994) CVD mortality rates rise gradually from west to east with the lowest death rate in British Columbia. Hypertension is the leading risk factor for mortality worldwide which is a major risk factor for (CVD). Evidence shows that reduction in blood pressure level achieved by reduction in sodium intake can be translated in decreased risk of stroke, myocardial infarction and congestive heart failure. In this talk I will present my preliminary results of estimating the effect of dietary sodium on cardiovascular events in Canada.
Spatial Targeting of Homeless Populations in Vancouver and Surrounding Communities
Jordan Ginther - B.Sc. student (Computing Science, SFU)
February 13, 2009
Homelessness is a pressing issue that affects every North American city. Several enumeration methodologies have been applied to try and gain abetter understanding of the demographics of homelessness. These enumeration techniques, whilst providing rough estimates of numbers, rarely, if ever, address the spatial distribution of homeless populations in an urban or suburban context. Potentially this spatial factor, when combined with proper demographics, can provide a valuable tool in the fight to reduce homelessness. This reduction comes as a result of the ability to spatially target services provided to the homeless. In this analysis, we develop a suitability model to identify potential refuge sites of the homeless in the Metro Vancouver region. With the aid of GIS we combine a series of data sets including transportation networks, land use, crime, health, and environmental to predict possible locations of homeless refuge sites. This analysis can, in turn, be used to better provide much needed services to help reduce the homelessness problem in North America.
