Browsing by Author "Jayawardena, C."

Now showing 1 - 6 of 6

Embargo
Adding Commonsense to Robotic Application Using Ontology-Based Model Retraining
(IEEE, 2022-10-04) Pradeepani, M. K. T.; Jayawardena, C.; Rajapaksha, U. U. S.
In terms of the level of technological capability in the world today, the use of automated robotics is common in various fields. There are large projects going on in many industries that collaborate between robots and other robots, as well as humans and robots. In hospital environments, care for people with medical needs and their needs and used to make appropriate suggestions to their problems. Robots can also be found in certain areas that can respond quickly as an emergency rescue agent. Furthermore, robots, which can be seen in the hotel industry as waiters and as farm assistants in agriculture, have a great tendency to be used as multi-tasking agents in many fields. In each of these areas, robots must co-operate with humans. In that situation, the importance of the exchange of mutual knowledge between robots-robots and between humans-robots comes into the picture. What matters here is not only the quantitative vastness of knowledge but also the ability to understand each other in the same medium. Although the common sense that people need in their day-to-day work is completely obvious to humans, the commonsense knowledge domain needs to be implanted in robots. Whatever concept is defined for adding commonsense to robotics, it should be a consistent concept that can be logically constructed so that it can be understood by a machine. As will be discussed later in the paper, different methods have been used in various related works to add a different kind of domain knowledge to robotics. The objective of this paper is to provide an improved retrained model for robotics in order to give them the ability to act more human-like when performing tasks. By using the proposed model robots are able to answer the incomplete command or inquiries related to a given context. One of the objectives of this work is to use the ontology-based, commonsense-support existing knowledge base as a mechanism to retrain and build a new model.
Embargo
Comparative analysis of the application of Deep Learning techniques for Forex Rate prediction
(2019 1st International Conference on Advancements in Computing (ICAC), SLIIT, 2019-12-05) Aryal, S.; Nadarajah, D.; Kasthurirathna, D.; Rupasinghe, L.; Jayawardena, C.
Forecasting the financial time series is an extensive field of study. Even though the econometric models, traditional machine learning models, artificial neural networks and deep learning models have been used to predict the financial time series, deep learning models have been recently employed to do predictions of financial time series. In this paper, three different deep learning models called Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN) and Temporal Convolution Network (TCN) have been used to predict the United States Dollar (USD) to Sri Lankan Rupees (LKR) exchange rate and compared the accuracy of the models. The results indicate the superiority of CNN model over other models. We conclude that CNN based models perform best in financial time series prediction.
Open Access
Determinants of stock market index movements: Evidence from New Zealand stock market
(Faculty of Graduate Studies and Research, 2017-01-27) Dassanayake, W.; Jayawardena, C.
his study examines the impact of a selected macroeconomic variables on the New Zealand stock market(S&P/NZX 50) index. We use exchange rate, interest rate, inflation rate and foreign stock market index (S&P 500 index) to evaluate their influence on the New Zealand stock market (S&P/NZX 50) index. Daily data from January 2014 to September 2016 are evaluated. Unit root tests, cointegration tests, vector error correction model (VECM) and Granger causality test are employed to examine both long run and short run dynamic relationship between these variables. The study finds that there is no statistically significant long run causality from inflation rate, exchange rate, interest rate and S&P 500 index on the New Zealand stock market index. However, S&P 500 index has a strong significant short run Granger causality to the New Zealand stock market index.
Embargo
Effectiveness of Stock Index Forecasting using ARIMA model: Evidence from New Zealand
(2021 3rd International Conference on Advancements in Computing (ICAC), SLIIT, 2021-12-09) Dassanayake, W.; Ardekani, I.; Gamage, N.; Jayawardena, C.; Sharifzadeh, H.
Time series of stock market indices are dynamic, interdependent, and considered sensitive to many factors. Thus, the prediction of such indexes is always challenging. A comprehensive review carried out by the authors finds that no attempts have yet been carried out to test ARIMA models’ predictive efficacy applied to the New Zealand financial markets. Thus, technical analysis based ARIMA prediction models are developed and empirically tested on the New Zealand stock market (NZX50) index. Daily NZX50 index data are used, and the forecasting precision of the models is assessed based on Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Square Error (RMSE]. Our study finds that ARIMA (1, 1, 0) plus intercept is the best order forecasting model out of the models we examined. Once a substantiate algorithm training is implemented, formulated ARIMA models could be successfully employed to forecast the time series of other stock market indexes or the same index for varied periods. Future researchers could compare the forecasting efficiencies of ARIMA with a deep-learning model such as long short-term memory (LSTM). The presence of limited published research of ARIMA applied to the financial markets of New Zealand validates the need and the contribution of this paper.
Embargo
Ontology based Optimized Algorithms to Communicate with a Service Robot using a User Command with Unknown Terms
(2020 2nd International Conference on Advancements in Computing (ICAC), SLIIT, 2020-12-10) Jayawardena, C.; Rajapaksha, U. U.S.
In real world applications, seamless integration of heterogeneous robots is very important to complete a task given by high level user instruction with unknown terms to all robotic devices simultaneously. In this research, we have used the technologies in Semantic Web mainly with the use of the ontology to represent the meaning of the unknown terms in the given high level instruction. If a user has given an instruction in domestic environment as “clean My Room 01 while finding my key for the car” to clean different locations with different capabilities and there can be robot who does not the meaning of the “key”. The robot can get the meaning of the unknown term by communicating with the semantic analyzer which is working with the ontology. According to our analysis we have proved that the object represented by the unknown term can be detected more accurately with compared to existing object detection algorithms since our ontology can represents more concepts related to the given object. The results indicate that if number of unknown terms in the command are increased then the time taken to process the command also be increased.
Embargo
Smart Monitoring and Disease Detection for Robotic Harvesting of Tomatoes
(2021 3rd International Conference on Advancements in Computing (ICAC), SLIIT, 2021-12-09) Pasindu, I.; Viraj, S.; Dilshan, R.; Kalhara, A.; Senaweera, O.; de Silva, R.; Jayawardena, C.
Tomato is a one of the most popular produced and extensively consumed vegetables in the world. Typical agricultural systems make extensive use of human labor which is more costly and less effective. This research explores the minimization of human labor through automation. The diseases infected by tomato plants are hard to detect. Identifying these diseases in advance would save the cultivation of the disease from spreading, thereby saving the crop.It is also a difficult task to recognize the ripe harvest and experienced labor is required. The efficiency of the harvesting method will be increased by automating the identification process of ripened fruits. Manually picking tomatoes can cause some harm to the fruits during plucking due to inconsistencies in human labor. Such damage will be reduced through a better implemented robotic scheme. This paper presents the development of autonomous system for tomato harvesting and disease detection.