Lakehead University Knowledge Commons

Knowledge Commons is an open access repository for scholarship and research produced at Lakehead University. It is a free and secure repository for LU faculty, students, staff, and researchers to preserve and present their scholarship.

Communities in KnowledgeCommons

Select a community to browse its collections.

Now showing 1 - 3 of 3

Recent Submissions

A leakage-aware and reliability-focused evaluation framework for multiclass Alzheimer’s Disease staging using MRI slices
(2026) Boyle, Tanner; Bajwa, Garima; Akilan, Thangarajah; Alkhateeb, Abedalrhman; Cava, Dominique
Two-dimensional magnetic resonance imaging (MRI) studies of Alzheimer’s disease (AD) can overestimate performance when slices from the same subject appear in both training and assessment sets. This thesis evaluated a leakage-aware framework for multiclass AD staging based on subject-independent partitioning, subject-level prediction, calibration, and external validation. Three model families were compared on 347 OASIS subjects using subject-wise crossvalidation: a convolutional neural network (CNN), a Vision Transformer (ViT), and a Hybrid CNN–ViT architecture. A matched slice-random ablation quantified leakageassociated inflation. Supporting analyses examined imbalance-aware objectives and subject-level reliability. Generalizability was tested on 502 ADNI subjects after OASIS model-selection decisions were fixed. Slice-random evaluation contaminated 98.3% of assessment subjects and inflated macro- F1 by 0.252–0.288. Under subject-wise evaluation, the Hybrid achieved the highest OASIS macro-F1 (0.711). On ADNI, the ViT achieved the highest external macro-F1 (0.718), reversing the internal ranking. These findings show that evaluation protocol and independent cohort testing can materially change conclusions about model quality. Keywords: Alzheimer’s disease, ADNI, data leakage, MRI, subject-wise cross-
Three-phase transformerless series-connected current source inverter with capacitor voltage balancing control method
(2026) Wu, Kechen; Wei, Qiang; Zhou, Yushi; Dekka, Apparao
In medium-voltage and high-power applications, the series-connected current source converter (SC-CSC) is a good candidate. However, most existing series-connected CSCs require transformers. Recently, a single-phase transformerless series-connected current source converter has been proposed. It eliminates the bulky and costly transformer, achieving lower cost and smaller system size. In this work, the three-phase version of this inverter is proposed and investigated. Since the original single-phase topology cannot be directly extended to the three-phase system, the original topology is modified to eliminate circulating currents. In practical implementations, mismatches in capacitors can destroy the inherent voltage balance of the inverter. To address this issue, a voltage balancing control strategy is introduced to regulate the capacitor voltages under parameter mismatch conditions. The proposed control method restores the capacitor voltage balance without modifying the topology or affecting its current balancing characteristics. The operation principles of the proposed three-phase inverter and the effectiveness of its capacitor voltage balancing strategy are validated through MATLAB/Simulink simulation results.
Evolutionary neural architecture search for automatic chord estimation: a reproducible, vocabulary-aware study of structured harmonic sequence models
(2026) Frost, Russell; Akilan, Thangarajah; Yassine, Abdulsalam; Carastathis, Aris; Zhou, Yushi
Automatic chord estimation (ACE) takes a musical recording and produces a time-aligned sequence of labels. These labels, such as C major, A minor 7, or G7/B, provide a compact description of a song’s harmony and support transcription, teaching, accompaniment, harmonic search, and large-scale musical analysis. Automating this process is difficult because the same harmony can be voiced in different ways, chords often share pitches, chord labels are contextual to surrounding harmonic information, detailed chord types may be rare, and human annotators may disagree. This thesis presents a reproducible evolutionary neural architecture search (NAS) framework for large-vocabulary ACE. Starting from a validated convolutional–recurrent baseline, the framework searches for improved convolutional and recurrent architectures while keeping the musical representation, chord vocabulary, training procedure, and decoder fixed. Multiple independent searches are conducted using a validation objective that balances recognition performance with model complexity, after which selected architectures are independently retrained and evaluated. On the McGill Billboard benchmark test set, the proposed model achieved a Weighted Chord- Symbol Recall (WCSR) of 63 49 0 36%, compared with 60 09 0 16% for the baseline, with a mean paired improvement of 3 39 0 42 percentage points across three initialization seeds. After topology freeze, the same model was retrained from scratch on a genre-stratified, song-disjoint split of Chordonomicon, with the corresponding chord progressions rendered as synthetic audio; it achieves 96 50 0 12% WCSR compared with 95 26 0 14% for the baseline, an improvement of 1 23 0 23 percentage points.
