American International University-Bangladesh

Over the past decade, there has been a global increase in the incidence of skin cancers. Skin cancer has serious consequences if left untreated, potentially leading to more advanced cancer stages. In recent years, deep learning based convolutional neural network have emerged as powerful tools for skin cancer detection. Generally, deep learning approaches are computationally expensive and require large storage space. Therefore, deploying such a large complex model on resource‐constrained devices is challenging. An ultra‐light and accurate deep learning model is highly desirable for better inference time and memory in low‐power‐consuming devices. Knowledge distillation is an approach for transferring knowledge from a large network to a small network. This small network is easily compatible with resource‐constrained embedded devices while maintaining accuracy. The main aim of this study is to develop a deep learning‐based lightweight network based on knowledge distillation that identifies the presence of skin cancer. Here, different training strategies are implemented for the modified benchmark (Phase 1) and custom‐made model (Phase 2) and demonstrated various distillation configurations on two datasets: HAM10000 and ISIC2019. In Phase 1, the student model using knowledge distillation achieved accuracies ranging from 88.69% to 93.24% for HAM10000 and from 82.14% to 84.13% on ISIC2019. In Phase 2, the accuracies ranged from 88.63% to 88.89% on HAM10000 and from 81.39% to 83.42% on ISIC2019. These results highlight the effectiveness of knowledge distillation in improving the classification performance across diverse datasets and enabling the student model to approach the performance of the teacher model. In addition, the distilled student model can be easily deployed on resource‐constrained devices for automated skin cancer detection due to its lower computational complexity.

The intriguing behavior of doped polyanilinine/graphene oxide (PANI/GO) offers a solution to the pivotal problem of device stability against moisture in perovskite solar cell (PSC). Tunable bandgap formation of doped PANI/GO with an absorber layer allows effective flexibility for charge carrier conduction and reduced series resistance further boosting the cell performance. Herein, the L9 Orthogonal Array (OA) Taguchi-based grey relational analysis (GRA) optimization was introduced to intensify the key output responses. Furthermore, this work also delved into incorporating a Pb-free absorber perovskite layer, formamidinium tin triiodide (FASnI3), and concomitantly eluding the environmentally hazardous substance. The numerical optimization supported by statistical analysis is based on experimental data to attain the utmost peak cell efficiency. Taguchi’s L9 OA-based GRA predictive modeling recorded over one-fold enhancement over experimental results, reaching as high as 20.28% power conversion efficiency (PCE). Despite that, the PCE of the structures is severely affected by interface defects at the electron transport layer/absorber (ETL/Abs) vicinity, which is almost zero at merely 1 × 10¹⁴ cm⁻², manifesting that control measures need to be taken into account. This work deduces the feasibility of ETL/Abs stack structure in replacing the conventional Pb-based perovskite absorber layer, while maximizing the potential use of doped PANI/GO as a hole transport layer (HTL).

Breast cancer is one of the world's significant health challenges. There is a need to have better techniques in the early diagnosis of this disease to increase patients' survival rates. This paper introduces a robust approach for breast cancer detection by integrating deep-learning and classical machine-learning techniques into a custom lightweight neural network model. This approach is based on integrating conventional machine-learning and deep-learning paradigms using a range of data pre-processing steps, including data cleaning, label converting, scaling, and feature selecting to improve the given dataset's readiness for training. The proposed model demonstrated impressive accuracy in breast cancer detection compared to individual classifiers by achieving an overall accuracy of 97.54 %. Additionally, integrating eXplainable Artificial Intelligence (XAI) techniques gives the application interpretability and transparency for clinicians to make sense of the feature's importance and individual prognosis. It is a more accurate and easy-to-understand tool for clinicians, making it a better use of faulty or confusing reference values. This study presents the need to learn more about making deep learning and eXplainable Artificial Intelligence (XAI) complementary approaches to breast cancer diagnosis and treatment research.

Perovskite materials have garnered significant attention within a very short period of time by achieving competitive efficiency. In addition, this material demonstrated intriguing optoelectronic properties and versatile applications. Although they have confirmed amazing efficiency in solar cells at the laboratory scale, mass commercial manufacturing of perovskite solar cells (PSCs) is still a problem due to their poor longevity. Researchers have identified several intrinsic and extrinsic factors contributing to the instability of perovskite compounds and PSCs, and various approaches are being used to increase material quality and stability in order to extend the lifespan of PSCs. Despite these challenges, the potential of perovskite materials in revolutionizing solar energy remains a central point of scientific investigation and development. In this review, a comprehensive analysis is provided to discern the intrinsic and extrinsic factors contributing to the degradation of PSCs which certainly helps us to understand the underlying degradation mechanisms. In addition, we discussed some novel approaches that have already been adopted to augment the stability of the devices.

