The research domain of UWSN (Under Water Sensor Networks) is gaining wide attentionthese days. The sensors are deployed on the sea surface or seabed for monitoring, data collection orsurveillance purpose. Unlike terrestrial networks, the UWSNs use acoustic signals for data communica-tion. The speed and bandwidth of acoustic signals is low compared to radio signals which are used inconventional terrestrial WSNs (Wireless Sensor Networks). As a result, protocol design in UWSN mustbe addressed separately than terrestrial networks. In this work, we have reviewed TDMA (Time Divi-sion Multiple Access) protocols for channel access. As TDMA is a collision-free method, it is preferredover other methods like random access methods or handshaking based medium access protocols. Someopen issues are also proposed as follows: (i) It is found that very less or no work is done in the area ofjoint scheduling and topology formation for UWSNs. (ii) The problem of balancing schedule length ofsink-rooted trees in multiple sinks UWSNs is also unattempted.
The Covid-19 app based on Chest X-rays & CT scans is an integration of Machine learning and android. Our group developed an android application that uses an ML model and predicts if the image captured or uploaded by the user of a Chest X-rays or a CT scan has Covid19. The main tools used for the development of this project were android studio & Firebase for android app development, Jupyter notebook & google Colab as a code editor along with google drive to fetch the data. The main Libraries used for training ML models were Tensorflow, Keras and the method used was Transfer Learning with the help of InceptionV3 pre-trained model on the Imagenet dataset In coping and fighting against COVID-19, the most critical step is to effectively screen and diagnose infected patients. RT-PCR is considered to be a reliable test for detection of coronavirus, but the problem with RT-PCR test and Elisa tests is that they take a lot of time to generate results, Since Covid is highly contagious and spreads through Human-to-Human Interactions it’s Crucial to detect it as early as possible to stop the transmission. Corona is an Infectious disease that affects the Lungs similar to Pneumonia, Deep Learning and Machine Learning models have produced significant results in the past for pneumonia detection in Lungs in this project we tried the same approach where we trained the Deep learning model on more than 13000 images of CT scans & 9000 images for X-rays and used the model to build an android application for users, the application is user friendly and easy to use. Since radiology scanning takes far less time to generate results as compared to the RT-PCR & ELISA tests, we believe our app will be convenient and it'll help people in the early detection of Covid.
The approach adopted to handle hard problems is known as metaheuristics. The problem is considered as hard if discovering the optimal solution for it may not be always possible within the stipulated time. Discovering a single solution to a problem is easy and can be accomplished extremely fast, but finding the best possible solution to the same problem is very long. Optimization algorithms are intended to bridge this gap. The research paper aims at solving the problems for finding the optimal solution for two popular metaheuristic algorithms, GWO (Grey Wolf Optimization) and FF (Firefly) algorithms. Both the metaheuristics algorithms, GWO and FF algorithms are studied and implemented. The two technical features comprised of metaheuristic algorithms are exploration and exploitation. The optimal solution has been evaluated alongside Makespan and Utilization Rate for both GWO and FF algorithms. The lower value of the Makespan and higher Utilization Rate is always desirable. Both the algorithms have been modified via using mathematical functions to enhance the readings concerning performance evaluation parameters. The GWO is been modified via developing a hybrid version comprising GWO and PSO (Particle Swarm Optimization) algorithms denoted as the Hybrid Modified GWOPSO algorithm. The FF algorithm too has been modified and is denoted as a Modified FF algorithm. The conducted modifications have been measured via different performance evaluation parameters. Finally, the fault tolerance factor is considered and the modified versions Hybrid Modified GWOPSO and Modified FF are hybridized to develop a new hybrid algorithm Hybrid GWOFF (Hybrid Grey Wolf Firefly) algorithm and its performance have been evaluated with and without fault tolerance.
