Search

An electroencephalogram (EEG) is a measurement that reflects the overall electrical activity in the brain. EEG signals are effective for biometric authentication and robust against malware attacks and any kind of fraud activities due to the uniqueness of the signals. Significant progress in research on EEG-based authentication has been achieved in the last few years, with machine learning being extensively used for classifying EEG signals. However, to the best of our knowledge, there has been no investigation into the overall progress made in such research. In this paper, the literature on the various factors involved in state-of-the-art biometric authentication systems is reviewed. We provide a thorough comparison of different machine learning biometric authentication techniques. The comparison criteria include the research objectives, machine learning algorithms, computational complexity, source of brainwaves, feature extraction methods, number of channels, and so on. Alongside the discussion of existing works, directions for future research are suggested to improve authentication accuracy. This paper provides an in-depth discussion of different advanced biometric authentication techniques, and a vivid picture of state-of-the-art machine learning-based biometric authentication techniques using EEG.

The research investigation was carried out in the experimental area of Sher-e-Bangla Agricultural University, Sher-e-Bangla Nagar, Dhaka-1207, during the late Rabi season Mid December to May 2018 to study the genetic variability and character association of yield components in Maize with 35 maize genotypes as experimental materials that were laid out in Randomized Complete Block Design (RCBD) with three replications. The research work was oriented to calculate and estimate the yield factor through analyzing the mean performances, variability, heritability, genetic advance, correlation coefficient and path analysis involving the yield contributing characters. The maximum yield per plant (117.51 g) was estimated in the genotype G12 (Pacific) and the minimum yield per plant (51.89 g) was recorded in the genotype G17 (Dekalb Super). The phenotypic coefficient of variation was observed higher than the genotypic coefficient of variation for all yield contributing characters. Heritability was also shown higher compared to genetic advance in comparative mean study for all the characters involved. Significant and positive correlation was observed for grain yield per plant with plant height (0.774), cob length (0.956), grain per row (0.703) and 100-seed weight (0.365). In path analysis study, positive direct correlation was recorded in days to male flowering (0.247), cob breadth (1.078), number of grains per cob (6.874) and 100-seed weight (0.595).

The Resource-Constrained Project Scheduling Problem (RCPSP) is a challenging optimization problem. In RCPSPs, it is very common to consider homogeneous activities, which means all activities require all types of resources. In practice, the activities are often singular because they usually require one single resource to execute an activity. The existing algorithms may be used for solving this variant of RCPSPs with a simple modification. However, they are computationally expensive due to unnecessary resource constraints. In this paper, we propose a customised evolutionary algorithm integrated with three heuristics for the singular activities. The first heuristic is based on the earliest start time with an aim to rectify an infeasible schedule. The second heuristic is based on neighbourhood swapping which is used to find the best possible alternatives. The third heuristic is used to further enhance the quality of the schedule. The performance of the proposed framework has been tested by solving a wide range of benchmark problems and the obtained results revealed that the proposed approach outperformed the existing algorithms. In addition, statistical and parametric testing show the value and characteristics of the proposed approach.

Background Self-medication practice (SMP) is widely adopted among residents of such countries as Bangladesh. However, the major determinants of SMP have only scarcely been evaluated. Therefore, this cross-sectional study aimed to fill this knowledge gap by assessing the prevalence and determining factors of SMP. Methods A convenience sample of individuals residing in Savar (Dhaka, Bangladesh) >16 years of age and suffering from an illness >2 months underwent face-to-face interviews during January-February, 2019. Results Among 520 interviewees fulfilling entry criteria, the prevalence of SMP was reported as 60.2%. Drug shops, previous knowledge or prescription, and consulting with peers (friends or relatives) were the most sought sources of SMP, whereas simple illness, higher consulting fees, unsatisfactory health-care services, and delayed access were reported as the motives. From the regression analysis, the sociodemographic risk factors of SMP were reported as being older (compared to

