21793 research outputs found
Sort by
Automated Data Preparation using Semantics of Data Science Artifacts
Get PDFData preparation is critical for improving model accuracy. However, data scientists often work independently, spending most of their time writing code to identify and select relevant features, enrich, clean, and transform their datasets to train predictive models for solving a machine learning problem. Working in isolation from each other, they lack support to learn from what other data scientists have performed on similar datasets. This thesis addresses these challenges by presenting a novel approach that automates data preparation using the semantics of data science artifacts. Therefore, this work proposes KGFarm, a holistic platform for automating data preparation based on machine learning models trained using the semantics of data science artifacts, captured as a knowledge graph (KG). These semantics comprise datasets and pipeline scripts. KGFarm seamlessly integrates with existing data science platforms, effectively enabling scientific communities to automatically discover and learn from each other’s work. KGFarm’s models were trained on top of a KG constructed from the top-rated 1000 Kaggle datasets and 13800 pipeline scripts with the highest number of votes. Our comprehensive evaluation uses 130 unseen datasets collected from different AutoML benchmarks to compare KGFarm against state-of-the-art systems in data cleaning, data transformation, feature selection, and feature engineering tasks. Our experiments show that KGFarm consumes significantly less time and memory compared to the state-of-the-art systems while achieving comparable or better accuracy. Hence, KGFarm effectively handles large-scale datasets and empowers data scientists to automate data preparation pipelines interactively
The Impact of Green Metrics on Inventory Transshipment
Get PDFGreen is associated with life and is becoming increasingly engrained in not just life, but the way people do business as well. In recent years, a growing number of business operations have adopted various green metrics to limit their carbon footprints and environmental pollution, drive sustainable operations, contribute more to sustainability projects, and appear more socially responsible within the industry and host communities. While these initiatives target reducing carbon footprints, their impact on daily operations in a sharing economy is yet to be explored. In this thesis, I performed a thorough review of green supply chains, recent green practices, and metrics adopted in various organizations, followed by a comparative study to analyze the impact of operational decisions in inventory and transshipment when green metrics are considered. I extended the classical inventory transshipment model with two newsvendors’ retailers by allowing the retailers to incorporate direct or indirect green metrics as part of the objective function. In this setting, I explored three central research questions: 1) How would the adoption of green metrics impact the expected profit and equilibrium order quantities under inventory transshipment? 2) Would green metrics negatively or positively impact the coordinating transshipment prices? 3) What is the impact of direct vs. indirect green metrics on expected profit and equilibrium order quantities? Based on extensive numerical simulation, I find that when the profit margin is high, the impact of green metrics is limited—there is almost no change to a slight decrease in expected profit and the equilibrium order quantity when green metrics are considered. However, when the profit margin is low, the green metrics may improve the expected profits while reducing equilibrium order quantities. Interestingly, introducing green metrics does not affect coordinating transshipment prices, irrespective of profit margins. Direct versus indirect metrics have a limited impact on equilibrium order quantity and expected profit. My study contributes to the research by identifying the operational benefits of adopting green metrics. As an extension, this work may create a foundation for further work to determine the cost and benefits of implementing green metrics in practice and the key trade-offs in sustainability or social responsibility
Assessing the Impact of Air Leakage on the Hygrothermal Performance of Wood-Frame Walls Under Historical and Future Climates
Get PDFAir leakage is a crucial factor when assessing the hygrothermal performance of wood-frame walls since it can lead to moisture accumulation during the cold season. The seriousness of this prob-lem may change in a warmer climate in the future and hygrothermal simulations are widely used as a tool to predict this effect. However, since 2D models are required for detailed air leakage as-sessment and the high number of input variables leads to having to conduct thousands of simula-tions for a single type of building cladding, downsizing the simulation grid to the lowest number of cells is a crucial task to help ensure reduced computational time. Using a hygrothermal simulation tool, the steps needed to build the smallest 2D grid were explained; as well, convergence and ac-curacy of the results were evaluated and the functional relations between air leakage rate and air permeability of the insulation were clarified. Hygrothermal simulations were performed for wood-frame walls having brick veneer and stucco cladding for three Canadian cities: Whitehorse, Van-couver, and Ottawa. While the air leakage rate has a significant impact on the inner surface of OSB, wind driven rain is the key factor on the outer surface. The performance of stucco cladding is worse than brick in all cases and the future climate may reduce the risk of mould growth on the in-ner surface of OSB in all cities. The results also show that the simulation time can be reduced by 90%, with negligible loss of accuracy, when comparing fine to optimized meshes
An Industrial Study on Predicting Crash Report Log Types Using Large Language Models