Agriculture and allied industries play an important role in the development of our nation. In India more than 55% of people make a living from farming. Crop yields are an essential aspect of every farmer’s day. It depends on many factors like soil quality, seeds, planting practices, humidity, fertilizers and pesticides. Besides all factors, diagnosing soil quality is a fundamental and essential task in farming, as it provides background knowledge of the soil and its physical, chemical and biological prominence. Hence, soil analytics is inevitable that gives information about the present nutrient availability or the need of the nutrients for effective cultivation. It helps to interpret the physico-chemical properties of soil nutrients and to classify the nutrient content as very low, low, medium, high, or very high based on pH values. Thus, predictive analytics based on the soil parameters offer precise and sensible solutions for soil fertility problems and enable suitable decision on crop cultivation. This study attempts to exploit the benchmark classification algorithms from data mining to classify soil samples of Tiruppur district using pH levels. The prediction of pH levels is important to know the nutrients availability in the soil. Classification algorithms like Logistic Regression (LR), Bernoulli Naive Bayes (BNB), Decision Tree (DT), Extra Tree (ET), Random Forest (RF) and K-Nearest Neighbor (KNN) are used to evaluate and predict the pH values. After the comprehensive evaluation, this study determined that the performance of the DT and RF model for pH prediction is high compared to the other algorithms in terms of accuracy. Further, the classifiers performance has improved by possessing feature scaling techniques like normalization and standardization. Results showed that the prediction accuracy of KNN and BNB with feature scaling outperforms the other algorithms.
Image understanding and language generation have always been a difficult task in the field of Artificial Intelligence. Automatic Image Caption Generation is concerned with the task of understanding the image and generating a caption for it. In this paper, we represented our research work that uses the Deep Learning technique to create Punjabi captions for a given image and its associated news document. High-level features of the images are extracted using the pre-trained VGG-19 (Visual Geometry Group) model. These image features are merged with features of news text which are extracted using LSTM (Long Short Term Memory). The proposed model augments keywords from associated news text to generate suitable captions. Using both BLEU scores and human evaluations, we show that the proposed method is successful in generating intelligible and suitable captions.
The communication media has undergone major technological advances in recent decades, from telegraph services to the internet of things. These rapid changes have impacted the routine of humanity and its way of seeing the world in an impactful way. Currently, we are entering the fifth generation (5G) of mobile telephony, whose architecture promises extensive exchanges of messages in a short period of time, which was still difficult to obtain in the networks of the previous generation. Given this new technology, this work proposes to investigate its impacts and possible advances in the area of Internet of Things, an emerging technology that extends mobile networks to so-called smart objects.
Numerous research works has focused on the positioning multiple mobile sinks in the network area to improvise the lifetime of wireless sensor network. The experimental results of the research works clearly indicates that the lifetime of the network will be increased several folds on exploiting multiple mobile sinks in the network rather than static sinks. The proposed paper employs multiple mobile sinks to increase the lifetime of network. The research work focuses on finding the best energy rich locations for placing multiple mobile sinks in a way it delays the expiry of one hop sensor nodes. The paper contributes an Integer Linear Programming (ILP) formulation for the positioning of mobile sinks. In an attempt to find the best location to place the mobile sinks, the paper discretizes the solution space as a replacement for the entire network. The energy rich positions for positioning the mobile sinks are found through the solution of the ILP formulation. As an alternative of running the ILP on a continuous search space, the solution space is discretized and ILP is implemented on the discretized space. Experimental results indicate that the proposed ILP model maximizes the network lifetime to a great extent in comparison with the static sink method.
Clone Attack is the very dangerous attack of all the active attacks of Wireless Sensor Network. In this attack, the adversary capture a sensor node from the network with help of that adversary can create any number clone nodes. Clone nodes are controlled by the adversary, which has all the secret keys. So that clone attack is most harmful threat which is very difficult to detect. In mobile sensor network, the nodes move from one location to another regularly. So it is a complicated task to find the clone nodes which exist in the mobile sensor networks. The proposed Central-node Based Clone Detection(CBCD) protocol detect the clone nodes in efficient manner and within minimum detection time compared with existing EDD and XED protocols. The advantage of proposed CBCD protocol include 1) High clone detection ratio; 2) Minimum communication overhead; 3) Minimum End-to-End delay; 4) Minimum latency time; 5) High throughput so on.
This research paper has been described the fuzzy logic and hybrid fuzzy logic-based systems, which is utilize to create knowledge-based frameworks in chemical engineering. In this proposed study, we have proposed fuzzy logic-based methods such as Fuzzy sets, Fuzzy C-Means (FCM), Subtractive Clustering (SC) and integrated approach of SC and Artificial Neural Network Fuzzy Inference System (SC-ANFIS) for calculating the rate of chemical reaction. After comparing the results of these proposed methods with other exiting methods such as classical fuzzy logic, FCM and subtractive clustering (SC) methods, gives get the better result of SC-ANFIS. The root mean square error (both training and testing data) of SC-ANIFIS is less as compared to existing FCM method.