Background: Cholecystectomy is one of the commonest surgeries in medical practice. Sometimes malignant condition cannot be assessed pre-operatively. Histopathology require for tissue diagnosis. Objective: To assess whether the histopathogical examination is really needed for all bladder specimens resected for cholelithiasis and cholecystitis. Materials and Methods: This observational study was carried out in the General Surgery Unit, Department of Surgery, Bangabandhu Sheikh Mujib Medical University, Dhaka from May, 2016 to April, 2017. A total number of 95 patients were included in this study. Results: Total of 95 gallbladder specimens were submitted for histopathological examination where 33.7% were males and 66.3 % were females with mean age of 46+12.86 years. Chronic cholecystitis was the most common finding in 91.6% of cases whereas Xanthogranulomatous 2.1%. Six of the gall bladders showed adenocarcinoma (6.3%), of which four had changes on ultrasound and macroscopic gall bladder analysis (MGAS) disturbances were found in 5 cases and 1 case was found incidentally on histopathological examination. On ultrasound, only four (66.66%) and peroperative macroscopic examination by surgeon only 5(83.3%) had a clue of carcinoma. Conclusion: It could be concluded that , routine histopathological examination of gallbladder after cholecystectomy is necessary as USG sometimes can not differentiate neoplastic lesion from inflammatory one. KYAMC Journal Vol. 10, No.-4, January 2020, Page 184-187

Background: Preoperative biliary drainage before pancreaticoduodenectomy is a controversial issue. Proponents are in favor of preoperative biliary drainage by ERCP with stent to reduce surgical jaundice with an anticipation of better surgical outcome. Objective: Compare the outcome with or without pre-operative biliary drainage before pancreaticoduodenectomy. Materials and Methods: This observational comparative study was conducted in department of Surgery and Hepatobiliary and pancreatic surgery of BSMMU. Twenty three patients presented with obstructive jaundice due to periampulary carcinoma who subsequently underwent pancreaticoduodenectomy were selected by purposive sampling and finalized by eligibility criteria. Results: Patients with preoperative biliary drainage (PBD) group required a longer operative time (mean 4.12 hours versus 3.83 hours) and had more intra-operative blood loss (mean 662 mL versus 495 mL) compared with non PBD group (P=0.009 and 0.010). No differences were found with respect to operative mortality (4.3%) and incidence of pancreatic leakage (P=0.281). PBD was significantly associated with positive bile culture (P=0.019) and high incidence of wound infection (p=0.029). Conclusion: Preoperative biliary drainage did not increase major postoperative morbidity and mortality but associated with increased operative time, intraoperative blood loss, and incidence of wound infection. Preoperative biliary drainage should be used selectively in patients undergoing pancreaticoduodenectomy. KYAMC Journal Vol. 10, No.-4, January 2020, Page 196-201

Background: Recent studies suggest that surgical diathermy shows better clinical outcome in the context of incision time, wound related postoperative pain, postoperative wound infections, and length of postoperative hospital stay and cosmetic outcome of scar in cases of elective surgical patients. Objectives: Compare the efficacy and safety of surgical diathermy versus conventional surgical blade for making skin incisions in elective mid-line laparotomy and to evaluate weather cutting diathermy is an effective and better alternative to surgical blade incision. Materials and Methods: This prospective study was carried out in the department of surgery at BSMMU, Dhaka over a period of one year. Sample size was 64 with a follow up duration for 6 month. In Group I (D), skin incision was taken with diathermy, and in Group II (S), incision was taken with surgical blade. Results: Compared with a scalpel incision, cutting diathermy resulted in significantly shorter incision times and reduced post-operative wound related pain (P =

Sesame (Sesamum indicum L.) is one of the oldest oilseed crops and important for high nutritional quality as well as medicinal value. Fifty diverse sesame genotypes were evaluated to study genetic variability. The results revealed that the genotypes were a significant variation in most of the studied characters. In all cases, the phenotypic variances were much higher than genotypic variances suggests a higher level of the environmental effect on the expression of these characters. The highest genotypic coefficient of variations (GCV) was observed in seed yield per plant while the highest heritability was exhibited by hundred seed weight followed by days to 80% maturity, pods per plant, number of branches per plant and seed yield per plant. The genotypic correlation with seed yield per plant showed a significantly strong positive with days to 50% flowering, plant height and number of pods per plant at both the genotypic and phenotypic level. The path coefficient analysis showed that pods per plant and seeds per pod were the most important contributing traits to seed yield. The 50 sesame genotypes were grouped into five clusters. The highest inter-cluster distance was observed between the cluster III and V while the lowest inter-cluster distance was observed between the cluster III and IV. Among 50 sesame genotypes G7, G36, G38 and G46 might be suggested for future hybridization program for the improvement of sesame yield.

