You will earn 6 research credits over 8 weeks, conducting faculty-supervised, hands-on, directed study research projects with results that will culminate in the preparation of a research paper. You will complete a minimum of 240 hours on research in and out of the laboratory.
Faculty mentors will work closely with you to direct your continued growth and knowledge development in the chosen research topic discipline.
| Course ID | Title | Credits | Syllabus |
|---|---|---|---|
| LIME RSLW 392S | International Independent Research in STEM Fields | 6 |
Endometriosis (EDS) is a debilitating, chronic, inflammatory, estrogen-dependent gynecological condition characterized by the presence of endometrial tissue-like lesions outside the uterus (often attached to the peritoneal wall or ovaries). Commonly associated with symptoms of chronic pelvic pain and infertility, EDS drastically impacts quality of life and imposes significant economic burden ($22 billion a year in the US [Al-Lami et al. 2024]) due to annual healthcare costs and productivity losses. Immunological dysfunction is proposed to be a critical facilitator of endometriotic lesion growth. Hence, there is a pressing need to explore the cellular and immunological factors that underlie this dysfunctional state and facilitate survival, invasion and vascularization in endometriotic lesions. A 3D-multicellular spheroid model containing human endometriotic epithelial (12Z cells), stromal cells (HESCs) and primary macrophages will be generated. RNA-sequencing of magnetic bead sorted macrophages will determine the activation state of macrophages in our 3D endometriosis spheroid model.
Goals of the research: What is the activation state of macrophages when in direct culture with endometriotic cells? Are macrophages M2-like or M1-like in terms of phenotype in our spheroids? How does this mimic the in vivo EDS immune microenvironment?
Desired qualifications: Practical experience in cell culture techniques is preferred but not necessary, as full training will be provided.
Europe generates over 25 million tons of plastic waste annually, with less than 30% recycled; the rest is incinerated or landfilled, exacerbating climate change. The traditional “take, make, waste” economy is unsustainable, and alternative waste-management solutions are crucial. Carbon recovery from plastics via thermal treatment under anoxic conditions can exceed 50%, significantly reducing CO2 emissions compared to incineration. This project aims to develop scalable methods to convert plastic waste into porous carbon materials (PCMs) for use as sulfur-hosting electrodes in lithium-sulfur (Li-S) batteries. The most commonly used consumer plastics such as low-density polyethylene (LDPE) and polyethylene terephthalate (PET) will be converted into carbon powders and processed into battery electrodes. The charge storage mechanism will be studied through operando X-ray diffraction and Raman spectroscopy. Findings will guide the design of PCMs with optimized structures, enhancing Li-S battery performance. This approach tackles both plastic waste management and energy storage challenges.
State of the Art Research in the Field:
Conversion of Waste Plastic to Useful Carbons. The carbon content for plastic materials can be ~86%, therefore there is a high incentive to produce value added carbons from waste plastics. Initial reports have demonstrated that various catalytic and thermal methods of preparation can be successful. However, further research is required to investigate the numerous parameters, which influence quality and yield. A recent review paper concluded that carbonization is a feasible route for the reutilization of waste plastic, in terms of its environmental impact. Combustion methods in air result in significant amounts of CO2 being released to the atmosphere (~3 tonnes of CO2 per tonne of PE plastic). The C recovery, for carbonization methods under anoxic conditions (without oxygen), is >50% meaning that at least half of the C contained in plastics can be recovered. Consequently, far less CO2 would be released, compared to presently used incineration methods.
