ArgusEye: A Federated Learning Based Intrusion Detection System with a Focus on Privacy Preservation
“ArgusEye is a privacy-preserving Intrusion Detection System (IDS) designed to detect network anomalies and attacks without compromising user data. Traditional IDS solutions typically rely on centralized machine learning models that require the collection of raw data from client nodes. While effective, this centralized approach raises serious privacy concerns, increases the risk of data breaches, and potentially violates compliance requirements in sensitive domains such as healthcare and finance.
ArgusEye addresses these limitations through the integration of Federated Learning (FL) and Differential Privacy (DP). Federated Learning enables decentralized model training by allowing each client device to train on its local data, sharing only model updates, and not the data itself, with a central server. To enhance privacy further, Differential Privacy is applied by adding noise to the aggregated updates, making it difficult to infer any single user’s contribution.
The system was developed and evaluated using a widely used benchmark dataset and was designed to simulate multiple clients operating in a federated learning environment. Experiments were conducted to study the impact of differential privacy on model performance and training time. ArgusEye achieved strong detection metrics, which demonstrate that federated learning, when combined with privacy-preserving technologies, can offer a practical and effective solution for intrusion detection in privacy-sensitive environments.”
VRoxel: An adaptive context aware WFC algorithm with streaming Marching Cubes
“VRoxel is an innovative procedural generation system that addresses critical challenges in creating large-scale 3D environments for VR applications. The system tackles three fundamental problems: boundary coherence issues that create visible seams between independently generated chunks, the inability of current algorithms to operate at multiple scales simultaneously, and the performance-quality trade-off that forces developers to sacrifice either visual fidelity or the 72+ FPS required for immersive VR experiences.
The system introduces novel boundary-aware Wave Function Collapse (WFC) algorithms integrated with a hierarchical constraint system and streaming Marching Cubes mesh generation. The boundary-aware WFC extends traditional constraint propagation with explicit cross-chunk constraint sharing through innovative boundary buffers, eliminating the notorious “”stitching problem.”” The hierarchical constraint system operates simultaneously at global, regional, and local scales, enabling generation of environments with both macro-level structures like mountain ranges and micro-level details.
VRoxel’s streaming Marching Cubes implementation converts voxel data to optimized meshes with predictive chunk loading based on user movement patterns and adaptive Level-of-Detail (LOD) systems. The two-phase implementation—2D Python prototype for algorithm validation followed by full 3D Unity implementation with parallel processing—demonstrates effective solutions for maintaining structural coherence while achieving real-time performance requirements.”
TremorAid. An Intelligent Parkinson’s Companion: Tracking, Predicting and Personalizing care
“Parkinson’s disease (PD) is a progressive neurodegenerative disorder affecting over 10 million people globally, including approximately 92,000 in Sri Lanka. Characterised by symptoms such as tremors, rigidity, and bradykinesia, PD requires early detection and continuous monitoring to enhance patient outcomes. Traditional symptom-tracking methods like paper diaries or generic digital tools often lack consistency, accessibility, and suitability for those with motor impairments.
TremorAid is a mobile-based application designed to bridge these gaps by enabling PD patients to log symptoms, track tremor progression, and identify environmental triggers. With an intuitive interface tailored to users with motor difficulties, TremorAid promotes autonomy through seamless symptom logging, access to educational resources, and meaningful, data-driven insights.
Leveraging artificial intelligence, the app classifies tremors, visualises symptom trends, and provides personalised symptom trajectories for proactive disease management. A key feature is its future symptom analysis module, which forecasts progression based on historical patient data, empowering users and healthcare professionals to make informed decisions about care and treatment adjustments.
While TremorAid does not support direct communication with healthcare providers, it enables data sharing via visual dashboards that support clinical evaluations. Future iterations may include wearable device integration and improved predictive models for enhanced monitoring and insight.
TremorAid exemplifies how digital health innovations can address the challenges of chronic disease management. By combining AI-driven insights, accessibility, and a patient-centric approach, it contributes meaningfully to advancing personalised care in Parkinson’s disease.”
Predicting the efficiency of apparel production using ML model
“This project focuses on predicting the efficiency of apparel production lines using machine
learning, with the goal of classifying lines into Efficient, Low Efficient, or Inefficient
categories. Efficient production is critical in the apparel industry, yet identifying
underperforming lines in real time remains a challenge due to the complexity of operational
factors like SMV,lead times, and production hours.
The project adopts a data-driven approach using a historical production dataset. After
preprocessing and exploratory analysis, key features were selected and engineered to enhance model performance. PyCaret, a python based machine learning library, was used to train and compare several classification algorithms, including Logistic Regression, Random Forest, LightGBM, and XGBoost. Based on accuracy, F1-score, and AUC, the XGBoost classifier was chosen as the final model due to its superior performance and robustness.