Ferrocement is composed of cement mortar reinforced with small diameter closely spaced steel wire mesh to form a thin section conforming high performance of serviceability. The present study investigates the performance of ferrocement panels focused on the mechanical properties, water absorption and durability. A series of specimens were cast with single and double mesh layers. Flexural performance was carried out following sixty days in temperature cycle and in corrosion cell. These results was compared with the controlled sample. The test results shows that the flexural strength performance was reduced by 52% and 35% for single and double layer wire mesh samples respectively followed by corrosion environment. First crack load also changed after completion of sixty temperature cycles. This load is 27.3%, 42.3% and 31.8% of failure load for controlled sample, sample in room air after every temperature cycle and samples in room air following quenching after every temperature cycle respectively for single mesh layer sample, and for double mesh layer sample these values are 38.9%, 30.1% and 17.7%. Early first crack is found for the samples following quenching and cooled in room air; however, both types of samples are in low absorption level. This data represents that double layer mesh specimen exhibits better when compared to single layer mesh specimen in strength and corrosion parameters.

The present study was conducted involving 62 F3 genotypes of Brassica napus L. at the experimental farm of Sher-e-Bangla Agricultural University, Dhaka, Bangladesh to ponder the genetic variability, phenotypic, genotypic and environmental coefficient of variation, heritability and genetic advance, correlation, path coefficient and genetic diversity analysis in a randomized complete block design (RCBD) with sixty-two genotypes (treatments) with three replications during November 2014 to February 2015. The investigations aimed to select the best segregating genotypes for the yield improvement of Brussica napus (rapeseed). Analysis of variance indicated that the genotypes were found significantly different for all the characters considered. The relative phenotypic coefficient of variation (PCV) was higher than the genotypic coefficient of variation (GCV) for all the traits investigated. The high GCV value was observed for the number of siliqua per plant (NSP), plant height (PH), silique length (SL), number of seed per silique (NSS) and seed yield per plant (SYP) indicated high broad sense heritability. The significant positive correlation with seed yield per plant (SYP) was found in plant height (PH) (0.368**), the number of primary branches per plant (NPB) (0.332**), number of secondary branches per plant (NSB) (0.382**), number of silique per plant (NSP) (0.549**), and siliqua length (SL) (0.037**). The results of path coefficient analysis uncovered that plant height (PH) (0.582), days to 50% flowering (50F) (0.390), days to maturity (DM) (0.575), number of primary branches per plant (NPB) (0.678), number of secondary branches per plant (NSB) (0.182),and thousand seed weight (TSW) (0.289) had a positive direct impact on seed yield per plant (SYP) and thus it was concluded that these traits could be exploited for the enhancement of yield potential of rapeseed. This study showed that based on the agronomic performance execution, genotypes G8, G14, G19, G21, G47, and G55 might be proposed for future hybridization program in Bangladesh and this could help rapeseed breeders to upgrade their breeding activities.

The confidentiality of wireless multicast network is notably compromised by the concurrent existence of shadowing and fading. To overcome this security threat, mathematical modeling of a multicast scenario is proposed in this paper over κ-μ/ inverse gamma (IG) composite fading channel where transmitted confidential information from source is under malicious attack of multiple eavesdroppers. The well-known physical layer security approach is adopted to analyze the impact of system parameters on secrecy performance in terms of novel expressions of secure outage probability for multicasting, probability of nonzero secrecy multicast capacity and ergodic secrecy multicast capacity. The generalization of security analysis is achieved by demonstrating Rayleigh, Rician, Nakagami-m, κ-μ fading and Gaussian channels as a special scenarios of κ-μ/IG distribution. At the end, analytical results are justified by performing Monte-Carlo Simulations.

