AUB ScholarWorks (American Univ. of Beirut)
Not a member yet
33634 research outputs found
Sort by
Automatic Reconstruction of Glass Relics Using Manifold Learning
Get PDFThe risk imposed by manual reassembly on valuable broken relics necessitates automating this process by leveraging computer vision for 3D data acquisition, and data science to extract features of interest from high dimensional data. This thesis proposes a solution for the automatic reassembly of broken glass relics. The solution first relies on digitizing the broken shards. After that, contours of shards are extracted and segmented. Next, the proposed system maps the segments into the space of manifolds to quantify the similarity in their local geometry and uncover pairwise matches among them. Finally, a global optimization step finds the overall solution of the reassembly problem and the shards are aligned to visualize a digitally reassembled relic. This digital solution is then used in an application that runs on a head-mounted AR device to guide users through the process of sequentially reconstructing the real relic. The focus of this thesis is on the reconstruction part of the problem. The proposed system is discussed and verified over a dataset of broken glassware. Experiments on ten manually broken glass relics validate the success of the proposed approach by estimating the correct position of each shard in the reassembled relic. Moreover, the performance of the proposed system was tested for two extreme scenarios: missing shards and intruder shards. The system showed robust performance in finding matches among shards, however, the alignment of shards around missing pieces was effected
A Citizen Engagement Application to Address Issues Related to Legal Violations and Environmental Prejudice
No full textThis report reviews existing literature on crowdsourcing, citizen engagement, evaluation methodologies, and application ratings to inform the development of a novel citizen engagement application designed to address legal violations and environmental prejudice.
Based on this literature review, we worked on the development and implementation of a citizen engagement application designed to address legal violations and environmental prejudice. By leveraging mobile technology and crowdsourcing, the application empowers individuals to report a wide range of issues, from environmental violations such as pollution and illegal dumping to noise pollution. The platform facilitates the submission of detailed whistleblowing including photos and pinning locations, which are then relayed to relevant authorities for prompt action. The application aims to foster a collaborative relationship between the public and domain exports, enhancing transparency and accountability. This approach not only addresses immediate violations but also promotes long-term environmental stewardship and legal compliance. Through case studies, user feedback, and platform evaluations, this report demonstrates the application's impact on improving community involvement and environmental protection. Additionally, it discusses potential future enhancements that pushes this app up to another level. By examining the success and challenges of this initiative, the paper emphasizes the potential of citizen engagement applications to drive significant social and environmental change, highlighting the importance of empowering individuals to actively participate in safeguarding their communities and the environment
Regulating the Residential Solar Boom in Lebanon: Policy Frameworks and Strategic Initiatives for Photovoltaic Transition
Get PDFThe economic and financial crisis in Lebanon that started in 2019 has exacerbated the shortfalls of the Lebanese energy sector. This is due to the emerging inability to continue funding the imports of fuel oil, the main component that the sector relies on, with hard currencies. Amidst increased and critical power outages, Lebanon has witnessed a significantly observable boom in the installations of distributed solar PV to achieve independence from the failing fuel-powered grids, especially in the residential sector. Installations, however, have been chaotic and unregulated, with the government not playing its role in instating the proper legislative, institutional, and policy environment for these installations to take this transition in a strategic approach. With the current trend, the Lebanese energy sector might be headed in a direction that requires even more reforms down the line. This study aims to assess both the active and dormant laws and institutes that are responsible for regulating distributed solar PV installations and to perform a comparative analysis with other countries across the globe to draw policy recommendations for Lebanon. This study will offer a better understanding of what needs to be done in Lebanon in the upcoming years and aid in instating relevant reforms and policy recommendations
Remote Sensing Using Hyperspectral Imaging for Paint Thickness Evaluation and Degradation Assessment
No full textProtective coatings are essential in many industries because they protect structures by isolating the underlying materials, or substrate, from harmful environmental factors. When coatings begin to fail or degrade, the exposed substrate deteriorates. Substrate deterioration and corrosion demand maintenance with high costs and may pose potential safety risks. Typically, paint condition assessment techniques rely heavily on visual inspection, which is limited to detecting mechanical or advanced stages of degradation. Other techniques like eddy current gauges are used to measure paint thicknesses, as thickness reduction is a sign of paint degradation, but this contact method is not practical for assessing large surface areas. Many industries tend to carry out scheduled repainting based on operating hours without considering the actual paint condition. Spectroscopy methods, like Fourier-transform infrared spectroscopy (FTIR), are an effective and reliable method for paint assessment, but impractical in the field setup due to scalability issues. On the other hand, hyperspectral imaging (HSI) which is classified under spectroscopy techniques has emerged as a new method with a high potential for paint condition assessment. In our study, we have developed a comprehensive framework that combines HSI and machine learning (ML) to predict both paint thickness and the condition of the paint under accelerated ageing tests. Aluminium plates painted with aliphatic polyurethane-based paint have been prepared with different paint thicknesses and various ageing states. The HYSPEX SWIR 384 camera for data collection within a spectral range of 930-2505 nm and a QUV chamber was used to age the samples according to ISO 11507 Method A standard with a maximum period of 1600 hours.
