Projets novateurs réalisés

ESROP – KMUTT – Time series Data Modeling and Analysis with a focus on financial data

This project focuses on improving transformer-based models for quantitative finance (QFin) by addressing key limitations such as inadequate long-term memory, lack of sentiment analysis, and difficulties in handling multimodal data. While transformers have achieved breakthroughs in fields like protein folding and natural language processing, their application in finance still faces challenges. Existing financial transformers struggle with capturing long-term dependencies, relying mainly on structured technical data while neglecting external factors like news and market sentiment.

This research aims to enhance financial transformers by integrating better memory mechanisms, sentiment analysis techniques from NLP models like BERT and T5, and multimodal learning to improve predictive accuracy and adaptability in volatile market conditions.

Automated optimization of robotic tasks and transitions using graph-based approaches

In recent years, machining with robots has become a trend in the manufacturing industry. The concept offers an economical solution for medium to low accuracy machining applications. However, due to the complexity of the robot kinematics, planning for these paths is challenging. Jabez Technologies has developed a semi-graphical approach that can program large robot-paths. This approach has been very well received by the industry and has proven to be extremely robust in practice. However, this approach is semi-automatic and cannot work without user input. Emerging applications, on the other hand, are demanding a fully automated solution. We, therefore, propose to build a customizable automated planner framework on top of the existing semi-graphical approach. The combination will allow the company to provide fully-automated solution with high flexibility and confidence.

L2M QC Spring 2025 | Tools for the automation of auditing of smart contracts for blockchain applications

The increasing adoption of blockchain technology, particularly in decentralized finance (DeFi), highlights the critical need for automated and efficient auditing tools for smart contracts and transaction monitoring. Traditional manual auditing methods are slow, labor-intensive, and insufficient to manage the growing complexity of smart contracts. By integrating AI-based techniques, this project aims to develop automated tools that can detect vulnerabilities in smart contracts in real-time, enabling faster and more reliable security audits. Additionally, these tools will provide continuous transaction monitoring, ensuring compliance and preventing fraudulent activities. The innovation lies in combining machine learning algorithms with blockchain security to scale audits, improve accuracy, and enhance the overall security of decentralized applications. Key stakeholders include blockchain developers, auditing firms, DeFi platforms, regulators, and end users, all of whom stand to benefit from more secure and efficient blockchain ecosystems.

Chelating bis(phosphinidenes) with cyclic alkyl(amino)carbene backbones: synthesis and coordination chemistry

This project focuses on developing new types of catalysts by designing and testing innovative molecules (ligands) that can better interact with metals and tune their properties. Catalysts are chemical species that help speed up chemical reactions, making them more efficient, selective, and sustainable. They are essential in many industries, from pharmaceuticals and petrochemicals to energy production, conversion, and storage. The proposed work will explore a promising new class of ligands called cyclic alkyl(amino)carbene-stabilized phosphinidenes, which build on the strengths of existing phosphinidene ligand systems to create ligands imparting improved stability and reactivity to the metal centers they bind to. This collaborative work will bring together the combined expertise of the two participating research groups in stable carbene and phosphorus chemistry, respectively.
By fine-tuning the structure of the phosphinidene ligands, the goal is to create metal complexes that can perform challenging chemical transformations more efficiently. This could lead to breakthroughs in green chemistry, helping reduce waste and energy consumption in industrial processes. It also provides valuable training for the training of highly qualified personnel in cutting-edge chemical techniques, fostering international collaboration in the field of advanced materials and catalysis.

Aerial eco-phenotyping: Characterization of tree fitness in response to water use

The main objective of this internship is the acquisition of field data in Australia, using drones equipped with RGB and thermal sensors to map and monitor vegetation in mine sites undergoing restoration. These data will be used to feed an AI model to identify plant species and assess their performance and resilience to environmental stresses. This project is part of an international research collaboration between the Université de Montréal (UdeM), the University of Western Australia (UWA) and the Kings Park Science Program, a renowned research center in plant ecology and ecosystem restoration. This internship is part of a wider collaborative project aiming to develop methods for monitoring and evaluating ecological restoration initiatives, particularly at mining sites, which will be relevant to all three organizations contributing to this project.

The Diagnostic Potential of Co-rotating Interaction Regions in Dense Hot-Star Winds

Wolf-Rayet (WR) stars exhibit strong, high-velocity winds with small-scale stochastic clumps and large-scale structures like Co-rotating Interaction Regions (CIRs). CIRs are spiral-shaped density enhancements, which propagate through the wind and induce variability in spectropolarimetric signals. WR6 (EZ CMa, WN4b) is a well-studied target with a stable 3.76-day periodicity in its wind variations. Recent observations, including linear spectropolarimetry, reveal distinct patterns in the Stokes Q-U plane, suggesting a complex, asymmetric scattering envelope.

Current modeling efforts, such as those by Dr. Richard Ignace, provide proof-of-concept frameworks for interpreting spectropolarimetric data. However, these models require refinement and validation through deeper observational insights. The proposed research seeks to advance the synergy between observations and modeling to uncover the mechanisms behind these variations, a critical step in understanding massive-star evolution and feedback.

Projet de recherche – Occupation temporaire d’espaces urbains vides à Montréal. Augmentation des campements urbains et de l’urbanisme temporaire en situation de crise du logement abordable.