Lithium–Sulfur Batteries. Li–S batteries are one of the most promising “beyond Li-ion” energy storage systems. However, many issues need to be addressed before widespread commercialization can be possible, including: (i) The low electronic conductivity of sulfur (S) and lithium sulfide (Li2S). (ii) The large volume changes (80%) which occur during cycling. (iii) The polysulfide shuttle effect. High-order lithium polysulfides (LPS) (Li2Sx (6 < x ≤ 8)), which are formed during Li–S battery operation, are highly soluble in the electrolyte. High-order LPS can migrate toward the anode, react with Li metal, reduce to lower-order LPS, and then migrate back to the cathode to form high-order LPS again, and so on, leading to a so-called polysulfide shuttle effect and resulting in low Coulombic efficiencies. To mitigate these issues, PCMs with sufficiently small pore sizes (≤ 1 nm) will be targeted to confine the soluble high-order LPS and mitigate polysulfide shuttling issues. The other advantages of using S-infilled PCMs will include: (a) enhancing the conductivity of sulfur and (b) providing high specific surface areas, which offer more electrochemical sites for increased S utilization.
Charge Storage Mechanism. The charge storage mechanism for Li–S batteries containing carbons derived from waste plastics will be determined through a correlation of results from electrochemical testing and advanced structural characterization.
Desired qualifications: Some experience with material synthesis is desirable but not crucial.
Relevant majors: Materials Science, Physics, Chemistry
The demands for healthcare services, by aging populations, is placing significant strain on European systems. In Ireland, national guidelines mandate that hip fracture patients admitted to emergency departments be transferred to orthopedics wards within four hours for optimal outcomes. The Irish Hip Fracture Database indicates that fewer than 25% of patients follow this pathway. To address this, interpretable machine learning models such as logistic regression, random forest etc. will be developed to predict patient length of stay, useful for better resource management of key resources such as beds, staff, and financial costs for running hip fracture events. This approach facilitates in advanced forecasting of resource demands for hip fracture events, enabling timely treatment. The proposed research will conduct exploratory data analysis to create a dashboard of hip fracture incidence in Ireland, highlighting population demographics, emergency department arrivals, and seasonal peaks.. This can ultimately lead to improving patient outcomes.
Desired qualifications: Python programming for Data Science and Machine learning such as pandas, numpy, sci-kit learn, matplotlib
Relevant majors: Data Science, Computer Science
Microfluidic array devices have revolutionized fields such as biomedical research, diagnostics, and chemical analysis by enabling precise control and manipulation of small fluid volumes. The design and development of these devices using 3D printing technology offer significant advantages, including rapid prototyping and customization. This project aims to explore innovative approaches to creating microfluidic arrays through advanced 3D printing techniques at the Bernal Institute in the University of Limerick. By leveraging the precision and versatility of resin-based 3D Printing with a 18 um resolution, this project seeks to enhance the functionality and scalability of microfluidic devices, paving the way for new applications in personalized medicine, environmental monitoring, and beyond.
Can resin 3D Printing be used to manufacture complex microfluidics devices for production of lipid and polymeric nanoparticles for pharmaceutical applications?
Desired qualifications: Experience with 3D CAD modeling is an advantage
Experience with 3D Printing/additive manufacturing is an advantage
Training in 3D CAD Modelling (Solidworks) and Resin 3D Printing will be provided for the successful candidate.
Relevant majors: Aerospace Engineering, Mechanical Engineering, Biomedical Engineering, Industrial Systems Engineering, Materials Science
MXenes are a remarkable family of 2D materials that have the potential to revolutionize the field of energy storage, particularly in the development of high energy density Li-sulfur batteries (LSB). With their exceptional properties, including high conductivity, mechanical strength, hydrophilicity, versatile morphology and abundant surface functionality, their application in LSBs can address several critical challenges associated with this technology including overcoming the shuttle effect through chemical and physical confinement of polysulfides, and negating the insulating effects of sulfur due to their excellent conductivity.
While MXenes are indeed a family of materials with exceptional properties, a major challenge needs to be addressed before they can be effectively applied commercially. The production of MXenes often requires the use of harsh chemicals, such as hydrofluoric acid, which poses safety concerns and environmental risks. This project will overcome this issue through the development of a sustainable, solvent-free solid-state synthetic approach to MXene production, unlocking their full potential through tailoring of their properties for high-energy density LSB applications.
