http://dlib.iit.ac.lk/xmlui/handle/123456789/1362 2026-04-13T12:16:33Z 2026-04-13T12:16:33Z Dissanayake, D.D.M.S.M.B. http://dlib.iit.ac.lk/xmlui/handle/123456789/1536 2023-01-26T05:46:09Z 2022-01-01T00:00:00Z

Evaluation of brute-force techniques to improve the defense against brute-force induced attacks Dissanayake, D.D.M.S.M.B. If someone wants to securely store data indefinitely, it is nothing more than encrypting it with a key. However, this has its own limitations where an attacker could try many keys until he/she breaks the encryption directly by guessing the right key. Year by year, the processing power increases from mobile processors to quantum computers. And the brute-force attack on encrypted data is also consuming very less time in the current state; compared with the techniques, technologies and resources used in earlier years. And in the future, in the worst case, there will be a time, where the HTTPS traffic might be brute-forced in a considerable time. The real threat of this is, the brute-force attacks are completely relying on the processing power rather than the attacker’s skills. Alternative strategies must be taken into consideration in order to prevent or at least minimise being exposed to brute-force assaults as processing power grows dramatically over time, along with cloud and distributed computing. The goal of this research is to develop a brute-force prevention technique regardless of encryption standard that will at the very least guarantee that the encrypted content cannot be decrypted throughout the typical human lifetime, regardless of how powerful the CPU or GPU is.

2022-01-01T00:00:00Z Ratnayake, Pabasara http://dlib.iit.ac.lk/xmlui/handle/123456789/1525 2023-01-23T09:02:10Z 2022-01-01T00:00:00Z

NOSTRADAME : Predicting outcome of NBA games for sports betting using Machine Learning Ratnayake, Pabasara "The NBA is the most famous basketball league in the world. The league has five hundred plus players currently registered in their thirty available teams. The NBA itself is worth more than two billion dollars. The games in the NBA are televised internationally. Sportsbooks tend to keep an eye on these games because betting on NBA games has become popular among their peers. With betting on NBA games being famous, people tend to use different methods to get the upper hand when placing the bet. A score/ win predictor is the best option for this. This system is built through the use of machine learning and the analysis of previous systems and projects. The system uses individual player data and uses it to determine how the absence of a player will affect the team when up against other teams. Overall team statistics data are taken into consideration along with individual statistics, which are used to give the user a decent and accurate team score as well as an individual player score to place their bets or to know who will be the victor of the game. "

2022-01-01T00:00:00Z Abewickrama, Gayani http://dlib.iit.ac.lk/xmlui/handle/123456789/1524 2023-01-23T08:58:32Z 2022-01-01T00:00:00Z

Analyzing and Predicting the Market Price of Fruits and Vegetables Abewickrama, Gayani Is there a method or a source for consumers to obtain an idea of the pricing of fruits and vegetables before they go to the supermarket? The price increases and decreases of fruits and vegetables on the market are quite unexpected these days. Machine learning may be used to predict fruit and vegetable market prices. There are several research articles based on it. However, there is no reliable technique for predicting the market price of fruits and vegetables. In this study, attempted to help the client by predicting fruit and vegetable prices. These projections will be based on historical pricing data as well as the criteria. This report will go through the problem's history and definition. Following that, will talk about software requirement specification and the initial design. The implementation of the core function, technology selection and video demo will next be discussed. Finally, project delivery strategies and processes are established. In this research mainly aim to study the previous prediction systems and technologies to find a research gap and to implement a solution for the problem. According to the findings of previous research done on machine learning there is no system to find the predicted market price of fruits and vegetables. In this research the implementation using Arima model is explained as a solution for the problem as well as the testing and evaluation of the system also included.

2022-01-01T00:00:00Z Jesuthasan, Tony http://dlib.iit.ac.lk/xmlui/handle/123456789/1523 2023-01-23T08:26:27Z 2022-01-01T00:00:00Z

Petrichor: Semi Supervised Fine-Tuning of Textual Entailment Based Boolean Question-Answering for Pre-trained Language Models Jesuthasan, Tony Pre-trained Language Models (PLM) have taken the Natural Language Processing (NLP) domain by storm since its inception during the latter years of the previous decade. Consisting of two training stages: pre-training (unsupervised) and fine-tuning (supervised), these language models require quite a large amount of annotated data for the latter process. Procurement of such data is expensive and an immensely time-consuming process, thus hampering the use of these powerful models especially in domains where annotated corpora is scarce. Boolean Question-Answering, an NLP task, is known for being notoriously difficult in nature to solve as it relies on the textual entailment between a question and passage to infer an answer. Obtaining a labelled dataset of this sort is extremely difficult as it requires long hours of human expertise, combing through each question and its relevant passage before deducing the right answer. This hindrance also limits the use of PLMs to solve the aforementioned NLP task. These aspects present the need for a solution capable of solving the labelled data requirements of the fine-tuning process of a language model and the Boolean Question-Answering task. A promising approach that has demonstrated the ability to reduce the annotated data requirement across any medium or task is semi-supervised learning. This dissertation presents Petrichor – a hybrid architecture that pairs a PLM with a Generative Adversarial Network (GAN) for semi-supervised fine-tuning of textual entailment based Boolean Question Answering to solve this gap. Experimental results indicate that the proposed architecture is capable of producing similar performance levels (F1-scores between 70-80%) when utilizing either 100% or 10% labelled data samples of a dataset. Benchmarking results showcase Petrichor’s ability to outperform functionally similar models that suffer massive performance drops when radically reducing annotated data quantity.

2022-01-01T00:00:00Z