Dementia is a common neurological disorder that substantially impacts the global population. Primarily, it affects elderly individuals who experience a correlation between memory decline and cogni-tive function. Regrettably, no effective medications are currently available for the prevention of dementia. Doctors recommend that the early detection of this illness can assist the patient in mitigating the progression of dementia. This research uses six machine learning (ML) classifiers on the Dementia Patient Health, Prescriptions ML Dataset to detect dementia, following vital data preprocess-ing. The Light Gradient Boosting Machine (LightGBM) classifier achieved the highest accuracy of 98%. Therefore, to explain the model's predictions, two Explainable Artificial Intelligence (XAI) explainers LIME (Local Interpretable Model-Agnostic Explanations) and SHAP (Shapley Additive Explanations), have been used. Both the LIME and SHAP algorithms have effectively explained the underlying causes of dementia. This research aims to assist clinicians and doctors in accurately identifying patients with dementia by providing them with a comprehensive explanation. This will enable the prompt treatment and mitigation of the effects of dementia.

The objective of the max-cut problem is to cut any graph in such a way that the total weight of the edges that are cut off is maximum in both subsets of vertices that are divided due to the cut of the edges. Although it is an elementary graph partitioning problem, it is one of the most challenging combinatorial optimization-based problems, and tons of application areas make this problem highly admissible. Due to its admissibility, the problem is solved using the Harris Hawk Optimization algorithm (HHO). Though HHO effectively solved some engineering optimization problems, is sensitive to parameter settings and may converge slowly, potentially getting trapped in local optima. Thus, HHO and some additional operators are used to solve the max-cut problem. Crossover and refinement operators are used to modify the fitness of the hawk in such a way that they can provide precise results. A mutation mechanism along with an adjustment operator has improvised the outcome obtained from the updated hawk. To accept the potential result, the acceptance criterion has been used, and then the repair operator is applied in the proposed approach. The proposed system provided comparatively better outcomes on the G-set dataset than other state-of-the-art algorithms. It obtained 533 cuts more than the discrete cuckoo search algorithm in 9 instances, 1036 cuts more than PSO-EDA in 14 instances, and 1021 cuts more than TSHEA in 9 instances. But for four instances, the cuts are lower than PSO-EDA and TSHEA. Besides, the statistical significance has also been tested using the Wilcoxon signed rank test to provide proof of the superior performance of the proposed method. In terms of solution quality, MC-HHO can produce outcomes that are quite competitive when compared to other related state-of-the-art algorithms.

Background: Skin cancer, particularly melanoma, poses significant challenges due to the heterogeneity of skin images and the demand for accurate and interpretable diagnostic systems. Early detection and effective management are crucial for improving patient outcomes. Traditional AI models often struggle with balancing accuracy and interpretability, which are critical for clinical adoption. Methods: The SmartSkin-XAI methodology incorporates a fine-tuned DenseNet121 model combined with XAI techniques to interpret predictions. This approach improves early detection and patient management by offering a transparent decision-making process. The model was evaluated using two datasets: the ISIC dataset and the Kaggle dataset. Performance metrics such as classification accuracy, precision, recall, and F1 score were compared against benchmark models, including DenseNet121, InceptionV3, and esNet50. Results: SmartSkin-XAI achieved a classification accuracy of 97% on the ISIC dataset and 98% on the Kaggle dataset. The model demonstrated high stability in precision, recall, and F1 score measures, outperforming the benchmark models. These results underscore the robustness and applicability of SmartSkin-XAI for real-world healthcare scenarios. Conclusions: SmartSkin-XAI addresses critical challenges in melanoma diagnosis by integrating state-of-the-art architecture with XAI methods, providing both accuracy and interpretability. This approach enhances clinical decision-making, fosters trust among healthcare professionals, and represents a significant advancement in incorporating AI-driven diagnostics into medicine, particularly for bedside applications.