Can tailored 2D nanostructured MXenes be synthesized using a sustainable, solvent-free approach?
Relevant majors: Chemistry, Physics, Materials Science
Cancer cell profiling helps identify specific molecular features and vulnerabilities in tumors. By studying how breast cancer cells respond to new drugs, researchers can discover biomarkers that indicate how well a treatment will work or identify tumors that are more likely to be sensitive to these drugs, ultimately improving the chances of treatment success.
Transcriptomics (measuring RNA transcripts) is a powerful tool for evaluating the effects of new drugs on gene expression in cells. Next-generation sequencing (NGS) enables us to sequence and quantify the mRNA transcripts of thousands of genes simultaneously, providing a comprehensive snapshot of gene expression patterns in response to drug treatment. Using NGS transcriptomics we will explore gene expression in breast cancer cells treated with a new anti-cancer epigenetic drug, developed in Dr Brown's group. This profiling reveals how the drug kills breast cancer cells, and helps identify patients where this drug will be most effective.
The development of generative AI systems has led to an increase in the use of such tools within the arts. Currently there is much discussion about the use of such tools and whether machine-created art is really art. This project will look at the current state of the art in generative AI, its use in the creation of artworks and the potential for it as a tool to augment artists rather than directly replace them. The project will involve user research with artists on the use of AI in art, as well as evaluation of existing generative AI tools with artists.
Desired qualifications: Useful if students have knowledge of generative AI tools. Some experience with user research would be useful. An interest in the creative arts would be beneficial.
Relevant majors: Digital Arts, Computer Science, Product Design, Psychology
Concussion is the most common form of mild traumatic brain injury and can cause long-lasting effects. Although the exact mechanisms are unknown, concussion is recognized as a subset of neurological injuries involving complex pathophysiological processes. There are many different causes of concussion, and not all of them involve direct impact. Non-impact situations, such as whiplash—one of the most common injuries following a car crash—can also cause concussions.
Developing more accurate models of concussive injuries will help researchers identify key markers post-injury, improving diagnostic processes and prevention methods. This project aims to develop a mechanical concussion rig to apply a non-impact rotational acceleration injury to a model skull, accurately mimicking concussion-inducing scenarios. It will explore current methods for studying concussions and address the limitations of previous iterations of this model developed at UL.
Alterations to the UL model will be designed and developed through the lens of biomechanical principles. The design can involve improvements aimed to improve the realism of the concussion rig as well as to increase the number of variables that can be tested (e.g., head gear, musculature changes). Testing will be performed to determine whether the implemented design changes improve the biomechanical mimetics and repeatable results.
Desired qualifications: Necessary: Engineering, optional: design methods, biology
Relevant majors: Biomedical Engineering
Eco-friendly, biodegradable electronics are crucial for reducing harmful electronic waste. However, the reliance on lead-based Pb(Zr,Ti)O3 (PZT) ferroelectric materials for transducers and actuators presents a significant challenge. Promising lead-free, eco-friendly alternatives include K0.5Na0.5NbO3 (KNN), (Na,Bi)TiO3/BaTiO3 (NBT/BT) composites, and (Ba,Ca)(Ti,Zr)O3 (BCTZ).
This project aims to develop fully solution-processed, biodegradable piezoelectrics, employing a screen-printing approach to create piezoelectric devices at low temperatures, and enabling their integration with eco-friendly substrates and electrodes. This approach involves using a printable ink to screen print potassium niobate (KNbO3) piezoelectric layers and sintering at low temperatures. The films will be poled, parallel plate capacitor and beam resonator will be designed from free-standing films and films on flexible biodegradable substrates to evaluate their dielectric and piezoelectric properties using impedance analysis and the berlin court method.