In this paper, physical layer security over α – κ – μ fading distribution is investigated for a long distance wireless communication scheme. This evaluation is undertaken considering the existence of multiple receivers and eavesdroppers in the network. In order to diminish the possibility of wiretapping and to maintain perfect secrecy, it must be assured that no eavesdropper is able to overhear private information from the multicast receiving channels. This condition is ensured by deriving the novel closed-form expressions of probability of nonzero secrecy multicast capacity using multinomial theorem and the ergodic secrecy multicast capacity using the well known Meijer-G function. Afterwards, advanced numerical analysis of the obtained security matrices has been performed to exhibit how the physical parameters of α – κ – μ fading channel can be tuned to achieve maximum security. Finally, Monte-Carlo simulation is conducted in MATLAB environment to verify the accuracy of the derived expressions.

Resource-Constrained Project Scheduling Problem (RCPSP) is a well-known challenging optimization problem. Over the last few decades, many different approaches have been proposed for solving different variant of RCPSPs. In those problems, it is usually assumed that the activities are homogeneous, that means, all activities are similar or they require similar types of resource over the entire project horizon. However, in real-world applications, a project consists of heterogeneous activities where the activities really need different kinds of resource. The existing algorithms, which are mostly developed for RCPSPs with homogeneous activities, can be applied to the projects with heterogeneous activities with a simple modification. However, they are computationally expensive as they need to perform many unnecessary computation. So, in this paper, we propose a customized approach for RCPSPs with heterogeneous activities based on a memetic algorithm. The proposed memetic algorithm is basically a genetic algorithm combined with a local search. The performance of the propose algorithm has been tested by solving a good number of problem instances and a few real-world problems, and comparing the results with an appropriate state-of-the-art algorithm. It is found that propose approach obtains high quality solutions within a considerably lower computational effort.

This paper presents a comparative study to recognize faces from a customized dataset of 10 identities of different celebrities using Convolutional Neural Network based models such as AlexNet, VGG16, VGG19 and MobileNet. These pre-trained models previously trained on ImageNet dataset are used with the application of Transfer Learning and Fine Tuning. For our experiment we used Keras API with TensorFlow backend written in Python. The performance analysis includes training, validation, and testing on different images created from original dataset. The validation accuracy of VGG19 model is found better than the other three but MobileNet model showed better test accuracy.

Over the last few decades, many solution approaches have been developed for solving different variants of resourceconstrained project scheduling problems (RCPSPs). In most of them, it is assumed that a project consists of some homogeneous activities that require all types of resources over the entire project horizon. On the contrary, many real-world projects consist of heterogeneous activities that use different types of resources at different time instants during the project execution. The application of existing approaches, developed for RCPSPs with homogeneous activities, in solving RCPSPs with heterogeneous activities is computationally expensive. In this paper, we propose a heuristic embedded genetic algorithm to address RCPSPs with heterogeneous activities. Two heuristics are proposed to obtain high-quality feasible solutions. The first heuristic is based on priority rules while the second one based on a new neighbourhood swapping matrix. To evaluate the performance of the proposed algorithm, we solve a number of real-world and modified test problems, and the obtained results are compared with an existing algorithm. It is found that the proposed approach obtains highquality solutions with a significantly lower computational time compared to other algorithms.

The increasing development of the biomedical system and smart technology has a major impact on smart devices. A smart wheelchair is one of them to be improved with the blessings of this modern technology. In this paper, a smart wheelchair topology is proposed which is operated by a hand movement device and a smartphone. It comes with a lot of advanced features for people with disabilities who cannot walk or travel without the help of others. It is a hand-held wheelchair in which the gyro sensor and accelerometer are used and the Bluetooth phone control module is used to make it automatic. Users will wear a gesture system in their hands, and by moving the hand, the wheelchair will move forward, backward, left, and right. Arduino Mega and Arduino Nano are used as controllers. In this paper, the minimum threshold angle is compared with a microcontroller-based wheelchair where this proposed wheelchair started working with 5° fewer angles for forwarding and backward movement and 3° fewer angles for the left and right movement. Moreover, the linearity of this proposed wheelchair is −0.7, 0.045 & −0.03 when the sensitivity is 0.6102, 0.5214 & 0.55 for X, Y, Z axis respectively. The Sonar sensor is used here to prevent a safe movement. Using this design dimension and configuration, a prototype was eventually built and evaluated at various stages for performance evaluation.