Get PDFSoftware crashes and failures take a fair amount of effort and time to resolve. Software developers
use information submitted in crash reports (CRs) to conduct root cause analysis of faults. The
problem is that CRs often lack all the information required. Automatic prediction of CR fields can
therefore reduce the crash resolution process time. In this thesis, we use CR headings and
descriptions to predict the type of log files that should be attached to a CR. Our approach is to use
multilabel learning algorithms to train a machine learning model using a dataset from Ericsson’s
CR database to predict the type of log files based on CR headings and descriptions. We use three
different pre-trained language models Bert, Telecom Bert, and Word2Vector to extract feature
vectors from CR headings and descriptions and then feed these vectors to three different multilabel
learning algorithms, namely Binary Relevance (BR), Classifier Chain (CC), and Neural Network
(NN). Then, we compare the performance of different feature sets. We found that the use of
headings alone with pre-trained language models Bert and Telecom Bert results in the best average
AUC (0.70). The use of descriptions and headings and descriptions together as features resulted in
an average AUC varying from 0.65 to 0.70. In general, the algorithms showed no significant
difference in their performances, but the choice of features impacts the performance. Also, the
performance of predicting each type of log is influenced by the use of keywords in headings and
descriptions that describe these files. We found that log types with a clear definition such as Key
Performance Indicators (KPI) Logs, Post-mortem Dumps (PMD), and execution traces can be
predicted with higher accuracy
Electrified Natural Gas Pyrolysis to Produce Low-Carbon Hydrogen
Get PDFElectrified plasma-assisted natural gas pyrolysis (PNGP) emerges as a promising technology for low-carbon hydrogen production in this thesis, encompassing process simulation, economic evaluation, and environmental impact assessment. The technical analysis demonstrates an impressive carbon yield of approximately 95% from natural gas, showcasing highly efficient decomposition to carbon particles and hydrogen. However, PNGP requires a specific energy of 16.2 kWh/kgH2, which surpasses conventional steam methane reforming (SMR). Economic evaluations reveal the hydrogen minimum selling price for PNGP to be 0.4/kg, reduces PNGP’s minimum selling price by $1/kgH2, enhancing its competitiveness with SMR. Environmental assessments underscore PNGP’s potential for mitigating greenhouse gas (GHG) emissions when integrated with renewable
electricity sources. Comparisons with SMR and PEM electrolysis emphasize the importance of carbon black production credits in bolstering PNGP’s economic and environmental performance.
Furthermore, a comparison of the cost of avoided/captured greenhouse gas (GHG) emissions between PNGP and SMR with CCS shows that PNGP offers competitive advantages, particularly
when the carbon black production credits are considered in the system. At certain electricity and natural gas price levels, PNGP’s cost of avoided/captured emissions becomes more favorable compared to SMR with CCS. This finding highlights the potential of PNGP as a promising option for low-emission hydrogen production, considering both its technical performance and environmental
sustainability
Hacking AI Governance: Exploring the Democratic Potential of Canada's Algorithmic Impact Assessment
Get PDFAmid growing concern over the adoption of artificial intelligence systems, algorithmic impact assessments (AIAs) have increasingly been proposed as a means of measuring and mitigating the impacts of AI. Proposed AIA methods vary significantly in their approaches, but even within this heterogeneous group, the AIA tool released by the Government of Canada in 2019 stands out. This AIA tool—an open-source, online questionnaire platform—represents one of the first-ever attempts at putting principles of “responsible AI” into practice. In this research-creation thesis, I explore Canada’s AIA tool as a media object, looking at the online questionnaire as a strategic opportunity to intervene in the growing debates about AI governance. Building on methods in critical making and civic hacking, this project includes the creation of both a critical guide to the AIA tool (aia.guide) and a series of “AIA hackathon” workshops designed to explore the tool’s use by the Government of Canada and its potential in the broader AI governance context. Informed by a deep ambivalence over the technology (Bucher 2019), I argue that the AIA tool is largely performative but also represents an important site for tactical intervention. In particular, I argue that collaborative processes of questionnaire design may prove to be effective methods for participatory and community-based AI governance
The Impact of Sustainable Transit Availability on Health Inequality in Canadian Cities, 2006 to 2016
Get PDFFocus on sustainable transit has grown in recent years, as Canada invests in active and public transport (Infrastructure Canada 2023), and plans to reduce CO2 pollution (Government of Canada 2022). While the health benefits of commuting by walking, biking or even public transit may seem clear, the impact on health inequality within cities is less so; some researchers claim that strong transit systems equalize access to health care (Abu-Qarn and Lichtman-Sadot 2022), while others argue that uneven implementation of sustainable transit may lead to gentrification in transit-accessible neighbourhoods, leading to worse outcomes for vulnerable residents (Tehrani, Wu, and Roberts 2019). This investigation seeks to determine the impact of sustainable transit availability on health inequality in Canadian cities using cross-sectional regression analysis. We use the gap in the hospitalization rate between the highest and lowest income quintiles as a proxy for health inequality. Walkability is found to be related to a smaller gap, while bikability is associated with a wider gap. This may be explained by bikability and transit being associated with gentrification, mitigating any positive effects they may have had on the gap in hospitalizations