Leakage-aware sentiment and multimodal stock-index analysis
(2026) Hu, Junjun; Yassine, Abdulsalam; Alkhateeb, Abedalrhman; Deng, Yong
Short-horizon stock-index forecasting remains difficult because returns are noisy, non-stationary, and only partly explained by historical prices and public news. This thesis investigates whether financial-news sentiment and multimodal market representations improve three-class direction classification, while asking when high accuracy reflects genuine forecasting rather than target construction. To address these questions, it develops a staged, leakage-aware experimental framework. A finance-domain language model first transforms dated headlines into lagged daily sentiment features. The primary forecasting experiment combines these features with historical market variables to classify the genuine three-trading-day log return from 𝑡 to 𝑡 + 3, using fixed ±0.5% thresholds. Candidate look-back windows of 5, 10, 15, and 20 trading days are compared using validation macro-F1 only; the 10-day window is then evaluated once on the locked test. The study also evaluates a smoother EMA-50 trend-state task and reconstructs a published ViT–TFT–HOG design that fuses numerical sequences, candlestick images, visual descriptors, and candle geometry across multiple stock indices. Chronological splits, simple baselines, feature ablations, per-market diagnostics, and matched target controls are used throughout. On 232 locked test examples, the Temporal Transformer reaches 37.93% accuracy, 0.3788 macro-F1, and 0.0393 MCC. The majority baseline has higher accuracy at 41.38% because of class imbalance, but its macro-F1 is only 0.1951. A matched ablation reaches 0.3808 macro-F1 without sentiment, so prior-day sentiment does not provide measurable incremental forecasting value in this experiment. Trading outcomes are mixed across markets and do not establish consistent economic value. The EMA-50 task is substantially easier because its labels are smoother and more persistent. The reconstructed multimodal model achieves stable accuracy above 93% for the reported market-state task and transfers well to additional indices. However, a deterministic audit reveals that the prices defining this target are already available in the input window; when the endpoint is moved into the genuine future, performance falls close to chance. These findings show that high classification accuracy can measure state reconstruction rather than forecasting skill. The main contribution is a reproducible, target-explicit evaluation that preserves strong reconstruction results without overstating future-return predictability.
The Listening, The Stillness, The Call, The Echo: re-Storying educational systems to address anti-Indigenous racism in education
(2026) Mathias, Amanda; Korteweg, Lisa; Peltier, Sharla Mskokii
Lakehead University’s MEd portfolio can be a collection of creative, scholarly, and technical products created for professional use, allowing the development of resources that connect theory to practice in educational work. This portfolio includes an introduction/overview and four Chapters. The four chapters are the literature review, a resource that benefits education (a policy guideline), a relational response (Human Rights), and a reflection. This portfolio is organized through the Maang/Loon teaching as four interconnected movements: The Listening, The Stillness, The Call, and The Echo. These movements structure the portfolio chapters: literature review, policy/guideline reflection, relational response guide, and critical reflection. Together, they support a re-storying approach to anti-Indigenous racism grounded in Miigiwewin, Miingweziwin, Anishinaabewin, Odenang Inquiry, and relational accountability. In this portfolio, the Maang/Loon framework guides how the work is approached, while re-storying names the process of challenging anti-Indigenous racism in northern Ontario provincial schools. Within this organizational framework, the research question and subquestions ask how a re-storying framework can support educational systems while amplifying Indigenous student, family, and community voices. Through Listening, Stillness, Call, and Echo, this framework offers a cyclical and relational way to move from deep listening toward accountable future action. Chapter One - Literature Review - The Listening This chapter reflects the framework's Listening phase. Wisdom, knowledge, and understanding are sought to uproot, seed, and nurture social justice in education. The literature review is rooted in Indigenous Storywork, wholistic theory, Miigiwewin, Miingweziwin, Anishinaabewin, Odenang Inquiry, anti-Indigenous racism, decolonizing education, Two-Eyed Seeing, and relational accountability. It gathers the knowledge, teachings, and scholarship needed to understand the roots of the problem and the possibilities for transformation towards social justice in education. The literature review synthesizes academic sources, identifies major themes, tensions, and contributions across the literature, including the limits of inclusion, the systemic nature of anti-Indigenous