Early diagnosis of skin diseases, mostly in developing countries, is necessary since it has serious impact on people’s quality of life. In this study, an automated model is proposed to classify skin diseases for dermoscopic images using texture analysis. There are some challenges for automating the classification of skin lesions such as existence of hair in the lesion, dark and highlight regions in dermoscopic image and different types of skin tone. In this study, these challenges are addressed with class imbalance issue, an internecine and common problem of skin disease image datasets. By combining different techniques or algorithms an effective approach is introduced to allay those challenges. The performance of the proposed approach is tested for two dermatological skin conditions viz. Solar Lentigo, Lentigo Simplex using k-th nearest neighbor (kNN) and support vector machine (SVM) as classifier.

Facial expressions have been studied extensively for the analysis of human sentiment properly. A human emotion recognition system through recognizing human facial expression is proposed in this paper. After preprocessing, segmentation of the facial expression regions is done in a unique yet effective and easy way to segment the left eye, right eye, nose, mouth properly from the facial region. 2D Gabor filter is used for the extraction of features from the expression regions. For reducing the dimension of the extracted features, downsampling and Principal Component Analysis (PCA) is used. For carrying out the classification task multiclass Support Vector Machine (SVM) is used for its ability to handle complex problems in high dimensional spaces. Three publicly available facial expression dataset was used to evaluate the performance of the proposed system. Finally, performance on these datasets by the proposed method is compared to previously attained performance by different methods which indicate that the proposed method attains state-of-the-art performance.

Facial expression is an effective method of nonverbal communication and expressing feelings. A method to recognize emotions through facial expressions is proposed in this paper. After some preprocessing of the input image, the facial region is segmented into four expression regions according to the highly effective proposed segmentation method. Features from these segmented parts are extracted using a fusion of Histogram of Oriented Gradients (HOG) and Local Binary Patterns (LBP). Reduction of the dimension of the feature vector is done using Principal Component Analysis (PCA). For classifying the features and thus the expression images, multiclass Support Vector Machine (SVM) is used. The performance of the proposed method is measured using three publicly available and highly used datasets (JAFFE, CK+, RaFD). Finally, achieved performance is compared with performance on these datasets by other available methods to indicate that the proposed method succeeds in achieving state-of-the-art performance.

Inability to look after and take care of own self in a healthy way is referred to as self-care problems. Physical disabilities are limitations in physical capacity, mobility of a person. Motor disability is defined as the loss of function of any body part. Self-care problems detection is an important sector of the exceptional children health care system because depending on the type of self-care problems, the treatment approach is chosen. Expensive involvement of professional therapists and a long time is required to detect self-care problems. To predict the self-care problems of children with physical and motor disability in advance, an expert system is proposed using machine learning. To do the task, the only available self-care activities dataset SCADI (Self-Care Activities Dataset based on ICF-CY) is used. Principal Component Analysis (PCA) is used to reduce the feature vector size. Different machine learning classification algorithms are used to find out the prediction accuracy. Prediction performances on different partitions of the dataset are analyzed. Finally, a system with K Nearest Neighbor (KNN) is proposed to predict self-care problems.

Website classification is a convenient starting point for building an intelligent web browser and social networking sites that can understand the favorite categories of a user and also detect adult or harmful websites perfectly. Classifying the web sites using the information of the Uniform Resource Locator (URL) is an important and fast technique. A perfect result is needed for URL classification to make it usable in the real world applications. So we have proposed an improved approach for URL classification that is able to provide a better result. We have introduced the word-based multiple n-gram models for efficient feature extraction and multinomial distribution for Naive Bayes classifier under the Random Search pipeline for hyperparameter optimization that finds the best parameters of the URL features. The experimental result of our research is compared with the result of previous research works and we have shown a better result than the existing result. Our experimental result provides 88.77% in recall and 87.63% in F1-Score which is the best performance so far.

A framework to recognize human emotion through facial expression recognition is proposed in this paper where segmentation of the expression regions (right eye, left eye, nose, mouth) are done manually in an easy yet effective, unique manner by analyzing many facial expression images and the possible positions of the expression regions in those images. For feature extraction from the segmented parts 2D Gabor filter is used with multiple frequency and orientation. Redundant features from the extracted features are eliminated using downsampling. Finally, Extreme Learning Machine (ELM) is used to handle the classification process. For performance evaluation of the proposed method, four different datasets (JAFFE, CK+, RaFD, KDEF) have been used and impressive correct recognition rate on these four datasets indicates the capability of the proposed system to recognize human emotion through facial expression recognition of front-facing images.