(i) Can screen printed structures possess piezoelectric properties
(ii) How piezoelectric and ferroelectric properties vary with low temperature sintering (120°C-350°C)
Desired knowledge: Crystal structure, Drude free electron model and quantum mechanical free electron theories, dielectric properties of materials
Novel products such as ebikes often include features that are innovative and manufacturer specific. Sometimes products can only be repaired by the original manufacturer. To realize a circular economy and break patterns of material consumption and reliance on hazardous and critical raw materials it is essential that products are durable and repairable. The European Research Council published repairability guidelines and indexes for many products. The Right to Repair movement captured the imagination of ebike enthusiasts. The degree to which standardization and repairability indexes should be available to consumers is currently under debate globally. Additionally, battery failure in ebikes often causes the product lifetime of a bike to end prematurely. Therefore, repairability and replaceability of ebike batteries is a key environmental action, reducing a product’s overall environmental impact.
This project examines use profile and battery degradation over time, ease of replacement of batteries in ebikes and elongation of ebike product lifetimes. Using the ISCycle fleet on the University of Limerick campus the student will apply the central approaches of repairability indices on the fleet of ebikes.
Desired qualifications: The student needs to have an appreciation of sustainability principles, circular economy and industrial ecology perspectives. The student needs to be able to apply techniques employed by the European Research Centre and UNITAR in terms of mapping product lifetimes
Relevant majors: Product Design, Environmental Science, Engineering
The R@ISE (Research at Immersive Software Engineering) Project will develop the next-generation platform for Low-Code No-Code (LC/NC) software development. LC/NC democratizes software development by empowering individuals, groups, and organizations without expert knowledge in the software development or programming to create applications that suit their needs. Education and training will represent a key component of the LC/NC platform to ensure that all stakeholders can reap the benefits of the LC/NC approach in their own contexts. This project will primarily involve research into the needs of key stakeholders and their current levels of familiarity with software development to create tailored training curricula. Students will work with the supervisory team, R@ISE researchers, partner organizations, and other stakeholder groups to identify knowledge gaps and construct engaging and accessible education and training resources and curricula to enable stakeholders to engage effectively with the LC/NC platform that is being developed by the R@ISE project.
Desired qualifications: Experience with Java and Python
Relevant majors: Data Science, Computer Science
This project will involve the development and testing of software applications that can be used to analyze and extract information from historical records. Students will build on technologies already developed for the analysis of historical mortality records and will work together with the supervisor and research team to enhance the techniques and apply them to other types of historical records. While rooted in Computer Science and Data Science, the project is interdisciplinary, exposing students to a number of different fields and faculties.
Relevant majors: Data Science, Computer Science
A wide variety of bacteria use the microbial communication system called quorum sensing (QS), which allows them to modify their behavior collectively in response to changes in cell density. This communication is mediated by small molecules accumulated during microbial multiplication and involves the production, secretion, and detection of extracellular signaling molecules, known as autoinducers.
The communication mediated by QS can be interrupted in several ways: by inhibiting the autoinducer synthesis, through enzymatic degradation of autoinducers, or by competition for binding to receptor proteins, ultimately inhibiting the target gene expression, mediated by interfering molecules called quorum sensing inhibitors (QSI)
This study aims to evaluate the ability of some artificial and natural non-nutritive sweeteners (NNS) to inhibit QS and its possible implications for the gut microbiome.
Hydrothermal carbonization (HTC) is a wet thermochemical method used to enhance the dewaterability of various feedstocks, such as biowaste, and to concentrate plant nutrients. In this process, the feedstock is heated in a sealed reactor, resulting in the formation of three distinct fractions: solid residue, liquid rich in organics, and gas. Since HTC requires an external heat source, it is essential to evaluate the energy demands dictated by process conditions such as temperature, residence time, reactor load, and stirring. Additionally, the forms and distribution of nutrients in the resulting products must be assessed. Specifically, nitrogen fractions in the products will be analyzed, including total nitrogen (N), total organic nitrogen, total inorganic nitrogen, Total Kjeldahl Nitrogen (TKN) and elemental analysis.
