Chittagong University of Engineering & Technology

About: Chittagong University of Engineering & Technology is a education organization based out in Chittagong, Bangladesh. It is known for research contribution in the topics: Renewable energy & Dielectric. The organization has 1200 authors who have published 1444 publications receiving 10418 citations. The organization is also known as: Engineering College, Chittagong & Bangladesh Institute of Technology, Chittagong.

An Automated Candidate Selection System Using Bangla Language Processing

Abstract: Recruiting or selecting the right candidates from a vast pool of candidates has always been a fundamental issue in Bangladesh as far as employers are concerned. In the case of candidate recruitment, different government organizations, nowadays, ask the applicants to submit their applications or resumes written in Bengali in the form of electronic documents. Matching the skills with the requirements and choosing the best candidates manually from all the resumes written in Bengali is very difficult and time-consuming. To make the recruitment process more comfortable, we have developed an automated candidate selection system. First, it takes the CVs (written in Bengali) of candidates and the employer’s requirements as input. It extracts information from the candidate’s CV using Bangla Language Processing (BLP) and Word2Vec embedding. Then, it generates an average cosine similarity score for each CV. Finally, it ranks the candidates according to the average cosine similarity scores and returns the dominant candidate’s list.

Imaging techniques to extract information: New born baby's skin birthmark

Abstract: Birthmark on human skin can be used for digital identification specially for new born baby, because their skin usually scars, tattoo, and others unexpected lesion free. The available methods to extract skin birthmark information take long time and depend on clinical process. Computer based Imaging program can be the solution to minimize time delay and reduce the complexity. Extracting information of skin birthmark is critical and complex process because of skin tone variation and shape of different organs. Different types of imaging operation such as edge detection for shape analysis, morphological operation for getting clear view, and mathematical image enhancing operations are applied for the extracting process to visualize the identifying information of new born baby. By comparing with other approaches, the successfulness of this new technique has been demonstrated for the forensic study where the average success rate is 94.75%.

Toward Embedding Hyperparameters Optimization: Analyzing Their Impacts on Deep Leaning-Based Text Classification

Abstract: In the last few years, an enormous amount of unstructured text documents has been added to the World Wide Web because of the availability of electronics gadgets and increases the usability of the Internet. Using text classification, this large amount of texts are appropriately organized, searched, and manipulated by the high resource language (e.g., English). Nevertheless, till now, it is a so-called issue for low-resource languages (like Bengali). There is no usable research and has conducted on Bengali text classification owing to the lack of standard corpora, shortage of hyperparameters tuning method of text embeddings and insufficiency of embedding model evaluations system (e.g., intrinsic and extrinsic). Text classification performance depends on embedding features, and the best embedding hyperparameter settings can produce the best embedding feature. The embedding model default hyperparameters values are developed for high resource language, and these hyperparameters settings are not well performed for low-resource languages. The low-resource hyperparameters tuning is a crucial task for the text classification domain. This study investigates the influence of embedding hyperparameters on Bengali text classification. The empirical analysis concludes that an automatic embedding hyperparameter tuning (AEHT) with convolutional neural networks (CNNs) attained the maximum text classification accuracy of 95.16 and 86.41% for BARD and IndicNLP datasets.

Design and Optimization of Efficient $\mathrm{FeS}_{2}$ Solar Cell

Abstract: Nowadays, one of the key priorities in photovoltaic solar cell technology is reducing fabrication cost, at that time maintaining high conversion efficiency. The FeS 2 (pyrite) is an ecologically viable semiconductor material which has very suitable optoelectronics properties for the economically benign solar cell. So, nowadays researcher shows interest in FeS 2 based solar cell. In this research work, numerical analysis of FeS 2 solar cell is done with CdS as a potential buffer layer. FTO is used as a TCO layer to get better performance. In the study, the proposed ultra-thin cell structure was found with a conversion efficiency of 17.28% with $\pmb{V_{oc}=0.56}$ V, $\pmb{J_{sc}=37.99}\ \mathbf{mA/}\mathbf{cm}^{2},\pmb{FF=81.89\%}$ . The thermal stability of the cell is also investigated.