Hyperparameter tuning was applied to optimize the model, which achieved a final accuracy of 79% and a macro F1-score of 0.84.To ensure generalizability, 10-fold stratified cross-validation was used, and the model was tested on unseen data.
A Flask web application was developed to make the model accessible to non-technical users. This allows production managers to input planning data and receive immediate efficiency classifications, supporting proactive decision-making.
Overall, the project demonstrates how machine learning can enhance operational efficiency in apparel manufacturing. Though effective, the system could be improved by incorporating real- time ERP integration and expanding the dataset for broader applicability.”
Work Sphere (Elevating Hybrid Employee’s Engagement in Online Working through Multi-model Involvement Recognition with Explainable AI)
“The COVID-19 epidemic changed IT industry processes by hastening the introduction of remote work. In virtual workplaces, detecting employee engagement has become essential. Conventional monitoring ignores behavioral measures like mouse activity and screen duration, as well as subtle engagement signs like head movements and facial expressions. Leadership is unable to comprehend employee difficulties and offer prompt assistance because of this detection gap. Resolving this issue will help organizations develop healthier cultures by empowering leaders to address engagement issues, build relationships, and create supportive settings even when people are physically separated.
In order to solve this issue, we created a multi-model engagement detection system that uses facial expressions, eye gazing, and head posture as crucial indicators. In order to identify two emotion kinds using face ROIs which are crucial for engagement analysis we developed innovative CNN architectures. Class imbalance in the complicated emotion dataset was handled by a parallel model that concentrated on basic emotions, offering a thorough engagement detection method. The most important facial ROIs indicating engagement were found using XAI algorithms, which also confirmed predictions.
Accuracy, precision, recall, and F1-score criteria were used to assess the CNN model in a variety of employee engagement scenarios. The results showed that both basic and complicated emotion prediction models were able to identify patterns of workplace engagement with 61.5% accuracy. Real involvement levels in remote work environments were successfully predicted by combining mouse and screen time monitoring with eye gaze and face angle estimation.
“
BlurFix : Efficient Image Blind Motion Deblurring Using Generative Adversarial Network
Most real-world images are degraded by motion blur due to either camera shake or fast moving objects. This prevails as a great challenge for many applications in computer vision such as object detection and scene understanding, which really require clear images. Blind motion deblurring aims to restore sharp images from blurry ones without any prior knowledge of the blur kernel. Real and effective deblurring remains a big challenge due to detail retention and avoiding artifacts. Even though the recent advances in deep learning and, particularly, Generative Adversarial Networks have shown some promising work for this task, they utilize a high computational cost. This work presents a Mobile Vision Transformer (MobileViT) enhanced with gradient Wasserstein GAN penalty (WGAN-GP) efficient blind motion deblurring framework pairing a spectrally normalised discriminator for stable training with a U-Net style generator with MobileViT blocks for multi-scale feature fusion and global-local representation learning. The proposed architecture combines important developments including a hybrid CNN-Transformer generator using inverted residual blocks and skip connections to preserve high-frequency details, a perceptual loss formulation including VGG-16 features alongside adversarial and L1 losses (Mean Absolute error) and lightweight patch-based discrimination using spectral normalising for enhanced training stability. Using the GOPRO dataset, experiments show that our model maintains a lower computational complexity than equivalent GAN-based deblurring techniques while obtaining competitive PSNR/SSIM measurements.
ChoonPaan OptimAI – Optimising Operations and Improving Profitability in Sri Lanka’s Mobile Bakeries
“Sri Lanka’s mobile bakery sector is an essential part of the informal economy, however, it faces growing challenges in maintaining operational efficiency and long-term profitability. These businesses often rely on outdated manual processes and intuition-based decision- making, leading to inefficiencies that hinder sustainability and competitiveness in a rapidly evolving market. To address these concerns, the project ChoonPaan OptimAI was initiated to explore how digital transformation can enhance business outcomes for mobile bakery vendors in this niche industry.
The project began with extensive research, including stakeholder interviews and surveys, to uncover the operational pain points and technology gaps within the mobile bakery ecosystem. The system design was driven by these insights, leading to the development of a prototype platform that integrates AI-powered features and a user-friendly interface tailored to vendor needs. The primary objective was to explore how data-informed route planning and inventory suggestions could improve decision-making and overall profitability.
ChoonPaan OptimAI offers a holistic solution by embedding intelligence into the daily operations of mobile vendors, replacing inefficiency with adaptability and insight. Rather than simply digitising existing routines, the solution envisions how vendors interact with their business data, transforming informal decision-making into a structured, insight-driven process. It introduces a framework where operational patterns, environmental factors, and past business performance are synthesised to support more responsive and resilient choices. This transition towards proactive management equips vendors to better navigate uncertainties, optimise limited resources, and incrementally build towards long-term sustainability. By promoting smarter resource use and sustainable practices, this project not only supports business resilience but also reflects a broader shift towards digital empowerment in micro-enterprise sectors.