Recently heart disease has become the most common fatal diseases in the world. Early stage detection and treatment can reduce the number of cardiac failures, mortality of heart disease and cost of diagnosis. The healthcare industry collects a huge amount of these medical data, but unfortunately, these are not mined. Discovery of hidden patterns and relationships from this data can help effective decision making to predict the risk of heart disease. The main objective of this research is to develop a Robust Intelligent Heart Disease Prediction System (RIHDPS) using some classification algorithms namely, Naive Bayes, Logistic Regression and Neural Network. This article reviewed the effectiveness of clinical decision support systems by ensemble methods of these three algorithms.

Mental condition and sentiment of a person can be analyzed through facial expressions. An emotion recognition system is proposed by recognizing facial expressions. Input images are preprocessed and then proposed image segmentation method is applied to segment a facial image into four parts that contribute highly in representing facial expressions. Features are extracted from the segmented parts using a fusion of Histogram of Oriented Gradients (HOG) and Local Binary Patterns (LBP). The dimension of the feature vector is reduced using Principal Component Analysis (PCA). Finally, Artificial Neural Network (ANN) is used to classify the facial expressions properly. The proposed system is tested using three widely used facial expression datasets (JAFFE, CK +, RaFD). At last, the achieved performance is compared with other facial expression recognition systems to justify that the proposed method succeeds in achieving state-of-the-art performance.

Bangla handwritten digit recognition is a convenient starting point for building an OCR in the Bengali language. Lack of large and unbiased dataset, Bangla digit recognition was not standardized previously. But in this paper, a large and unbiased dataset known as NumtaDB is used for Bangla digit recognition. The challenges of the NumtaDB dataset are highly unprocessed and augmented images. So different kinds of preprocessing techniques are used for processing images and deep convolutional neural network (CNN) is used as the classification model in this paper. The deep convolutional neural network model has shown an excellent performance, securing the 13th position with 92.72% testing accuracy in the Bengali handwritten digit recognition challenge 2018 among 57 participating teams. A study of the network performance on the MNIST and EMNIST datasets were performed in order to bolster the analysis.

Facial expressions contribute highly in conveying the feelings of a person. An emotion recognition system through facial expression recognition is proposed in this paper. Preprocessed input images are segmented into four facial expression regions by following the proposed highly effective image segmentation method. 2D Gabor filter with different frequencies and orientations are used to extract features from the segmented parts. Reduction of the dimension of the extracted features is done using downsampling and Principal Component Analysis (PCA). Classification of the features is done using the artificial neural network (multilayer perceptrons with backpropagation). To evaluate the performance of the proposed method three widely used facial expression datasets (JAFFE, CK+, RaFD) are used. Performance on these datasets by the proposed method is compared with the performance on these datasets by other methods to indicate that state-of-the-art performance is achieved by the proposed method.

Ontology (Onto=Being and Logy=Knowledge, therefore the Knowledge of Being) has a significant impact on the study of natural language processing. By providing a formal representation of knowledge it ensures proper understanding of a particular domain. A comprehensive description to build a vocabulary on the given domain can never be feasible without ontologies; in other word conceptualizations. There have been a number of works in the recent times to perceive the ontological knowledge to build a strong vocabulary. Most of the existing ontology construction tools support construction of ontological relations (e.g., taxonomy, equivalence, etc.). But the main problem is that they do not support construction of domain relations, non-taxonomic conceptual relationships (e.g., causes, caused by, treat, treated by, has-member, contain, material-of, operated-by, controls, etc.) which are basically found in the text sources. The first notable work on this field is a Named Entity Recognition system developed by Stanford University. Stanford NER (also known as CRFClassifier) is a Java implementation of a Named Entity Recognizer. It can successfully recognize at most seven classes. In this research work we have utilized this NER, and proposed an algorithm which includes POS tagging, lemmatization, parsing and pronoun and co-reference resolution. We can squeeze out 22 classes from these seven primary classes. We have compared our work with an existing system named TextOntoEx. On the basis of performance matrices we have analyzed both of the works on a same natural language text. Result shows that our proposed system can find out more classes than TextOntoEx.