Various machine learning (ML) models were explored to predict paint thickness. Traditional methods like Nu-SVR and SGD performed well, but when their performance compared to DNN’s, with four hidden layers, DNN stood out. It reached an RMSE of 21.6 µm and R2 of 0.97, predicting thicknesses from 43 to 499 µm. For the degradation assessment, the 1D CNN model delivered the highest performance with an R2 of 0.94 and RMSE of 125 hours in predicting ageing hours from HSI data for 500- and 1000-hour aged paint samples. Extending the model to 250 and 800 aging hours, R2 achieved 0.90 and an RMSE of 157 hours. The obtained results demonstrated that our model captures well the relationship between paint thickness/degradation and the reflectance values. These results demonstrate the effectiveness of HSI for paint condition assessment. This solution can lead to more reliable inspections and better maintenance strategies. Future work will include field experiments on a large-scale structure
Untangling the GDF15-NETosis Connection: Implications for Targeted Therapy in Diabetic Cardiomyopathy
No full textBackground: Diabetic Cardiomyopathy (DCM) is a pathophysiological condition induced by diabetes mellitus and characterized by cardiomyocyte death, hypertrophy, and fibrosis. Diabetes-induced alterations in endothelial and cardiac muscle cells have been reported to be major causative factors in the onset and progression of DCM. An alteration in the expression of cardiac fibrosis markers has been identified as a consequence of DCM. Furthermore, accumulating evidence associates chronic hyperglycemia with oxidative stress as a pathway that facilitates DCM progression. There is increasing evidence for the expression of NOXs, particularly NOX4, in the heart, contributing to the pathogenesis of cardiac hypertrophy, fibrosis, inflammation, and progression towards heart failure (HF). Recently, NOX-generated reactive oxygen species (ROS) are proposed to mediate a neutrophil-specific cell death process known as NETosis which was found to be associated with diabetes, as neutrophils are the first responders and secrete neutrophil extracellular traps (NETs), which induce sustained inflammation. On the other hand, Growth Differentiating Factor 15 (GDF-15), a member of the transforming growth factor β (TGF β) cytokine superfamily that acts as a stress-responsive cytokine, and has been shown to regulate neutrophil arrest and counteract its recruitment. However, a direct correlation between GDF-15 and NETosis and the role of their crosstalk in diabetes-induced cardiovascular complications remains to be elucidated. Hypothesis: We hypothesize that GDF-15 inhibition using an AV-380 monoclonal antibody ameliorates diabetic cardiomyopathy by attenuating NETosis. We also inspect the influence of AV-380 on cardiac function, injury, fibrotic markers, and oxidative stress, particularly focusing on NOX4 and NADPH signaling. Materials and Methods: Two sets of animals were utilized to assess our hypothesis. C57BL/6J male mice were used in this experimental design. In the first animal model, they were divided into six groups: (1) Control mice receiving vehicle (saline), (2) Control mice receiving 7.5 mg/kg AV-380, (3) Control mice receiving 20 mg/kg AV-380, (4) type 2 diabetes mellitus (T2DM) mice receiving vehicle (saline), (5) T2DM mice treated with 7.5 mg/kg AV-380, and (6) T2DM mice treated with 20 mg/kg AV-380. The treatment duration was 8 weeks. T2DM was induced by five consecutive doses of 55 mg/kg streptozotocin (STZ) and fed a high-fat diet (HFD). This set of animals was followed by a similar animal model excluding Control mice receiving 7.5 mg/kg AV-380 or 20 mg/kg AV-380 and the duration of the treatment was extended to 15 weeks. Following sacrifice and organ harvest, functional, histopathological, and molecular analyses were performed on the left ventricle of all groups. Results: Our findings indicate that inhibiting GDF-15 with AV-380 can restore cardiovascular homeostasis in T2DM mice. This is evident through reduced proteinuria, decreased superoxide generation and NADPH oxidase activity, reduced IL-1β and IL-6 levels, as well as improved cardiac ejection fraction and fractional shortening in the treated diabetic mice. Interestingly, the inhibition of GDF-15 significantly reduced NETosis markers. Conclusion: Our findings highlight GDF-15 as a plausible therapeutic target in DCM, alleviating inflammation, oxidative stress, and structural malfunctions in the heart and restoring cardiovascular homeostasis. Further studies are warranted to clinically validate these findings and to assess the relevance of AV-380 as a novel therapeutic intervention for cardiovascular disorders
LEVERAGING A BILSTM-BASED EMOTION RECOGNITION TRANSFER LEARNING MODEL TO IDENTIFY ABUSIVE LANGUAGE PATTERNS FOR COMPLEX PHRASAL ANALYSIS IN CYBERBULLYING DETECTION
No full textIn the last two decades, the penetration of Social Networking Sites and Social
Media (SNS/SM) platforms has risen to include more than one-third of the global
population. However, the use of SNS/SM has produced both positive and negative
results so much so that there have been calls to researchers pay immediate and far
greater attention to these contradictory effects of SNS/SM capabilities. A key negative
impact is the fast and significant rise of cyberbullying.