Keywords: Mobile Bakeries, Artificial Intelligence, Sustainability, Route Optimisation, Inventory Forecasting, Micro-enterprises, Digital Transformation, Profitability”
ZKSafe: Enhancing Crypto Wallet Usability and Security Through Zero-Knowledge Proof-Based Authentication
Security and cryptocurrency wallet use are still at the center of the issues with blockchain adoption. Seed phrase-based physical wallets and private key management can result in loss, theft, and user mistake and provide entry points for mainstream use. Centralized key storage is convenient but exposes access vulnerabilities to breaches and unauthorized viewing. This study suggests a non-custodial crypto wallet with Zero-Knowledge Succinct Non-Interactive Argument of Knowledge (zk-SNARK) verification for improved security and convenience. The system offers ownership proof of the wallet without the disclosure of the private key, which maintains confidentiality and avoids utilization of the classic seed phrase recovery. The wallet uses modular design with backend features written in Node.js and Express.js, frontend in Angular, and MongoDB for safe storage of users’ data. The wallet transactions are processed by ethers.js, directly on the Ethereum blockchain. The private keys are encrypted by AES too, thus reducing exposure risk. The architecture is client-server with zk-SNARK proofs created and checked to verify the users and grant them access to their wallets. Future work will move proof creation to the client side so that private keys are never exported from the user device, giving additional security. The system was tested using functional testing, confirming key operations like wallet generation, verification, transaction handling, and secure key storage. The system was validated structured functional, integration and system level tests. The test scenarios included the core functionality of the wallet with the ZKP integration. Edge case testing was also conducted to ensure robustness against invalid inputs and unauthorized access attempts. All core features passed the validation tests under expected and abnormal scenarios. Although, performance and usability metrics were not captured due to lack of deployment the successful integration of wallet operation and ZKP shows the feasibility of the project.
Advancing Resume Grading Systems with NLP and Explainable AI
“Traditional resume analysing systems rely heavily on keyword matching and rigid formats, which can lead to qualified candidates being overlooked if they don’t conform to the system’s format. Additionally, these systems often function as “black boxes,” providing a final score without any transparency in the decision-making process. This lack of explainability can lead to biases in candidate evaluations, raising concerns about fairness and accuracy. To address these issues, this project introduces a more transparent, section-based resume scoring system.
A hybrid model which integrates Named Entity Recognition, transformer neural networks, and a point-based scoring system was developed. It provides granular evaluations of each resume section. Using Explainable AI (XAI) techniques, the system provides insight into how each section’s score is calculated, promoting trust and accountability. The approach also allows recruiters to adjust evaluation criteria, enhancing flexibility.
The initial Named Entity Recognition (NER) model showed significant results, with an F1 score of 86.84%, a recall of 88.24%, and a precision of 85.48%. Similarly, the transformer regression model demonstrated strong performance, with an R-squared value of 0.896428, a Mean Squared Error (MSE) of 0.194837 and a Mean Absolute Error (MAE) of 0.312830. These results indicate the effectiveness of the models and their potential for further improvement.”
TraceFlow: Enhancing Blockchain-based Traceability in Sri Lankan Fresh Mushrooms for Export and International Trade
“Abstract
The fresh agricultural produce industry faces significant limitations with its reliance on manual, outdated documentation-based traceability practices. This is no exception for the fresh mushroom industry. These limitations are further exaggerated in the presence of international markets, thereby limiting the export potential of local fresh mushrooms due to the lack of information accuracy and transparency. This project aimed to develop “TraceFlow”, a solution tailored explicitly for digital traceability of fresh mushrooms in the Sri Lankan small-holder farmer landscape.
A structured Hybrid agile project management approach was adopted throughout the development of this project to facilitate incremental development, stakeholder engagement, effective risk mitigation and rigorous testing. This was initiated from problem identification and industry research, moving to stakeholder requirement elicitation and gap analysis of existing practices/ solutions. The system architecture integrated a react-based web interface for admins and a flutter-based multilingual mobile application for daily harvest logging, verification, defect tracking and blockchain centric data immutability using solidity smart contracts managed via web3.js.
The key outcomes demonstrated a significant decrease in supply chain error and absence of information misplacement, strengthening the preparedness for European Union regulations and regulatory compliance. While successful, the study identified potential areas for improvements including IoT integrations, off-line data logging, smart analytics, suggesting valuable opportunities for future growth of TraceFlow.
Keywords: Traceability, Blockchain, Fresh mushroom, Supply chain, Export regulations
“