The people express their sentiments thought the Social Networks, and these sentiments can be measured to offer knowledge about tendencies and influences on Internet. In many scenarios, as in sport games, is important have the insight about what the people are thinking regarding a given player or product. This paper presents a metric based on phrase-level, in which is shown the importance of considering users´ characteristics, such as gender in sport scenario is shown. The calculation made from the word dictionary is adjusted by applying a correction factor based on the user´s profile characteristics. The results showed that considering the user's profile characteristics in the sentiment metric improves the sentiment analysis performance.
Question Answering (QA) system is becoming more popular with the introduction of Virtual Agents and Chatbots. Medium of QA system is generally either text or audio. There are differences between search engine and QA system. Generally searching is based on keyword matching. In case of web search, list of URLs is ranked based on location, user history, search preference etc. Sophisticated algorithms like page-rank is also involved there. On the other hand, QA system does not work on keyword matching primarily. It’s often possible that the query and the best answer have no term or a very small number of terms in common. QA system in English and other popular languages resolves the issues with the help of ontology, WordNet, machine readable dictionary etc. QA system in low resource languages suffers from lack of annotation, absence of WordNet, immature ontology. In this work, QA system in Bengali is developed using supervised learning algorithms. A collection of Bengali literatures, which was developed during TDIL (Technology Development of Indian Languages) project funded by Govt. of India, is used as the repository. Well known classification techniques like ANN, SVM, Naive Bayes and Decision Tree are employed in this work. The system has achieved 84.33% accuracy to return the exact answer. It has achieved 97.13% accuracy to return the string containing correct answer. Unavailability of structured dataset and poor resources were the main challenges for this work. QA system in Indian languages especially Bengali is very much useful not only for chatbots or virtual agents but also for the e Governance and mobile governance in West Bengal and Bangladesh. QA system in mother tongue gives opportunity to more number citizens to interact with the administration. Though the system is designed aiming towards Bengali language but it can be tuned to work for any language with minimum modification.
Classification has gained popularity in research community due to its wide variety of applications in different fields. The complexity in extraction of information from images and use of this information for classification task has made image classification a tedious job. The Proposed method of classification uses SIFT algorithm for extracting rotation and scale invariant features. Using the concept of Deep Neural Network sparse coding technique is used for generating codebooks from the extracted features. Sparse codes give an intermediate representation between local codes and dense codes. These codes have capability to extract information at different levels and with varying amount according to the type of input. Instead of using L1 norm another popular regularizer is used in this paper, which maintains group sparsity for non-overlapping groups. Pooling operation is applied on the sparse coded features. Bat algorithm is used on these pooled features for classification of medical images. Experimental results prove the fruitfulness of our proposed method in medical image classification.
This paper applies the basic compartmental SIR model to explain the transmission dynamics of an pandemic like COVID-19. COVID-19 has created an alarming situation for the whole human kind. Researchers around the world are trying to demystify the prevailed global health emergency and come up with some effective solution. We attempt to implement the basic SIR model to show the spread of the disease. Further, we compared the level of spread of the disease on basis of basic reproduction number. Basic reproduction number is very important measure in order to quantify the proliferation of an infectious illness over the time. One of main assumptions of the model is a recovered person generates immunity towards the disease. Then we have suggested the ways to control the outbreak and normalise the prevailing situation.
Mangoes, the king of fruits is globally exported and locally consumed on a large scale. During exports and local marketing, delivering good quality fruits and satisfying certain pre-defined standards is important. This post-harvest operation of quality checking, known as mango grading is usually performed manually. But manual grading can be in-consistent, erroneous and labor-intensive. A computer vision based grading solution will result in consistent and accurate sorting. Such a mango grading system based on external parameters namely ripeness, size, shape, defects was developed in this research work. Image processing techniques were applied to extract the color, geometric and shape related features. These features were further utilized by pre-trained random forest classifiers to determine the mango ripeness (unripe/mid-ripe/ripe), size (small/medium/large) and shape (well-formed/deformed) category. K-means clustering was applied for defect segmentation to determine the mango defect category as (non-defective/mid-defective/completely-defective). Final grading was performed using a grading formula that combines the parameter specific quality scores assigned, according to predicted categories. Ripeness, size and shape classification performed on a created dataset of Dashehari mangoes achieved a test accuracy of 100%, 98.19% and 99.20% respectively. Formula based integrated grading could grade mangoes with 88.88% accuracy.