Relevant majors: Chemical Engineering, Chemistry
The main objective of this research topic will be to determine the energy potential of solid (hydrochars) and liquid (liquors) products of hydrothermal carbonization (HTC) obtained in high-pressure chemical reactors in a variable pH environment. For this purpose, doses of a commercial acid catalyst will be selected for digested sludge, with a constant initial moisture content, under established reactor operating conditions: temperature and residence time. In addition to conducting HTC experiments, during the training the student will learn to perform the proximate and ultimate analyses of solid residue and liquid after HTC as well as determining a higher heating value by calorimetric technique, biomethane potential test and operation and analysis by gas-liquid chromatography.
Relevant majors: Chemical Engineering, Chemistry
Leaching minerals from hydrochar is an important process in the context of biomass conversion and sustainable energy production. Hydrochar is a solid product obtained from the hydrothermal carbonization (HTC) of biomass, which can be utilized for various applications, including as a soil amendment. Hydrochar typically contains a variety of minerals, including potassium (K), phosphorus (P), calcium (Ca), magnesium (Mg), and trace elements such as iron (Fe) and manganese (Mn). The mineral content can vary significantly depending on the type of biomass used and the conditions under which hydrochar is produced. Leaching refers to the process by which soluble substances are washed out from a solid material. Leaching can release essential nutrients into the soil, making them available for plant uptake. This can enhance soil fertility and support plant growth. However, excessive leaching may lead to a loss of beneficial minerals, reducing the effectiveness of hydrochar as a fertilizer.
Relevant majors: Chemical Engineering, Chemistry
Tetracyclines are one of the most prescribed families of antibiotics used in medical, agricultural, and poultry sectors. However, they are poorly absorbed by the body (20-50%), leading to their release into wastewater through urine and feces. Zr-based metal organic framework UiO-66 has been studied for the adsorption and photocatalytic degradation of tetracycline and has been shown to be an excellent adsorbent and catalyst for the reaction. In addition, the use of modulators helps engineer its porosity and structure to enhance photocatalytic activity. However, unmodulated UiO-66 does not demonstrate very high absorption in the visible region. This study aims to study the effect of varying modulators and MOD/Zr ratios on the photocatalytic degradation of tetracycline. Unmodulated UiO-66 and modulated UiO-66 (with two ratios of seven modulators) will be prepared and characterized. Each of the 15 catalysts will be tested for adsorption and photocatalytic degradation of tetracycline under direct sunlight. An optimal catalyst will be chosen, and the pH, catalyst, and tetracycline will be varied using RSM-designed experiments to determine the optimal reaction parameters. Radical inhibitors will also be introduced into the reaction medium under optimal conditions to determine the active radicals in the degradation. Finally, modulated UiO-66 will be collected and tested for re-usability.
Relevant majors: Chemical Engineering, Chemistry
The aim of this study is to design, 3D print, and assess microfluidic chip efficacy through optimization of the microstructure of the chip, material use, and orientation of the chip on the print bed. Microfluidics chips play a pivotal role in various scientific and medical applications, facilitating precise manipulation of tiny fluid volumes for tasks such as diagnostics, drug testing, and biotechnology research. This research is built on current work whereby chip molds are printed and the chip cast. By leveraging 3D printing's precision and adaptability while incorporating biocompatible materials, the results of this study can enhance low-volume chip production while ensuring compatibility with biological systems. By merging these advancements, the research could deliver superior, cost-effective, and customisable microfluidics chips, with profound implications for healthcare, biotechnology, and scientific research.
