3789 research outputs found
Sort by
Vapnik-Chervonenkis dimension in neural networks
Get PDFThis thesis aims to explore the potential of statistical concepts, specifically the
Vapnik-Chervonenkis Dimension (VCD)[33], in optimizing neural networks. With the
increasing use of neural networks in replacing human labor, ensuring the safety and
reliability of these systems is a critical concern. The thesis delves into the question
of how to test the safety of neural networks and optimize them through accessible
statistical concepts.
The thesis presents two case studies to demonstrate the effectiveness of using
VCD in optimizing neural networks. The first case study focuses on optimizing the
autoencoder, a neural network with both encoding and decoding functions, through
the calculation of the VCD. The conclusion suggests that optimizing the activation
function can improve the accuracy of the autoencoder at the mathematical level.
The second case study explores the optimization of the VGG16 neural network
by comparing it to VGG19 in terms of their ability to process high-density data. By
adding three hidden layers, VGG19 outperforms VGG16 in learning ability, suggesting
that adjusting the number of neural network layers can be an effective way to analyze
the capacity of neural networks.
Overall, this thesis proposes that statistical concepts such as VCD can provide a
promising avenue for analyzing neural networks, thus contributing to the development
of more reliable and efficient machine learning systems. The final vision is to allocate
the mathematical model reasonably to machine learning and establish an idealized
neural network establishment, allowing for safe and effective use of neural networks
in various industries
Adding time-series data to enhance performance of naural language processing tasks
Get PDFIn the past few decades, with the explosion of information, a large number of
computer scientists have devoted themselves to analyzing collected data and applying
these findings to many disciplines. Natural language processing (NLP) has been one of
the most popular areas for data analysis and pattern recognition. A significantly large
amount of data is obtained in text format due to the ease of access nowadays. Most
modern techniques focus on exploring large sets of textual data to build forecasting
models; they tend to ignore the importance of temporal information which is often
the main ingredient to determine the performance of analysis, especially in the public
policy view. The contribution of this paper is three-fold. First, a dataset called
COVID-News is collected from three news agencies, which consists of article segments
related to wearing masks during the COVID-19 pandemic. Second, we propose a
long-short term memory (LSTM)-based learning model to predict the attitude of the
articles from the three news agencies towards wearing a mask with both temporal
and textural information. Then we added the BERT model to further improve and
enhance the performance of the proposed model. Experimental results on the COVIDNews dataset show the effectiveness of the proposed LSTM-based algorithm
Advances in operations research models used in the gold mining industry
Get PDFThe topic addressed in this dissertation is a set of economically important operational
problems in the gold mining industry that are solved using mathematical models of operations
research. More specifically, the main objective of this thesis is to formulate and evaluate decision
support models for three important diverse challenges which were found to exist at an underground
gold mine in Northwestern Ontario: Newmont Goldcorp’s Red Lake Gold Mine. The challenges
discovered at Red Lake Gold Mine are not peculiar to that location but are economically relevant
to the underground gold mining industry in as whole. The mine at Red Lake provided a deeper
understanding of the problems and data sets.
The challenges modeled and solved in this dissertation are: i. minimizing freshwater used in the
processing of gold ore; ii. optimizing ore-waste material flow in an underground gold mine; and
iii. optimal dispatching of trucks and shovels in an underground gold mine. Each of the three
problems was treated with a formulation of the model which is innovative and the evaluation of
the results of each case study showed that improved decisions can result when these models are
used.
This dissertation shows that, for a single gold mine, problems of major economic importance can
be found, innovatively modeled, and solved using the methods of operations research. In addition,
since these problems are not peculiar to one gold mine, but are found in other gold mines, the
innovation of this dissertation is relevant to the underground gold mining industry as a whole and
therefore constitutes a minor but important advance in the practical knowledge in this industr
Effects of macrophyte cutting on a whole lake ecosystem
Get PDFMacrophytes are of significant importance to aquatic ecosystems, generating
primary production in nearshore environments and providing physical structure and
habitat for organisms in the littoral zone of lakes. Macrophyte cutting is a common
practice near human settlements, used to dampen the negative perceived effects they have
on human activities. As such, understanding the impacts of vegetation cutting on both the
lower and higher trophic levels can provide insight into impacts on the whole lake
ecosystem. Impacts of macrophyte cutting on the whole-lake ecosystem were assessed at
Lake 191 of the IISD-ELA, where 2 years of pre-experimental monitoring (1994 - 1995)
were followed by 3 years of macrophyte cutting (1996-1998). After cutting occurred,
macrophytes were allowed to re-establish and post-experimental monitoring occurred
from 1999 until up to 2003. Results from this experiment showed decreased light
penetration and decreased relative macrophyte biomass at 0.5m depth in 2000.
Phytoplankton community composition became more variable, and biomass increased
during macrophyte cutting. Daphnia pulex, Daphnia catawba, and Daphnia schoedleri
collectively and Diaptomus oregonensis saw the greatest biomass changes within the
zooplankton community. [...