Multi-Scenario Land Use and Land Cover (LULC) Change Projection Framework Using Markov Chain and PLUS Integrated Model
No full textThe spatial distribution of urban land use has undergone significant transformations due to rapid urbanization. Assessing the dynamic and complex interactions of land use and land cover (LULC) can help planners and policymakers understand the extent and effects of those changes. This study proposes a novel framework for land use and land cover (LULC) change through the integration of patch-generating land use simulation (PLUS) and Markov Chain (MC) model under different scenarios. Various simulations have been conducted for the island of Montreal, Quebec, Canada using regional land use types under the five shared socioeconomic pathways (SSPs) for the year of 2028. In addition, a comparative study was conducted between three major cities in Canada: Toronto, Ottawa, and Montreal, in which global land use types were used to project LULC change in 2030 based on historical trends. Different accuracy measures were calculated to validate our model and compared to the accuracy of other models reported in the literature. Our findings show that our model achieved a higher figure of merit (FoM) than other models and was able to simulate LULC change without the need for expert knowledge in the field. The results of this multi-scenario simulation and ecological, environmental effect study can be used as a reference for future regional territorial spatial planning and policy formulation. The integration of the PLUS and Markov Chain models is shown to be quite applicable to the projection and assessment of urban spatial land use patterns
Arithmetic and computational aspects of modular forms over global fields
Get PDFThis thesis consists of two parts. In the first part, we present a positive characteristic analogue of Shimura's theorem on the special values of modular forms at CM points. More precisely, we show using Hayes' theory of Drinfeld modules that the special value at a CM point of an arithmetic Drinfeld modular form of
arbitrary rank lies in the Hilbert class field of the CM field up to a period, independent of the chosen modular form.This is achieved via Pink's realization of Drinfeld modular forms as sections of a sheaf over the compactified Drinfeld modular curve.
In the second part of the thesis, we present various computational and algorithmic aspects both for the classical theory (over C) and function field theory. First, we implement the rings of quasimodular forms in SageMath and give some applications such as the symbolic calculation of the derivative of a classical modular form. Second, we explain how to compute objects associated with a Drinfeld modules such as the exponential, the logarithm, and Potemine's set of basic J-invariants. Lastly, we present a SageMath package for computing with Drinfeld modular forms and their expansion at infinity using the nonstandard A-expansion theory of López and Petrov
Assessing the Emerging Environmental Concerns from Bio-originated Organic Pollutants in Cropping Systems
Get PDFAgriculture plays a significant role in achieving the goal of carbon neutrality and emission reduction through practices such as crop residue management. Crop residues can be utilized to produce biodegradable mulches (BMs), which can increase crop production and carbon sequestration potential. However, agricultural health and safety are facing new challenges, particularly concerning bio-originated organic pollutants in cropping systems, including biogenic volatile organic compounds (BVOCs) and biodegradable microplastics (BMPs). The main purpose of the research presented in this dissertation is to assess these emerging environmental concerns, including the appropriate assessment of BVOC emissions and the degradation and fragmentation of BMs.
BVOC emissions were generally influenced by various factors, including temperature, drought, solar radiation, humidity, nutrient availability, carbon dioxide (CO2), ozone (O3), etc. Among these factors, growth length, air temperature, solar radiation, and leafage were found to be the most important variables affecting the spatial-temporal variations of methanol (MeOH) emissions from spring wheat during the growing period in a Canadian province. The seasonality of MeOH emissions was positively correlated with concentrations of carbon monoxide (CO), filterable particulate matter (FPM), and coarse particulate matter (PM10), but negatively related to nitrogen dioxide (NO2) and O3. Compared with paper mulch, bioplastic mulch contributed a higher amount of aromatic structure-containing chemicals and carboxylic acids, to the water environment, but released fewer and smaller plastic particles. After entering the soil-water environment, the rough microstructure and oxygenated functional groups on BMP surfaces played a crucial role in the adsorption of aromatic compounds and heavy metals from soils.
Scientometric analysis can provide researchers with an in-depth understanding of BVOC emission mechanisms, while also offering decision-makers insights into emission mitigation and environmental management. The newly developed BVOC assessment approach, designed to evaluate the biogenic MeOH emitted from crops during growing seasons, can help uncover the relationships between BVOC emissions and key influencing factors. The characterization and quantification of BMPs in cropping systems focused on examining the fragmentation and degradation of BMPs under UV irradiation using visual inspection and quantitative analysis. This dissertation offers scientific support for researching and further developing the impact of BVOC emissions and BMP generation on environmental management