Alors que la crise du logement atteint des niveaux critiques à Montréal, la gestion des espaces vacants soulève d’importants enjeux sociaux et urbains. Récemment, divers types d’occupations temporaires émergent : certaines, portées par des institutions ou des initiatives citoyennes, visent à revitaliser l’espace public à travers des projets culturels, sociaux ou économiques, tandis que d’autres, comme les campements urbains, résultent de la précarisation croissante et sont souvent perçues comme problématiques par les autorités. Ces occupations, bien que différentes, révèlent des tensions autour de la légitimité de l’appropriation de l’espace urbain et des politiques qui en régulent l’usage.
Ce projet de recherche analyse les représentations, discours et pratiques entourant ces occupations, en mettant en lumière les tensions, rapports de pouvoir et mécanismes de légitimation qui influencent leur reconnaissance ou répression. À travers une enquête ethnographique auprès des usagers, collectifs citoyens et acteurs associatifs, il donnera la parole aux personnes concernées pour mieux comprendre leurs expériences et stratégies face aux cadres institutionnels. En approfondissant ces enjeux, cette recherche contribuera à une réflexion plus large sur les politiques urbaines et la gestion de l’espace, afin d’intégrer la diversité des usages et besoins dans un contexte de transformation accélérée.

Non-invasive assessment of early molecular response and detection of minimal residual disease in diffuse large B-cell lymphoma (DLBCL) using cell-free methylomes

Research at the Lymphoma Lab is driven by the observation that the outcomes of lymphoma are extremely variable with some patients seem cured or enjoying long remissions while others experience relapse. Treatments that directly target the underlying therapeutic vulnerabilities of a patient’s lymphoma remain unavailable [1]. Lymphoma Lab wants to change this. A comprehensive understanding of the biological underpinnings of lymphoma can lead to improved therapies for patients. One promising approach is detecting mutations in circulating tumor DNA (ctDNA) to assess disease status non-invasively [2]. However, its effectiveness is limited by the small number of mutations detectable in targeted sequencing panels, reducing its sensitivity. Enzymatic Methyl-seq (EM-seq), a novel technique, enables the non-invasive profiling of tumor methylomes. Unlike mutation-based methods, EM-seq can analyze thousands of alterations without being restricted by predefined sequencing panels [3]. They hypothesize that methylomes from EM-seq will allow identification of lymphoma specific methylation signatures that can be quantified and tracked over time to determine treatment response. Lymphoma Lab aims to clarify the mechanisms of lymphoma pathogenesis, tumor evolution, and treatment resistance. The lab’s overarching goal is to enhance patient outcomes by deepening the understanding of treatment response variability and advancing personalized therapeutic approaches for lymphoma. More information on the lab can be found at http://kridel-lab.ca/Research/.

Improved Order Computation in Class Groups of Real Quadratic Fields

Cryptography is an important tool for safeguarding our data from attackers. The security of several modern cryptosystems relies on unproven properties of an algebraic structure called the class group of an algebraic number field. In the absence of proofs, tabulating class groups in order to generate numerical evidence of these unproven properties remains the best way to enhance our confidence in their truth and the security of the related cryptosystems. However, tabulating class groups in all but the simplest types of number fields remains a significant computational challenge. This project will devise improved algorithms for computing the order of an element in the class group of a real quadratic field, the simplest case of number fields where these challenges manifest. Order computation can be considered as a special case of computing the full class group. The results will be a significant step to improving algorithms for class group computation, eventually leading to the extended class group tabulations required to bolster our confidence in security claims of related cryptosystems.

ESROP – KMUTT – Context Aware AI Music Therapist for Elderly with Neurodegenerative Diseases

Neurodegenerative diseases like Alzheimer’s and Parkinson’s cause cognitive decline, motor impairment, and emotional distress. While pharmacological treatments exist, music therapy has shown promise in alleviating symptoms. This project enhances music therapy using AI, making interventions more accessible and personalized. This project aim to develop an AI-powered wearable system that personalizes music therapy based on emotional state and physiological responses. Using machine learning and biometric monitoring, the system will analyze user data to curate and deliver tailored music therapy in real-time.

Passive Temperature-Responsive Solar Energy Switch for Heat and Electricity Generation

Harnessing Solar Power for Smarter Energy Use:
This project introduces a passive solar energy switch that automatically adjusts how it uses sunlight based on temperature, without electricity or moving parts. It relies on a Fresnel lens that focuses sunlight. When temperatures are low, water condenses on the lens, causing solar transmission and heating. When temperatures rise, the water evaporates, restoring the lens’s ability to focus sunlight for efficient solar electricity generation.

Why It Matters:
Heating and cooling account for over half of global electricity use. Traditional systems consume large amounts of energy, leading to high costs and environmental impact. This technology offers a self-regulating, cost-effective, and scalable alternative for energy-efficient buildings and solar power systems.

Real-World Applications:

1. Smart Windows that adapt to sunlight, reducing energy use.

2. Solar Concentrators that enhance electricity generation.

3. Off-Grid Energy Solutions that work in varying conditions.

By integrating natural processes like condensation into solar energy management, this innovation enhances sustainability and efficiency in everyday applications.

L2M – Development of Self-Powered Portable Devices for Simultaneous Water Disinfection and Pollutant Removal Through Piezophotocatalysis

This project aims to provide clean drinking water to low-income families in remote areas by developing a self-powered, portable ceramic water filter. While the filter has shown strong pollutant removal in lab tests, real-world testing is needed to confirm its effectiveness. This project will focus on pilot testing, refining the product based on user feedback, and developing a strong market strategy. By partnering with organizations, we can accelerate commercialization, secure funding, and ensure successful market entry. The project will benefit partners by creating opportunities in the water filtration industry while improving public health through affordable and sustainable clean water solutions.