Desired qualifications: CAD skills (SolidWorks), 3D printing experience, CNC machining experience
Relevant majors: Biomedical Engineering, Product Design
Fast and accurate diagnosis is paramount to effectively treating illness. One of the most prevalent modes of diagnosis is based on obtaining a biopsy followed by histopathological staining of the tissue to screen for disease based on tissue composition and morphology. Histopathological assessment depends strongly on the individual judgment of trained specialists, with variations between individuals and laboratories frequently reported. (10.1634/theoncologist.11-8-868). Automated methods of histopathological assessment are therefore required to reduce interpretation variability and bias, thereby improving the speed and accuracy of disease diagnosis. However, automated analysis of histopathological images is complicated by high degrees of variability across patients and protocols, necessitating the use of machine learning methods that can be trained to identify key metrics within complex sets of images. This project will develop machine learning methods to automatically analyze a cohort of histopathological images obtained from patient biological samples.
Desired qualifications: It would be helpful, but not necessary for the student to have a background in image processing and/or machine learning.
The student will have the opportunity to develop lab-based skills in histological staining of biological tissue and computer-based skills in machine learning and image analysis.
Relevant majors: Biomedical Engineering
While taking blood pressure, with the use of manual aneroid sphygmomanometer, the nurse (or a healthcare practitioner) needs to keep the diaphragm of the stethoscope directly over the brachial artery (please see picture below, from Elsevier Clinical Skills, Blood Pressure Retrieval Augmented Generation (RAG) is a popular method for enabling Large Language Models (LLMs) to utilize external knowledge sources to improve their responses to user queries in a specific domain.
The aim of this project is two-fold:
(A) Familiarize the student with RAG and its benefits and limitations.
(B) Enable the student to put their acquired knowledge of LLMs and RAG into practice by developing a RAG-based chatbot.
The student is expected to design, develop, and deploy a RAG chatbot for the Handbook of Academic Regulations and Procedures at the University of Limerick.
As part of this work, the student is expected to develop a test dataset to formally evaluate and report the performance of their chatbot in terms of factual accuracy in response to user queries.
Desired qualifications: Prior knowledge of Python programming is essential
Prior knowledge of ML/NLP would be advantageous
Relevant majors: Computer Science
In the field of materials science, dynamic molecular crystals have attracted significant attention as a novel class of energy-transducing materials. However, their development into becoming fully functional actuators remains somewhat limited. In this study the student will grow piezoelectric amino acid based crystalline materials and explore the efficiency of conversion of electrical energy to mechanical work. A simple setup will be designed to determine a set of performance indices of these eco-friendly crystals as a class of dynamic molecular crystals. The ability of these crystals to reversibly bend will be assessed from the perspective of soft robotics applications, where grippers manipulate and assemble microscopic objects driven and controlled by applied voltages.
Relevant majors: Engineering, Physics
Our research team works with organic crystals of amino acids, peptides, and proteins, to generate electricity. The popular painkiller paracetamol, also known as Acetaminophen, can be made to crystallize in the correct shape to demonstrate this phenomenon, but it needs to be combined with other molecules. In this project the student will grow multiple crystal films of paracetamol, crystallized with different solvents and coformer molecules. The films that have the highest piezoelectric response, i.e. generate the highest voltage, will be integrated by the student into a large array and tested in real-world conditions for harvesting energy from pedestrian movement.
Relevant majors: Chemistry, Physics
Our research team works with organic crystals of amino acids, peptides, and proteins, to generate electricity. Many of these crystals can generate electricity from multiple applied stimuli, such as force and heat. In this project the student will design a custom rig that can heat up or apply force to a sample while measuring its electrical output, as well as control and measure the force/heat energy applied. The student will get to test their setup on a wide variety of eco-friendly crystal-based devices for integration into our Citizen’s Assembly project to embed these devices around Ireland to generate useful energy and reduce reliance on fossil fuels.
Relevant majors: Engineering, Physics
This project focuses on the fundamental studies on resonance properties of doped hydroxyapatite, synthesized via solid state reaction. The synthesized powders will be processed in screen printed form followed by calcination and sintering. The calcined and sintered films will then be subjected to impedance analysis as a function of frequency to study resonance piezoelectricity and other dielectric properties.