Cyberbullying has emerged as a serious act on social networking sites/social
media (SNS/SM) platforms in today’s digital society. Statistics underscore this whereby
42% of individuals indicate they have experienced cyberbullying, more specifically
38% of females and 54% of males have also experienced some form of bullying. This
form of dysfunctional social act is expressed via aggression, harassment, and toxic
behavior poses severe consequences to increasing penetration of SNS/SM.
Simultaneously, there has been a great awareness of the need for moderating
contents on (SNS/SM) to detect and reduce cyberbullying contents. It is also recognized
that human content moderation on SNS/SM is impractical and too costly. Therefore,
there is an increasing need for accurate methods of content moderation that are less
reliant on human judgement and instead employ sophisticated machine learning
methods. A key shortcoming of the current machine learning approaches to
cyberbullying detection is that their accuracy needs to be improved significantly to be
relied upon for practical deployment of non-human content moderation on SNS/SM.
Problem of using advanced data analytics and AI-based techniques in
cyberbullying detection in the service of reducing the latter has been extensively
studied. However, existing research method have faced challenges with the issue of
false positives whereby many identified instances may not be cyberbullying. For
instance, “I hate you” and “I hate thinking about the future” both contain the word
“hate”, yet the second sentence is a non-cyberbully phrase/sentence that contains the
word hate as a metaphor for personal discomfort with the uncertainty associated with
one’s future.
Indeed, more accurate detection can come from a deeper contextually sensitive
detection methods going beyond identifying simple hate words, whereby distinguishing
3
between harmful and innocuous expressions becomes the focal research problem. To
address this, we will conduct a complex phrasal analysis to identify cyberbullying
through the detection of the underlying emotion behind the seemingly toxic phrases.
This approach aims to mitigate false positive predictions, which can occur when relying
on hate keywords or an online hate corpus.
To achieve our goal, we employ an emotion recognition transfer learning model
to comprehend the underlying emotion trigger of cyberbullying and enhance its
detection. To accomplish this, we use a Bi-LSTM pre-trained emotion detection model
with a 92% accuracy on the training set. Then, we adapt the knowledge gained from the
first model to improve learning on a different but related task which is improving
cyberbullying detection by integrating elements of context based on emotion
identification within the expressed phrase. Based on extensive testing and training of
the model on the data, we propose that LSTM and CNN multi-label-based classification
model which is exhibiting an 80% accuracy on the training as superior approach to
cyberbullying detection.
To evaluate and validate our results, we randomly split our dataset into training
and validation sets, testing against an unseen dataset. A comparative analysis with
(Gencoglu, 2021) model reveals a significant improvement in performance using
McNemar’s test and the Paired T-test. Notably, our model shows an 18% improvement
on (Gencoglu, 2021) model.
In summary, this research has addressed the limitations inherent in keyword-based cyberbullying detection methodologies, thereby making a meaningful
contribution to the field of cyberbullying detection in SNS/SM research
From Knowledge Production to Impact: Research as Environmental Activism
Get PDFThis discussion is the third of AUB-NCC’s 2024 series of roundtables, titled “The New Roadmap to A Green Transition: Exploring Today’s Climate Actors” Roundtable discussion organized on September 17th, 2024, at the Basile Antoine Meguerdiche Conference Hall - IFI.The series is co-designed with: Nature Conservation Center, UN-Habitat, Cewas, and Synaps.Panel: Ms. Dana Hourany, Journalist, The Public Source ; Dr. Fadi El-Jardali, Professor & Director of Knowledge to Policy (K2P) Center, AUB ; Ms. Kristine Beckerle, Economic, Social and Cultural Rights Advisor, Amnesty MENA ; Ms. Maya Gebeily, Bureau Chief for Lebanon, Syria and Jordan, Reuters.This discussion marks the third session in AUB-NCC’s 2024 series of roundtables, titled “The New Roadmap to a Green Transition: Exploring Today’s Climate Actors.” The panel entitled “From Knowledge Production to Impact: Research as Environmental Activism” explored how research contributes to advancing the environmental agenda and whether its impact can be effectively measured. It examined the role of research and education as direct contributors to practical environmental solutions and their ability to reach a global audience beyond the academic community. The discussion also addressed whether citizens should engage with research centers, the tools that facilitate inclusive research, and the integration of knowledge translation into knowledge production. Additionally, it tackled the obstacles that hinder the translation of knowledge into actionable outcomes, emphasizing the importance of bridging the gap between research, innovation, and tangible environmental progress.The series is co-designed with: Nature Conservation Center, UN-Habitat, Cewas, and Synaps
Impact of Reactor Design on the Chemical Absorption of Carbon Dioxide