Biometric authentication has become a popular and required approach to increase dynamism and security of shared information or place. But single modality fails to fulfill the present demand of accuracy and security in some cases. That's why fusion of multimodal biometrics are used to improve the identification efficiency. This paper also deals with the development of such a weighted score level fusion technique by consolidating the significance of human iris and face. Here the algorithms developed by Daugman has been used for Iris recognition process and PCA has been used for the extraction and representation of the features of human face. Finally individual iris and face matching score have been merged using weighted sum rule. Identification of the individual is confirmed based on the comparison with the weighted score. The developed multimodal technique improves both the recognition accuracy and robustness of the system.

not available DOI: http://dx.doi.org/10.3329/sje.v7i0.14470 Stamford Journal of English; Volume 7; Page 145-158

Any message encoded in a certain discourse on the common ground of reference, purpose or aim that a discourse serves, runs the risk of semantic gap at the crucial moment of decoding. The response(s) may contradict the feedback, or else the feedback may bring about epistemic violence and thereby may subvert the projected meaning(s). Teaching T.S. Eliot’s poem, “The Love Song of J. Alfred Prufrock” as a part of an undergraduate course at a University (in Bangladeshi context) led us to the discovery of a threefold analysis (phase by phase) of the poem: a. constituent analysis, b. structuralist reading, and c. dialogic criticism, which was quite challenging. This paper aims at analysing the stated poetic piece as a discourse and setting it against the backdrop of a literature classroom at the tertiary level with a view to understanding how the aforementioned three-phase-analysis of a text can help the learners go beyond the limit of identifying with the objective reading of a discourse, and thus they pave their way to the undiscovered world of signification and eventually can relate to a social context where the meaning is multiple and the voices are polyphonic. Stamford Journal of English; Volume 6; Page 27-40 DOI: http://dx.doi.org/10.3329/sje.v6i0.13900

Different types of coolant are widely used in different metal cutting processes to improve the machining responses. But the suitability of using the correct cutting fluid is very important considering the concept of green environment. In this study, hot air is used as an alternative approach for hot machining process and coolant. Hot air is considered to initially heat the work-piece for easy machining operation. In this study, two different velocities of hot air has been applied during the machining of mild steel in turning process and its effect is investigated in terms of surface roughness. With the variation of different process parameters, it has also been observed that surface roughness at different cutting conditions using hot air is improved significantly. A clear comparison has been made to investigate the responses of surface roughness at different cutting conditions in between the hot air and normal machining processes. This procedure may be used as an alternative approach in the dry cutting re...

Face recognition has been gaining popularity for long time in various fields of human computer interaction. Moreover face recognition technique is widely used for automatic biometric security control, document verification, criminal investigation etc. In this paper we propose a new approach of using PCA based face recognition method for human verification. PCA based method seems to be interested due to its simplicity and better accuracy. In our proposed method two stages of authentication are performed for recognizing a single candidate individual. At first stage the candidate face is matched with all stored faces and only few best matched samples are isolated to use as second stage training samples. Here in both stages PCA is used for extracting significant features of face. Our proposed approach showed 1.5% better accuracy for ORL face database and 0.52% better accuracy for Face94 database.

This paper represents an Artificial Neural Network based approach for the recognition of English characters using feed forward neural network. Noise has been considered as one of the major issue that degrades the performance of character recognition system. Our feed forward network has one input, one hidden and one output layer. The entire recognition system is divided into two sections such as training and recognition section. Both sections include image acquisition, preprocessing and feature extraction. Training and recognition section also include training of the classifier and simulation of the classifier respectively. Preprocessing involves digitization, noise removal, binarization, line segmentation and character extraction. After character extraction, the extracted character matrix is normalized into 12×8 matrix. Then features are extracted from the normalized image matrix which is fed to the network. The network consists of 96 input neurons and 62 output neurons. We train our network by proposed training algorithm in a supervised manner and establish the network. Eventually, we have tested our trained network with more than 10 samples per character and gives 99% accuracy for numeric digits (0∼9), 97% accuracy for capital letters (A∼Z), 96% accuracy for small letters (a∼z) and 93% accuracy for alphanumeric characters by considering inter-class similarity measurement.