This study commences by delving into B-spline curves, their essential properties, and their practical implementations in the real world. It also examines the role of knot vectors, control points, and de Boor’s algorithm in creating an elegant and seamless curve. Beginning with an overview of B-spline curve theory, we delve into the necessary properties that make these curves unique. We explore their local control, smoothness, and versatility, making them well-suited for a wide range of applications. Furthermore, we examine some basic applications of B-spline curves, from designing elegant automotive curves to animating lifelike characters in the entertainment industry, making a significant impact. Utilizing the de Boor algorithm, we intricately shape the contours of everyday essentials by applying a series of control points in combination with a B-spline curve. In addition, we offer valuable insights into the diverse applications of B-spline curves in computer graphics, toy design, the electronics industry, architecture, manufacturing, and various engineering sectors. We highlight their practical utility in manipulating the shape and behavior of the curve, serving as a bridge between theory and application.

Even though no regulations require, public limited companies include management reports regarding internal controls in annual reports. Accountants and auditors are in a good position to suggest what degree of reporting is appropriate as they are directly involved in auditing financial statements and reviewing internal controls. This is a unique opportunity for management to discuss issues and concerns not communicated elsewhere in the annual report. From the very beginning there is a growing consensus as to what the content should include: financial statement presentation; purpose, nature and components of internal controls; roles of internal audit, independent auditor and audit committee. A significant number of companies studied acknowledge that “the systems are designed to provide only a reasonable assurance of meeting stated objectives.” If independent auditor’s attestation of such management reports were required; such a mandate would have a significant impact on roles of both the independent auditor and management.

This paper examines the relationship between foreign direct investments and economic growth of Bangladesh during the period 1972–2011. After reviewing the literature on the factors affecting the growth of the economy of the country, the paper empirically evaluates the most significant factors that may influence the growth of the economy of Bangladesh during the period of 1972–2011. This study evaluates the association between FDI and economic growth using multiple regression method by considering relationship between real gross domestic product, foreign direct investment, domestic investment and openness of the trade policy regime. The results indicate that domestic investments exert positive influence on economic growth whereas foreign direct investments, openness of trade are less significant.

Agriculture sector plays an important role in the Malaysian economy. Malaysia experiences deficit in food balance of trade but some of the agricultural products such as palm oil, fisheries etc. have competitive advantage. This paper aims to examine Malaysia’s export food market growth between 1996 and 2009 using shift-share analysis. Findings show that the major export commodities from Malaysia are animal or vegetable fats and oils and their cleavage products; prepared edible fats; animal or vegetable waxes (HS 15) during the said period. The increasing growth rate of Malaysia’s exports is found in the newly industrialized countries such as China, Iran, India and Ukraine due to their increasing demand for edible oil. However, influences of the trading agreements between these countries also cannot be denied.

BRAC provides microcredit to the landless and marginal borrowers to accelerate agribusiness activities in the rural areas. The prime objective of the study was to evaluate the impact of microcredit program on household income of the female borrowers of BRAC. Survey was conducted in the Gazipur district of Bangladesh. Primary data were collected from 417 borrowers who were engaged in agribusiness. Ordinary Least Square (OLS) technique was used to assess the impact of credit on household income. The study shows that the amount of microcredit received by the borrowers made a significant contribution in enhancing their household income. Besides credit, value of agricultural assets, compulsory saving, number of agribusiness pursued by household and training appeared as the key factors in determining income. The study also shows that non-institutional loan and operating cost of agribusiness adversely influenced the household income.

This empirical study attempts to test the unconditional capital asset pricing model. Two-pass regression models are employed using 86 randomly chosen companies of LSE during 1997 to 2015. A two stage approaches have been applied to investigate whether excess returns can be explained by the market risk. Based on empirical results of the first pass regression, among the 86 companies 81 companies are consistent with the prediction of CAPM except five companies. However, the estimated R-square of the sample companies are very low and indicate that market excess return has low explanatory power. In the second pass regression, empirical result shows that beta coefficient is negative and statistically significant which implies that rate of return has no linear positive relationship with beta. Further, coefficient of residual variance is also observed negative and statistically significant which violates the CAPM assumption as unsystematic risks are assumed to have no impact on rate of return. In conclusion, CAPM predictions are not consistent with the findings of this study; hence CAPM is violated and does not hold.

Identifying the EFL learners’ errors in writing has no longer been important but essential. As such, drawing the pertinent questions that what are the most common types of error committed by EFL learners in Bangladesh and what are the perceptions possessed by them concerning error correction, the article addressed the commonest errors committed by the learners and the perceptions of them toward error correction. Additionally, adopting the error analysis suggested by Ellis, the categorical presentation of the errors was also accomplished. This study comprised a corpus of EFL learners in the secondary level to enquire the commonest errors. Along with this, a student survey was carried out to reveal the perceptions of the students regarding error correction. The common errors identified were subjected to, grammar, misinformation, misordering and overgeneralization. Additionally, the study uncovered strong preference of the EFL learners to get their errors to be corrected by the teachers.

