http://dspace.ewubd.edu:8080/handle/123456789/231 2026-04-05T21:43:53Z 2026-04-05T21:43:53Z Newaz, Asif http://dspace.ewubd.edu:8080/handle/123456789/4442 2024-10-09T05:00:30Z 2024-09-27T00:00:00Z

Exploring New Frontiers in Imbalanced Learning: Data Complexity-Based Solutions Newaz, Asif Class imbalance is a frequently occurring scenario in classification tasks. Learning from imbalanced data poses quite a challenge which has instigated a lot of research in this area. Various techniques have been developed over the years to tackle this problem. These approaches are broadly classified into two categories: Data-level modification and Algorithm-level modification. In data-level modification, the original class distribution in the data is altered through resampling techniques. In algorithm-level modification, the traditional classification algorithms are adjusted to the imbalanced scenarios by changing the cost function and making them cost-sensitive (CS). A lot of different data resampling and CS techniques have been proposed by researchers in the past decade. To understand their strengths and weaknesses, a comprehensive experimental analysis is first conducted to obtain insights about these techniques. Several limitations have been identified that limit the performance of these approaches. Most of these techniques do not take into consideration data intrinsic characteristics that complicate the learning process. Several data difficulty factors have been identified in some previous studies which are rarely addressed in most cases. Moreover, the application of many of these techniques overfits the data and causes a loss of generalization, producing poor performance while testing. They are also unable to provide well-generalized performance on a wide range of imbalanced scenarios. In this study, novel strategies have been developed to address these issues. Solutions have been proposed to limit the effects of different data difficulty factors and enhance prediction performance. Moreover, attempts have been made to overcome the shortcomings of the established approaches and obtain better generalization. Three different methods have been proposed in this study. First, a novel data resampling technique that takes into consideration data intrinsic characteristics to effectively balance the dataset. Second, an instance complexity-based CS technique which is an advanced modification to the original CS approach. Third, a hybrid framework combining resampling and CSL. Rigorous experiments have been conducted on a wide range of imbalanced datasets to validate the performance of the proposed approaches. The results have been evaluated on eight different performance measures and compared with other state-of-the-art techniques used in imbalanced learning. Superior results have been obtained from the proposed techniques on different imbalanced scenarios. The results demonstrate the efficacy of the proposed models in learning from imbalanced data. To conclude, this research delineates new trajectories in the field of the imbalanced domain. New approaches have been proposed that introduce fresh perspectives and directions in imbalanced learning. The proposed strategies are remarkably successful, ensuring well-generalized performance when addressing imbalanced data. This thesis submitted in partial fulfillment of the requirements for the degree of Masters of Science in Computer Science and Engineering of East West University, Dhaka, Bangladesh

2024-09-27T00:00:00Z Ghyas, Qazi Adnan http://dspace.ewubd.edu:8080/handle/123456789/4093 2023-08-24T05:19:32Z 2007-12-06T00:00:00Z

Identification of Genetic Promoter through Stochastic Approach Ghyas, Qazi Adnan Analysis of a gene sequence, which is transcribed into RNA and then translated inti protein, is a difficult task. If this could be achieved, it would make possible better understand how the organisms are developed from DNA information. The behavior of gene is highly influenced by promoter sequences residing up stream or downstream of the Transcription Start Site (TSS). The promoter recognition pro, access is a part of the complex process where genes interact with each other over time and actually regulates the whole working process of a cell. This paper attempts to develop an efficient algorithm that can successfully distinguish promoters and non promoters by analyzing statistical data. A learning model is developed from the known dataset to predict the unknown ones. Results: We have developed an efficient algorithm that can successfully distinguish genes from non-gene sequences by analyzing statistical data. A learning model is initially developed to train the Support Vector Machine (SVM) to identify distinctive features between gene and non gene. Then this context was used to predict other foreign sequence by the SVM. Our system has been tested using standard plant prom data sequence from the EMBL and the performances are: 0.86 for the Sensitivity and 0.90 for the specificity. Identification This thesis submitted in partial fulfillment of the requirements for the degree of Masters of Science in Computer Science and Engineering of East West University, Dhaka, Bangladesh