racism, the importance of Indigenous-led pedagogy, and the ethical responsibilities of re-storying education. The Listening reflects the Maang teaching that meaningful leadership begins with a deep examination beneath the surface. In education, this means reviewing and reflecting on policy statements to examine the underlying structures, assumptions, histories, and stories that shape school systems. Before change can occur, the “water” of the institution—its norms, biases, silences, and power relations—must be understood through relational listening and critical reflection. In this chapter focused on academia and Indigenous scholarship, listening also amplifies Indigenous voice and wisdom by amplifying and/or “listening” to Indigenous academic scholars by citing their important, authentic, and meaningful work. These scholars are trailblazers and knowledge keepers. Chapter 2 - Re-Storying for Policy/Guidelines - The Stillness The second chapter is an applied practitioner inquiry focused on selected policy and guideline areas connected to anti-Indigenous racism and Indigenous student well-being. This task corresponds primarily to The Stillness. This chapter focuses on a particular set of public and approved documents to develop a focused reflection on omissions, misrepresentations, tensions, or areas requiring further development. This has led to specific recommendations for the formal creation of an Indigenous education policy and guidelines that address the following: Collaboration/Co-creation/Communication requirements, Commitments of Northern Ontario School Boards, Self-Identification, Land Acknowledgments, and Data & Data Sovereignty. The policy is a starting point, as directly related to student success and well-being. This concise analysis and set of considerations will inform future policy work in Indigenous education. The Stillness represents an intentional pause for reflection and critical inquiry. In this phase, institutions examine policies, practices, and assumptions through a relational lens to ensure that change is grounded in understanding rather than urgency alone. In this case, an academic publication, the “Stillness,” resists the colonial pressure to solve problems, rush toward solutions and action before real relationships, understandings, and true co-creation of a shared path can exist, with a two-eyed seeing philosophy. Chapter 3 - The Call - Re-Storying Relational Response Guide The third chapter is an applied framework product: a re-storying relational response guide for educators and/or administrators. Drawing on the literature review, policy reflection and recommendations, the guide will invite educators into the Maang/Loon re-storying framework and offer literature-informed principles, reflective prompts, and relational considerations for addressing anti-Indigenous racism through relational accountability. This chapter will share how one aspect of the policy (from chapter 2), specifically, uprooting Indigenous racism through the Ontario Human Rights Code (OHRC) honours Anishinabek rights to cultural absences without penalty. The chapter (guide) will support educators and administrators in supporting anti-Indigenous racism efforts by situating themselves in the work, listening, developing a better understanding, building relationships, and being active in re-storying. Honouring experiences, accreditation and properly recording approved cultural absences may produce a more accurate and respectful account of student participation, but the purpose of this chapter is not to measure improvement in attendance statistics. The Call represents a shift from reflection to action grounded in relational accountability. The Call, in this academic use, involves educator learning, policy reflection, Indigenous-led pedagogy, and practical tools that support systemic change. This chapter is intended to inform future educator learning and administrator reflection in the broader 2026–2027 funded project. Within the MEd portfolio, however, its purpose is conceptual and preparatory: to translate the literature and Maang/Loon framework into a practical but provisional resource for future community-led implementation and revision. Chapter 4 - Reflection - The Echo The Echo represents community accountability. In the Maang teaching, the voice is never projected in isolation. Maang pauses, listens, and waits for the echo before singing the next note. In educational systems, the Echo is the commitment to respect, responsibility, reciprocity, relationship, and reconciliation. The Echo becomes the framework's accountability mechanism. It reminds educational systems that the next action must be shaped by what returns from students, families, communities, Elders, Knowledge Keepers, and the Land itself. In relationship to this portfolio, the ECHO will be my personal reflection as Mother, Daughter, Aunty, Artist and Educator as a member of Teme-Augama Anishinabek. The portfolio's four interrelated chapters meet the expectations of the MEd portfolio by combining scholarly, applied, technical, and reflective components. Aspects of this portfolio may inform meaningful future change through the broader 2026–2027 community-led project.