In recent years, education quality and quality assessment have received a great deal of attention at Higher Education Institutions (HEIs) in Bangladesh. Most of the HEIs in Bangladesh face severe resource constraints and find it difficult to improve education quality by improving inputs, such as better infrastructure and modernized classroom facilities. Thus, in response to the present government’s demand to improve the quality of education at HEIs in Bangladesh, it is imperative to formulate plans that are more cost-effective. According to some previous studies, the quality of education depends largely on the teaching-learning process. These studies affirm that, with limited resources at hand, the employment of active learning in the classroom is one of the most effective ways to improve education quality. To conduct this qualitative research, we utilized multiple sources of data, including semi-structured and in-depth interviews, descriptive observations and self-administered questionnaires. This paper aims to explore three related issues: What are the various active learning strategies that can be employed by the instructors at HEIs in Bangladesh? What are the potential factors that can hinder the implementation process? Finally, what recommendations can be provided on how to successfully implement active learning strategies in the classroom? The findings suggest that a lack of teacher training and student prior experience in an active learning environment, large class sizes, excessive curriculum loads and students’ academic backgrounds are some common factors that can hinder the implementation of active learning in Bangladesh. The findings of this study can be instrumental for HEIs in Bangladesh as they aspire to improve their education quality.

The mental learning arrangement of children and the impact of visual attention make computer vision researchers extremely inquisitive. Understanding child psychology in the early stage is one of the significant factors for their smooth mental growth which includes thinking patterns, learning styles, and psychological development. Visual inclining is one of the primary drivers of mental advancement in a child where the visual system is constrained by the subject’s eye. In this context, the idea of visual attention in children could uncover new learning patterns by means of the visual framework which is addressed by this research. Previous researchers attempted to discover various aspects in this context such as kid improvement, conduct acknowledgment, and field of intrigue where they did not use computer vision and augmented reality technology adequately. This research presents a rigorous investigation to learn about children's psychological development through assigning interactive tasks to children. For this, this research developed a computer vision-based augmented reality system that gets a grouping of videos and identifies data from participating individuals. In this context, this research audited the coordination of eye, head, and hand developments amid the execution of a block copying assignment. Overall experimental investigation performed by this research presents a significant pattern to understand children's phycology. The impacts of the proposed investigation using computer vision-based augmented reality will contribute significantly to design improved education materials for children from an early stage.

This research investigates the heat transfer efficiency of a double-pipe heat exchanger (DPHE) using ethanol and water as working fluids. The study focuses on the Heat Exchanger Metric (HEM) as an efficiency measure. It explores the effects of flow rates, inlet temperatures, and fluid properties on the system’s performance. Experimental setups were designed to simulate real-world applications, and data was collected using Arduino Uno microcontroller, temperature sensors, pressure sensors, flow rate sensors, and motor controllers. The HEM efficiency was calculated based on measured temperature and pressure differences. The findings of this research are expected to provide valuable insights into the performance of HEMs in double-pipe heat exchangers, particularly when considering alternative working fluids like ethanol. A counter-current flow arrangement was more efficient than a co-current flow arrangement. The research provides valuable insights for selecting the optimal working fluid in DPHE applications and contributes to understanding HEM as a reliable metric for evaluating heat exchanger performance.

This study investigates the influence of various dimensions of brand image—perceived brand quality, brand loyalty, brand awareness, perceived value, brand personality, and brand communication—on consumer behavioral intentions, with trust acting as a mediating factor. Using survey data from 130 respondents and employing structural equation modeling (SEM) via Smart PLS, the research finds that brand loyalty, awareness, perceived value, personality, and communication significantly impact consumer intentions. Notably, trust mediates the relationships between perceived value and consumer intentions, and between brand communication and consumer intentions. However, perceived brand quality shows no direct impact on consumer intentions. The study acknowledges limitations such as convenience sampling and self-reported data, suggesting avenues for future research to enhance generalizability through diverse samples and longitudinal studies. Practical implications suggest that marketers should focus on strengthening these brand dimensions to build consumer trust and drive favorable behaviors, thereby potentially enhancing brand reputation and societal well-being. The study contributes to theoretical advancements by emphasizing trust as a pivotal mediator in brand-consumer relationships, offering valuable insights for strategic brand management in competitive market environments.