2007-12-06T00:00:00Z Saha, Pritom http://dspace.ewubd.edu:8080/handle/123456789/3563 2022-05-25T06:11:16Z 2020-01-05T00:00:00Z

Using Ethernet/Gigabit Passive Optical Network (EPON/GPON) Over Traditional Broadband Networks in Bangladesh, its Challenges and Solutions Saha, Pritom Accessing Internet is going to be a fundamental right like other basic human rights. No matter how often a user uses Internet, waiting for a web page to load is really an annoyance. To overcome this situation, Internet speed has increased significantly in the past decade to keep pace with the demand of users, new services and bandwidth-hungry applications. These demands include as multimedia content based e-commerce, video on demand, high definition TV, IPTV, online gaming, social media, etc. Lots of different technologies like gigabit Ethernet, mobile Internet, etc. that have been trying to cater and satisfy the user demand. FTTH(Fiber to the home) is one of the high-speed access technology which offers triple play service (i. e., data, voice and video) EPON/GPON is easier, faster and low cost. EPON/ GPON can transport Ethernet, ATM and TDM (PSTN, ISDN, E1 and E3) traffic. It has been widely accepted by Internet service providers and operators in terms of bandwidth upgrade, service bearer as well as passive network maintenance. High bandwidth capacity, high quality and multi-play services through a single channel presents a strong business opportunity for telecom operators as it provides support which includes voice, Ethernet, ATM, leased lines and others. FTTH offers simple operation, administration, maintenance and provisioning capabilities as well as end-toend service management. It not only provides substantially higher efficiency as a transport network, but also delivers simplicity and superb scalability for future expansion in supporting additional services. According to the contents above, we can see that there are own advantages and disadvantages of EPON and GPON. In performance comparison, GPON is better than EPON, while EPON has many advantages on time and cost. And now, EPON is still the mainstream of PON, and meanwhile, GPON is catching up with it. Going forward to the broadband access market, maybe it is not substitution between EPON and GPON. It is more likely that they are coexist to complement each other. For the users who have demands of multi-service, high QoS and security, as well as ATM backbone network, GPON seems to be an ideal. This thesis submitted in partial fulfillment of the requirements for the degree of Masters of Science in Computer Science and Engineering of East West University, Dhaka, Bangladesh

2020-01-05T00:00:00Z Muksud, Md. Turza Ibna http://dspace.ewubd.edu:8080/handle/123456789/3562 2022-05-25T05:51:26Z 2020-02-13T00:00:00Z

PANI: An App for Dynamic Irrigation Scheduling Muksud, Md. Turza Ibna Large tracts of land in the delta region of Bangladesh are left fallow or are cultivated with low input crops during the dry winter months. Surface water based irrigation opens up new opportunities for sustainable intensification by enabling the production of a high yielding crop. Irrigation in Bangladesh is still managed in a very traditional manner without considering scientific data on soil and weather which increase cost of farming and also lower productivity. Farmers usually depends on their past experience. In a traditional way, they forecast the weather as well as the moisture of the land. So the process the irrigation depends more on their guess. The simulation model has been integrated into a smart phone app called PANI (Program for Advanced Numerical Irrigation). PANI runs on a daily time step and uses forecasted weather data to predict irrigation needs one week in advance. I used the calibration of the Blainy-Criddle Equation so that only forecasted daily maximum and minimum temperatures are required. PANI addresses the needs of the irrigation service provider as well as of the farmer. Both will receive the field data, informing them as to whether a field needs to be irrigated or not. Since the water balance of PANI is based on the SVM algorithm, it can be used under most conditions, as long as weather and ground cover data are available. This thesis submitted in partial fulfillment of the requirements for the degree of Masters of Science in Computer Science and Engineering of East West University, Dhaka, Bangladesh

2020-02-13T00:00:00Z