BRAC University Institutional Repository
Preserving Knowledge, Advancing Research, Sharing Scholarship
A digital platform for collecting, preserving, and sharing BRAC University’s scholarly, academic, and institutional outputs.
Contribute to the Repository
Faculty members and students are invited to submit their research publications, theses, dissertations, and scholarly works to increase visibility, access, and long-term preservation.
Recent Submissions
Application of machine learning classifiers for predicting human activity
(Institute of Electrical and Electronics Engineers Inc., 2021-07-27) Alvee, Benjir Islam; Tisha, Sadia Nasrin; Chakrabarty, Amitabha; Department of Computer Science and Engineering
Involving machine learning in recognizing human activities is a widely discussed topic of this era. It has a noticeable growth of interest for implementing a wide range of applications such as health monitoring, indoor movements, navigation and location-based services. This paper compares the performance of various machine learning algorithms in the domain of human activity recognition. Data of different aged people is collected using a custom setup and custom hardware. The observed data are modeled using machine learning and neural network. As recorded human motions have variations and complexity, four dataset reduction techniques are used to manipulate the results. Best accuracy is obtained for SVM classifier with 99% accuracy and after applying PCA and SVD techniques the accuracy percentages increased to 100%. On the other hand, worst accuracy is obtained for Naive Bayes classifier before and after applying LDA technique for 100 components. The accuracy percentages are 77% and 98% respectively.
Multi criteria decision method based enterprise product ranking using fuzzy TOPSIS
(Institute of Electrical and Electronics Engineers Inc., 2021-07-27) Alvee, Benjir Islam; Chowdhury, Md. Tawfiq; Alam, Md. Golam Rabiul; Department of Computer Science and Engineering
The goal of this work is to focus on choosing product for an enterprise using fuzzy TOPSIS model. The initial focused product here in this research is laptop brand. An appropriate laptop is a vital part in a technology based era currently, which is expected to have the ideal quality and services at a reasonable price, at the same time. It is a multi criteria decision-making problem involving several criterion on which decision makers' knowledge is collected. Thus, based on the decision makers' given weights and ratings of criterion and available alternatives respectively, final ranking of the alternatives are obtained using Fuzzy TOPSIS.
Pose detection: Integrating machine learning with large vision models
(Institute of Electrical and Electronics Engineers Inc., 2025-01-01) Ashraf, Md Sadi; Akuthota V.; Paul, Tanay; Dass A.; Saha S.; Islam, Md Saidul; Chowdhury A.E.; Anwar A.S.; Roy P.; Department of Computer Science and Engineering
This paper presents a novel framework for automated yoga pose analysis that integrates computer vision with large Vision Models (LVMs) to provide detailed assessment and personalized feedback. Our system leverages Detectron 2 for pose detection and Qwen VL 2 for comprehensive pose evaluation, creating a pipeline that can identify misalignments and generate actionable guidance comparable to human instructors. In controlled evaluations across multiple yoga pose categories, our approach demonstrated superior joint accuracy (0.912) compared to established frameworks like MediaPipe (0.891) and AlphaPose (0.875). Most notably, complex poses such as inversions and deep twists showed the greatest differential benefit (53.2% improvement). Our findings demonstrate that the integration of advanced pose detection with vision-language models creates a synergistic effect that significantly enhances yoga learning outcomes. This work establishes a foundation for intelligent assistive systems in physical practice domains where precise form and alignment are critical for both effectiveness and safety.
Initiatives taken by NGOs and private companies to fight the COVID-19 pandemic
(Inderscience Publishers, 2021-01-01) Huda, S.S.M. Sadrul; Akhtar, Afsana; Maliha, Syeeda Raisa; BRAC Business School
In times where all other companies are on the verge of shutting down, app-based companies have found new meaning to their businesses through helping people. They have risen from the ashes like a phoenix and have set a marvellous example for all other companies about the meaning of life and how the goat of every company should be set. There is a saying that everything has a bright side, even a life-threatening pandemic like COVID-19 has a bright side to it. And one of those is the new definition of business and partnership. The outbreak of Coronavirus, countrywide lockdown and people in need has taught companies that making profit is not the only goal of business and partnerships. Helping people in times of crisis is even bigger than profit. You go after doing good for the people, profit will follow you anyway, and that trend has been initiated by app-based companies in Bangladesh. Copyright © 2021 Inderscience Enterprises Ltd.
Multi-user non-orthogonal waveform transceiver in RIS-assisted UAV communication: Design and implementation
(Inderscience Publishers, 2025-01-01) Faruk, Md. Omor; Sabuj, Saifur Rahman; Kamal, S.K. Tamanna; Jabin, Fowzia; Sadique, Joarder Jafor; Ullah, Shaikh Enayet; Department of Electrical and Electronic Engineering
Reconfigurable Intelligent Surface (RIS) has evolved as an energy-efficient wireless communication technology and its anticipation with Unmanned Aerial Vehicles (UAV) has extended the millimetre wave range communications by enhancing wireless coverage. In this research article, we propose a downlink multi-user RIS-assisted Non-Orthogonal Waveform (NOW) transceiver system that is energy-efficient. A NOW approach for increasing spectrum efficiency through the integration of Faster-Than-Nyquist (FTN) signalling in DFT-s-OFDM is provided. In addition, we have implemented QR-GSO pre-coding algorithm for reducing Multi-User Interference (MUI). The simulated system incorporates Low-Density Parity Check (LDPC) and Repeats and Accumulates (RA) channel coding schemes along with various types of signal detection techniques such as Zero Forcing (ZF), Cholesky Decomposition (CD)-based ZF, Lattice Reduction-based MMSE (LR-MMSE) and Minimum Mean Square Error (MMSE) to improve Bit Error Rate (BER). Results from simulations demonstrate that the suggested system is effective with regard to of BER, energy efficiency and attainable spectral efficiency. The ground users achieve an improved BER performance with a value of 1 u 10-4 at 10 dB signal-to-noise ratio under RA with CD based ZF, 8 dB under RA with CD-based ZF, 10 dB under RA with LR-MMSE and 12 dB under RA with ZF. Copyright © 2025 Inderscience Enterprises Ltd.