Subrina Akter

Bio: Subrina Akter is an academic researcher from International Islamic University, Chittagong. The author has contributed to research in topics: Moving average & Earthquake prediction. The author has an hindex of 3, co-authored 8 publications receiving 30 citations.

Significant HOG-Histogram of Oriented Gradient Feature Selection for Human Detection

Abstract: Detecting human efficiently is an important field of research and has many applications such as intelligent vehicle, robotics and video surveillance. Histogram of Oriented Gradient (HOG) is one of the eminent algorithms for human shape detection. HOG features are extracted from all location of a dense grid on an image region and use linear Support Vector Machine (SVM) to classify the combined features. Although HOG gives an accurate description of the contour of human body, it requires a large computational time. We studied the fundamental idea and consider features that have high percentage to contain edge. In this proposed method we used difference of Gaussian to obtain the edge percentage of each feature. Then a threshold is used to remove features with low edge percentage. Selected features then classified using linear SVM. Experiments on INRIA dataset demonstrate that the proposed method not only reduce the dimension of the HOG features but also outperforms. General Terms Pattern Recognition, Image Processing.

An Expert System to Detect Uterine Cancer under Uncertainty

Abstract: Uterine cancer is one of the conspicuous cancers for women in both developed and developing countries including Bangladesh. Now, in the world it is the sixth most common cancer among women and fourteenth most common cancer overall. The high occurrence of uterine cancer in women has increased significantly in the last years. This involves many factors to be measured and evaluated, which are related to the signs and symptoms of this disease. These factors usually expressed in quantitative and qualitative ways. In addition, a hierarchical relationship exists among these factors. Since qualitative factors cannot be measured in a quantitative way, resulting various types of uncertainties such as incompleteness, vagueness, imprecision. Therefore, it is necessary to address the issue of uncertainty by using appropriate methodology; otherwise, the conclusion to detect uterine cancer will become inaccurate. There exist many systems to address the issue presented in this paper. However, none of them is able to address the issue of uncertainty. Therefore, this paper demonstrates the application of a novel method, named belief rule-based inference methodology-RIMER, which is capable of addressing the uncertainties in both clinical domain knowledge and clinical data. This paper reports the development of a Belief Rule Based Expert System (BRBES) using RIMER approach, which is capable of detecting the presence of uterine cancer by taking account of signs and symptoms. The system has been validated by using real patient data and it has been observed that the results generated by the BRBES are more reliable than the manual system usually carried out by a physician.

Genetic Algorithm with Updated Multipoint Crossover Technique and its Application to TSP

Abstract: Genetic Algorithm (GA) is a promising method for optimizing the NP-hard problem especially the Travelling Salesman Problem (TSP). The reason of its popularity is for the ability to gain an ideal approximation in time. GA is usually based on the three artisans namely selection, reproduction and metamorphosis. The principal target of using GA is to determine the lowest total cost to travel all the nodes optimally. Consequently, this study introduces a novel crossover operator which optimizes the solution to the TSP. The suggested method started with two randomly selected parents and new offsprings have been generated by comparing cost. The overall methods, as well as the experimental outcomes, have also depicted here. The paper concludes that the newly introduced crossover operator outperforms various cross-over operators. It produced better result while experimenting on a set of instances from TSPLIB dataset.

Travel Time Prediction using Support Vector Machine(SVM) and Weighted Moving Average(WMA)

Abstract: Travel Time forecasting in highway system has appeared a vital issue for delivering travellers exact guidance about choosing their route. In this paper, a new method for forecasting travel time from historical traffic data using SVM and WMA is depicted. The proposed work has been divided into two parts: First one is classifying Travel Time depending on the traffic condition or velocity class using Support Vector Machine(SVM) and Second one is predicting Travel Time using modified Weighted Moving Average(WMA) method with a modified equation where the WMA method will be applied on the support vectors whose are generated after classifying time using travel Multi class all versus all SVM. Considering the same historical traffic data, the outcomes of previous methods also compare with the outcome of propose method. In this case, previous methods include Successive Moving Average (SMA), Chain Average (CA), and Artificial Neural Network(ANN). The comparison result proofs the better performance of SVM and WMA method than the previous methods. Keywords— Intelligent Transportation System(ITS); Support Vector Machine(SVM);Weighted Moving Average(WMA); Successive Moving Average (SMA); Chain Average (CA); Artificial Neural Network(ANN); Travel Time Prediction.