The use of citizen science data to predict the winter distribution of the snowy owl in Ontario
Get PDFAs an Anishinaabe person and environmental scientist, I feel I have an inherent duty to
be a steward of the environment. I chose to conduct research on Snowy Owls after
learning about the conservation issues affecting this species, namely habitat loss due to
climate change. I hypothesized that using citizen science data and bioclimatic
(BIOCLIM) variables would effectively predict the winter distribution of Snowy Owls
in the province of Ontario s. I created a species distribution model (SDM) based
upon occurrence data from the eBird and iNaturalist citizen science databases and
bioclimatic variables using the computer software Maxent. The occurrence data was
cleaned to a scale of 1-km2 using QGIS to mitigate impacts of sampling biases in citizen
science data. Though it had a relatively high AUC of 0.848, the final SDM was
inaccurate in predicting the winter distribution of Snowy Owls in Ontario. Sampling
biases inherent in citizen science data, possibly exacerbated by the permutation
importance of the chosen BIOCLIM variables, were found to heavily skew the results of
the SDM. The SDM indicated that Snowy Owls are more populous within Southern
Ontario than Northern Ontario, and within developed areas highly populated by humans.
Careful consideration of sampling biases and their magnitude of impact is recommended
when developing an SDM based upon data acquired via citizen science databases.
Further research with alternative methodologies is required to develop an appropriate
SDM of winter occurrence of Snowy Owls in Ontario
Plant mixture effects on fine-root biomass and its functional traits
Get PDFFine roots play a critical role in the uptake of soil water and nutrients and make a crucial
contribution to the carbon pool through their fast turnover rate and subsequent decay. Plant traitbased approach enables us to understand the plant growth strategy via the ratio of benefit to cost
of carbon investment, especially under global change. Increasing biodiversity loss threatens the
ecosystem productivity, which could further influence the fine-root functions. However, previous
studies have reported inconsistent responses of fine-root biomass and root functional traits to the
effects of plant species richness and functional trait dissimilarity (such as contrasting shade and
drought tolerance). The purpose of this dissertation is to reveal the possible mechanisms of
different responses of fine-root biomass (FRB) and root functional traits to plant species richness
and functional trait dissimilarity, and further to test whether these mixture effects would change
with water availability. [...
Multicriteria feasibility assessment of BioSuccinic acid production from lignocellulosic biomass
Get PDFThe goal of any lignocellulosic based biorefinery should be to produce a spectrum of marketable
products and energy utilizing all significant components of biomass. Depending on the maturity
of available technology, biorefineries can target high-value low-volume (HVLV), middle-value
middle-volume (MVMV), and low-value high-volume (LVHV) outputs provided they are
economically feasible. BioSuccinic Acid (BioSA), a MVMV product, has considerable potential
as a candidate for biorefineries based on an analysis we carried out initially. The total world
production of Succinic Acid (SA) in 2013–2014 was 38,000 tons, valued at 2.90 USD/kg SA and
with predicted worldwide market demand of 94,000 tons by 2025. However, the most crucial
challenge encountered with BioSA production is the cost of production compared to its
conventional fossil-based counterpart. About a decade ago, many industries were set up to produce
succinic acid from starch-based renewable sources when the cost of petroleum and petroleumbased products was higher than 100 dollars a barrel. The drop-in petroleum prices have led to the
closure or re-orientation of some of these industrial units. [...
Pre-feasibility study of applying a biomass-powered district energy system in Marathon, Ontario
Get PDFWith the energy price fluctuation the nation is currently experiencing, more and more people are
now looking into biomass as a substitute energy resource. Northwestern Ontario, with a history
of forestry operations and management for over a hundred years and a substantial net annual
growth of wood, has the potential to produce enough biomass to support the energy demand of
the local communities as well as take a portion of the national or international market. There
have been several previous studies within the region of Northwestern Ontario to assess the
possibility of applying biomass heating in remote communities to reduce the cost as well as add
energy supply stability. In this article, we examined the feasibility of applying a biomass-powered district energy system (DES) in Marathon, ON. A biomass-powered DES is proposed to
be constructed in the town center to supply the surrounding public buildings with heat. The cost
of the DES is 2,075,249 on fuel, which will make the return period of the initial investment 8.737 years.
The DES will also bring a GHG reduction of 3,712 tons annually
Disconnect between Indigenous traditional connection with water due to water insecurity
Get PDFPeople indigenous to what is now Northern Ontario have always had a traditional
connection and relationship with water. In recent year, that relationship has been put
through many obstacles amounting to water insecurity. The effects of water insecurity
have changed Indigenous people’s traditional connection and relationship with water,
causing a disconnect for various reason. This relationship must be fixed and made even
stronger than before with the help of all levels of government, non-Indigenous people
and Indigenous people alike, and a multi-step approach
The effects of knee bracing on reactive agility performance among healthy soccer players – a pilot study
Get PDFSoccer is the most popular sport in the world. The increase in the sport’s
popularity is paralleled with an increased prevalence of knee injuries. Knee braces are commonly
worn in athletic populations to prevent knee injuries. The biomechanics of wearing knee bracing
have been well documented, with studies showing reduced vertical ground reaction forces. One
concern with wearing knee braces, however, has been the possible effect on sports performance,
the research examining this topic has been confounding. Some studies have shown that agility
time has improved during an agility T-test, while others have shown no change in agility time.
To date, no studies have examined the effects of knee bracing on reactive agility performance.
The measurement of neuromuscular activity is known as electromyography. Electromyography
is also another area of interest with regards to sport performance. To date, there is limited
research on the effect of the application of a knee brace on the electromyography of various
lower extremity muscles during cutting maneuvers. More specifically, no studies have explored
the effect that knee braces may have on the peak muscular activation on the gluteus medius
during an agility task. Therefore, the purpose of this pilot study was to examine differences
between braced and non-braced soccer players on measures of reactive agility time (s), and EMG
activity (% MVC) of the GM, BF, and VL during the acceleration and change of direction phases
of the Y-shaped reactive agility test. [...