No full textChemical absorption of carbon dioxide from point sources remains one of the most promising technologies to offset greenhouse gas discharges from heavy industries. A recent study highlighted the promise of using screen-type static mixers in intensifying CO2 capture using alkaline solutions and new insights into their flow dynamics led to the development of a novel variant. These new mixers utilize carefully designed, and strategically placed, inserts downstream of a woven mesh to further improve their mixing action. Therefore, the current study attempts to compare the performance of these different mixers for the intensification of CO2 absorption into a sodium hydroxide solution under various operating conditions and design configurations. Using the same reactor, seven equidistant screen mixers were employed while only three novel mixers were used. Two different screen geometries were utilized in each case and the gas and liquid flow rates were changed to maintain a total flow velocity ranging between 0.67 and 1.7 m/s. Pressure drop, removal efficiency, and specific energy consumption of each design were compared at the various gas and liquid flow rates with the gas phase volume fraction ranging between 0.09 and 0.38. The pressure drop per element of the novel design was on average 25.3 % higher than the screen mixers. However, the total pressure drop across the reactor equipped with the novel mixer was approximately 46.3 % lower than the case of screen mixers because of the lower number of elements. The removal efficiency enhanced with an increase in the liquid and gas flow rates, with the liquid flow rate having a more pronounced effect. The mixer geometry also impacted the CO2 removal efficiency where using the novel mixers yielded higher efficiencies than screen-type static mixing elements with the enhancement being a function of a smaller screen open area. Removal efficiencies as high as 95% were reached within 361 milli-seconds of contact time between the phases. In addition, the novel mixer geometries were found to require 40 to 300% less energy to achieve CO2 removal rates similar to those obtained using screen-type static mixers
Accelerating Knowledge Graph Relationship Queries on GPUs
No full textLarge graphs and networks, referred to as semantic graphs, have become essential
for discovering relationships between entities, objects, or concepts in modern-day
applications across various fields such as medicine, engineering, and business. Hence,
identifying relationships among sets of two, or more entities represents a critical
challenge in numerous analysis, search, and identification applications.
This challenge corresponds to finding the Steiner Tree within a given set of en tities, a well-known NP-Hard problem. Despite extensive research and proposed
solutions, the focus has primarily been theoretical, leaving a significant gap for
practical applications. Consequently, there is a need for real-world and fast im plementations. In this paper, we propose a GPU-accelerated implementation of a
heuristic algorithm tailored to our specific application domain. This approach aims
to alleviate the complexity of the problem, particularly when dealing with large
knowledge graphs. Our solution guarantees having a significant speed-up time com pared to CPU execution time and provides significantly optimized procedures to
achieve such speed-up
KGFusionX: Linking, Combining, and Exploring Data Through Knowledge Graphs
No full textIn the realm of data exploration, the persistent challenges of data disconnection
and inconsistency often hinder the efficiency of data analysts, especially in terms of
data enrichment and aggregation. This thesis focuses on addressing the following
research questions: How can we improve data integration and reuse of data in a clean
and downloadable format to facilitate data analysis? Moreover, how can we
contextually expand data on the fly to leverage its value and enhance data exploration?
This work proposes KGFusionX, a knowledge graph centered framework that
recognizes the time-intensive nature of data enrichment and integration. The study
employs a backend implementation utilizing knowledge graphs to seamlessly connect
disparate datasets. Several datasets from Lebanon covering different domains (e.g.
health care, economy, education, and others) were converted and published as openly
accessible knowledge graphs in a triple store repository (749,500 triples). This
conversion allows efficient and fast aggregation of data because of the connections
generated by knowledge graphs. Also, it is integrated with open linked data sources that
serves as a resource to expand the data. The framework is showcased through an online
platform built with Streamlit that allows users to select, combine, and download tabular
data that can be used in other visualization exploration tools (e.g. PowerBI and
Tableau). The approach was evaluated by data analysts and two use cases. Potential
pickup of our platform was expressed by users who relied on the tool to analyze school
and university challenges in rural areas, in addition to boosting tourism in Lebanon. The
results demonstrated a significant improvement in data exploration efficiency, and
better visuals with the knowledge graph-driven approach proving successful in
overcoming the challenges posed by disconnection, inconsistency, and enrichment. This
research primarily contributes to streamlining data exploration using the high potential
of knowledge graphs to support data aggregation, data enrichment and visual data
analysis