Hydroxyapatite (HAp) possesses chemical resemblance to bone’s mineral content and unique properties such as biological activity and biocompatibility. Today, synthetic HAp is one of the most important materials used in hard tissue replacement applications as well as coating in cementless implants and bone graft substitute in orthopedic and dental applications. It is however, required that apart from chemical resemblance, synthetic HAp has resemblance in other physical properties for e.g. physicochemical, dielectric and piezoelectric properties. Such a material would possess the ability similar to natural bone and collagen in responding to mechanical stress to adapt growth of bone that can speed up healing where and when required but can also recede growth if or when not needed. The chemical, physical and biological properties of hydroxyapatite can be changed and tailored by suitable doping with various ions.
Relevant majors: Engineering, Physics
Brain Cancer is one of those few cancers with very high mortality and low five-year survival rate. The primary reason for this is the difficulty in diagnosing and monitoring the progression of cancer tumors noninvasively and in real time. This raises the need for a tool to diagnose the tumors in the earliest possible time frame. Raman spectroscopy is well-known for its ability to precisely represent the molecular markers available in any sample given, including biological ones, with great sensitivity and specificity. This has led to a number of studies where Raman spectroscopy has been used in brain tumors in various ways.
This research will explore the fundamentals of Raman spectroscopy and including conventional Raman, SERS, SORS, SRS, CARS, etc. that are used in cancer for diagnostics, monitoring, and even theragnostics, collating all the major works in the area. Also, the student will be introduced to nanomedicine and how Raman spectroscopy can be even more effectively be used in cancer research and the clinical level which would make them a one-stop solution for all cancer needs in the future. Further characterizations such as transmission electron microscopy and uv-vis spectroscopy will be carried out during the project.
Relevant majors: Materials Science, Biomedical Engineering, Physics, Biochemistry, Oncology
Classical data analysis involves data stored in a table, where each row represents an individual observation and each column is a numeric variable measured for that individual. While tabular data has been the focus of traditional statistical modeling for hundreds of years, in the modern world, we often encounter other data modes beyond tabular data, for example, free text, images, sound, video etc. However, machine learning methods (particularly deep learning) has been quite successful in handling multi-modal data. The aim of this project is to explore multi-modal machine learning models, assess the impact of considering the different data modalities, and compare with more traditional statistical methods.
Desired qualifications: The student should have a quantitative background, for example, in statistics, data science, or computer science. Some background in neural networks or handling unstructured data (images or text) would be helpful but is not necessary.
Relevant majors: Statistics, Data Science, Computer Science
Neural networks pervade modern society, and are now making decisions in a wide range of scenarios. Despite this, they have not historically been developed from a statistical perspective. However, in the context of statistical modeling, it is well known that one should consider inherent dependence that arises due to clusters/hierarchies of related observations, e.g., data spatially located within regions/countries, patients’ responses to a medical treatment monitored longitudinally over time, or multiple/replicate measurements taken on the same specimen. In particular, ignoring these structures leads to misleadingly precise predictions and statistical “mixed models” account for this. Therefore, the aim of this project is to examine the deterioration in the performance of neural networks on clustered data, compare them to statistical mixed models, and develop an initial mixed effects neural networks procedure.
Desired qualifications: The student should have a quantitative background, for example, in statistics, data science, or computer science. Some background in either neural networks or statistical mixed models would be helpful, but is not necessary.
Relevant majors: Statistics, Data Science, Computer Science
The following information is vetted and provided by the American Association of Collegiate Registrars and Admissions Officers (AACRAO) on the Electronic Database for Global Education (EDGE).
| Letter Grade | Percentage | Ranking | U.S. Equivalent |
| A+/A/A- | 70 - 100% | First Class Honours | A |
| B+/B/B- | 60 - 69% | Second Class Honours Upper | B+ |
| C+/C/C- | 50 - 59% | Second Class Honours Lower | B |
| D+ | 45 - 49% | Third Class Honours | C+ |
| D/D- | 40 - 44% | Pass | C |
| F | 0 - 39% | Fail | F |