http://dlib.iit.ac.lk/xmlui/handle/123456789/1615 Wed, 08 Apr 2026 18:28:18 GMT 2026-04-08T18:28:18Z http://dlib.iit.ac.lk/xmlui/handle/123456789/2202 Transfer Learning for Facial Forgery Detection using Attention-based Neural Networks Randika, Aruna "This paper discusses the problem of identifying artificially generated human-like faces that appear realistic due to advancements in AI technology. The challenge lies in detecting these images using specific parameters. The author proposes to automate one of these parameters to improve the identification process." Sun, 01 Jan 2023 00:00:00 GMT http://dlib.iit.ac.lk/xmlui/handle/123456789/2202 2023-01-01T00:00:00Z http://dlib.iit.ac.lk/xmlui/handle/123456789/2201 A Secure Firmware Over-The-Air Update Framework for Constrained IoT Devices Zoysa, Navod "IoT devices have a wide range of applications in several different domains in the modern world mainly as wearables and used in smart cities to name a few. Due to their nature of being deployed in remote places it is crucial that these devices are up to date and this is done through firmware over-the-air updates but this poses a security threat where if the firmware update process is not secured it could lead to attackers using man-in-the-middle attacks or injecting malicious firmware to compromise the devices. The solution to the above problem is to patch these devices with regular firmware updates with a strong emphasis on security due to the critical nature of the situation. The solution uses encryption to transmit the firmware binaries securely between the server and the device. It also restricts unauthorized users from attacking the device or the web application by adding both token-based authentication and mutual authentication. The proposed solution was tested on an ESP32 microcontroller with 4MB storage and 320KB of memory. With a firmware binary of size close to 1MB the device updates in 1 minutes and 39 seconds with 50-60% device load during the update." Sun, 01 Jan 2023 00:00:00 GMT http://dlib.iit.ac.lk/xmlui/handle/123456789/2201 2023-01-01T00:00:00Z http://dlib.iit.ac.lk/xmlui/handle/123456789/2200 LogSense: Optimization of the DevOps Lifecycle Through Efficient Log Anomaly Detection and Root Cause Analysis Jayasuriya, Lilan "Log anomaly detection and root cause analysis play a crucial role in the optimization of the DevOps lifecycle. However, existing methods struggle to efficiently handle the large volume and diverse formats of log data, resulting in delayed detection and diagnosis of issues. This research proposes a novel machine learning-based approach for efficient log anomaly detection and root cause analysis. The model is designed to extract features from raw log data, enabling accurate anomaly detection and efficient root cause analysis. Experiments were conducted on real-world log datasets, and the proposed method achieved high accuracy and recall rates in detecting anomalies and identifying root causes. The results demonstrate the effectiveness of the proposed method in optimizing the DevOps lifecycle through efficient log anomaly detection." Sun, 01 Jan 2023 00:00:00 GMT http://dlib.iit.ac.lk/xmlui/handle/123456789/2200 2023-01-01T00:00:00Z http://dlib.iit.ac.lk/xmlui/handle/123456789/2199 Container Distribution and Optimization Gomez, Darren "Containerization technology, epitomized by Docker, has revolutionized software development and deployment by encapsulating applications and their dependencies into lightweight, portable containers. Efficient distribution and optimization of these containers are crucial for maximizing resource utilization and minimizing operational overhead. This paper explores various strategies and techniques for container distribution and optimization within Docker environments. Distribution strategies encompass methods for efficiently transferring container images across networks, including image layer caching, multi-stage builds, and content delivery networks (CDNs). Optimization techniques focus on enhancing container performance and resource utilization, such as container orchestration with Kubernetes, auto-scaling based on resource demand, and container image size reduction through layer pruning and base image selection. Furthermore, this paper investigates advanced optimization approaches, including dynamic resource allocation, workload scheduling algorithms, and utilization of specialized hardware accelerators. Real-world case studies and performance evaluations demonstrate the efficacy of these strategies in improving scalability, reliability, and cost-effectiveness of containerized applications. By implementing the proposed distribution and optimization techniques, organizations can streamline their container deployment workflows, mitigate operational complexities, and achieve higher levels of efficiency and agility in modern software delivery pipelines." Sun, 01 Jan 2023 00:00:00 GMT http://dlib.iit.ac.lk/xmlui/handle/123456789/2199 2023-01-01T00:00:00Z