Emperical Computation of Solar Radiation and Determination of Regression Coefficients for Khulna City

Abstract: This study is accomplished to calculate global, diffuse and direct solar radiation empirically on a horizontal surface for the divisional district “Khulna” in Bangladesh (latitude 22o47΄N and longitude 89o34΄E) as well as to predict correlations for it by using several meteorological data for 32 years between 1980 and 2012. The global radiation is found to be maximum in the month of April and minimum in the month of December here. The estimated values of the Angstrom’s regression constants a and b are 0.2388 and 0.5228 respectively. The other regression constants were also computed and the correlations proposed for Khulna can be used in future for the estimation of global, diffuse and direct solar radiation if the meteorological parameters remain available.

Computer aided diagnosis of acute abdominal pain

Abstract: a correct decision in 84% and a combined bad diagnostic and management error rate of 3-2%.' We used a different approach, which requires the surgeon to categorise patients into management pathways at the time ofadmission (definitely needs operation, definitely does not require operation, uncertain). Laparoscopy was done in the uncertain group. We consider that this management approach to acute abdominal pain is more appropriate than a system based on diagnostic accuracy. We emphasise the point made by the authors that improved results during the study period are due not to the use of a computer but to accurate collection ofinformation and feedback of results to the doctors concerned. Improvement in this important area stems from interest, analysis, and feedback. Computers are one way of achieving this, rigorous analysis of decision making is another. We prefer the latter.

A framework of human detection and action recognition based on uniform segmentation and combination of Euclidean distance and joint entropy-based features selection

Abstract: Human activity monitoring in the video sequences is an intriguing computer vision domain which incorporates colossal applications, e.g., surveillance systems, human-computer interaction, and traffic control systems. In this research, our primary focus is in proposing a hybrid strategy for efficient classification of human activities from a given video sequence. The proposed method integrates four major steps: (a) segment the moving objects by fusing novel uniform segmentation and expectation maximization, (b) extract a new set of fused features using local binary patterns with histogram oriented gradient and Harlick features, (c) feature selection by novel Euclidean distance and joint entropy-PCA-based method, and (d) feature classification using multi-class support vector machine. The three benchmark datasets (MIT, CAVIAR, and BMW-10) are used for training the classifier for human classification; and for testing, we utilized multi-camera pedestrian videos along with MSR Action dataset, INRIA, and CASIA dataset. Additionally, the results are also validated using dataset recorded by our research group. For action recognition, four publicly available datasets are selected such as Weizmann, KTH, UIUC, and Muhavi to achieve recognition rates of 95.80, 99.30, 99, and 99.40%, respectively, which confirm the authenticity of our proposed work. Promising results are achieved in terms of greater precision compared to existing techniques.

SVM-based hybrid approach for corridor-level travel-time estimation

Abstract: The objective of this study is to develop an accurate model for corridor-level travel-time estimation. Different approaches, such as k-nearest neighbour (k-NN), gradient boosting decision tree (GBDT) and support vector machines (SVMs), were used in this study. Further, this study also developed a hybrid model combining a data-driven approach (SVM) and a model-based approach [particle filter (PF)] for corridor-level travel-time estimation. Both static and dynamic parameters, such as road geometry, intersection length, location information from Global Positioning System devices, dwell time etc. were used as influential factors for modelling. The proposed algorithm was tested on a study corridor of length 59.48 km, in the arterials of Mumbai, India. The data was collected using a probe-vehicle technique for five days during the morning peak period (from 8.00 am to 11.00 am) for two modes (car and bus). The mean absolute percentage error values obtained for the hybrid model for the two modes were: 9.96 (car) and 11.24 (bus). The performance of the proposed hybrid (SVM-PF) algorithm showed a clear improvement in accuracy in comparison to existing standard methods such as k-NN, GBDT and SVM.

Impact of Fuzziness Measures on the Performance of Semi-supervised Learning

Abstract: Usage of fuzziness in the study of semi-supervised learning is relatively new. In this study, the divide-and-conquer strategy is used to investigate the performance of semi-supervised learning. To this end, testing dataset is divided into three categories, namely low, medium and high-fuzzy samples based on the magnitude of fuzziness of each sample. It is experimentally confirmed that if the low-fuzzy samples are added from the testing dataset to the original training dataset and the model is retrained, then the accuracy can be improved. To measure the amount of fuzziness of each sample, four different fuzziness measuring models are used in this study. Experimental results support that improvement of accuracy is dependent on which fuzziness measuring model is used to measure the fuzziness of each sample. Wilcoxon signed-rank test shows that choosing a specific fuzziness measuring model is significant or not. Finally, from the Wilcoxon signed-rank test, the best model is chosen, which can be used along with semi-supervised learning to improve its performance.