The fact of appearing of the handheld devices offers forthright entree to the internet and social networking sites. Sentiment analysis and opinion mining is the study of sentiments or opinions shared by different users in social networking sites like, Twitter, Facebook, Reddit, Instagram etc., on diverse social phenomena. In this article, sentiment analysis of different tweets on the ongoing epidemic COVID-19, Corona virus disease is performed. COVID-19 is declared as epidemic by the World Health Organization (WHO) in the mid of March 2020. The statistical and machine learning based analyses are implemented on 40,000 tweets, which were collected in two different mutually exclusive time frames. Tweets are collected from Twitter site between 3/07/2020 to 11/07/2020 and 01/08/2020 to 06/08/2020, using Tweepy python library. Various Python based libraries are applied for data acquisition, data pre-processing and data analysis processes. As a data pre-processing phase initially sentences are cleaned. Then by calculating the polarity and subjectivity measures tweets are categorized into three groups (viz., negative, neutral, and positive}). Thereafter, in the later phase by applying the Term frequency-inverse document frequency (TF-IDF) feature extraction scheme with the help of uni-gram, bi-gram, and tri-gram techniques different features are extracted to prepare the datasets to feed it into the prediction models. 70% of the datasets are used to train Gaussian Naive Bayes (G-NB), Bernoulli's Naive Bayes (B-NB), Random forest (RF), and Support vector machine (SVM) classifiers to generate different prediction models. Finally, 30% of the data is tested on those learning models. Experimental results suggest that RF and B-NB models performed better than the other two classifier models. The execution computational cost of SVM is very high.
The domain of Medical images is escalating with the trend of digital image based diagnosis and treatment. When talking about Tumors and Cancers, medical images play significant role to identify the affected area with maximum precision. In this paper, Cross Bilateral Filter is used to focus on retaining the edges. The Muti-Modality medical images are firstly decomposed using cross Bilateral Filter and Wavelets (in parallel), followed by fusion of detailed parts by Fuzzy Logic Infererence System having 25 set of rules and approximate parts are fused with average rule. Lastly, the reconstruction is done to obtain the final fused image. To compare the results quantitatively as well as qualitatively, MR-T1, MR-T2 images when fused with proposed method, attained higher values for Standard Deviation (SD), Fusion Symmetry (FS), Correlation Coefficient (CC) and QAB/F and lower value of NAB/F.
Age-related Macular Degeneration (AMD) is a common eye condition that leadsto about 8% of all blindness worldwide. It is leading cause of vision loss among people age 50and older. Late stage AMD is divided into dry AMD and wet AMD. Drusen and ChoroidalNeovascularization(CNV) are the major causes of dry and wet AMD. OCTA is recent noninvasive and safer technique that provide detailed internal and external structural informationof CNVs. Detecting and analysing CNV is very effective for proper treatment and assessmentof wet AMD. The proposed work aim to provide automated method and application to detectand analyse CNV region in OCTA images. Proposed methodology is divided into two moduleswhich include CNV segmentation step based on connected component labelling algorithm andother one is CNV quantification step based on Otsu thresholding. This provides two importantquantification measures namely CNV area and CNV vessel density. Performance of proposedsystem was measured using various parameters and was found better when compared with stateof art methods. The jaccard similarity score was 0.9379 ± 0.023 and false negative rate was0.0026 ± 0.003.
Machine translation systems for Indian languages such as Bengali and others are commonly found. Classical machine translation systems involving Bengali are available for tourism, agriculture, medical and other domains. The performance of these systems are restrained by the linguistic knowledge, that are used to develop the rules. In the recent past, notable results have been achieved by systems using neural machine translation. Well known organizations like Google and Microsoft have started using NMT models. In this paper, we explore the design and implementation of an unexplored domain in Bengali, the aviation domain. It is implemented using a neural machine translation model. In order to implement it, we have used English to Bengali parallel corpus for the aviation domain which was developed specifically for this implementation. The corpus is a unique one with large number of aviation specific OOV words and phraseologies included in it. We have used the already developed aviation preprocessing tool, E-dictionary and transliteration tool for creation of the corpus and system. Ultimately we get the output model which generates our machine translated output file in Bengali. We then apply the aviation phraseology converter and transliteration tool on the output to get a post-processed output. The two versions of the output are compared using n-gram BLEU score. The results ultimately demonstrate that NMT output with the post processing exhibits better results.
Clustering of quantitative survey data is done in-order to identify the divergent and dominant behaviors of the respondents. It is intended to explore the general tendencies of the respondent groups. Popular clustering methods working on value based similarity are inappropriate for survey data due to its distinct properties. Since marking patterns in survey data represents respondent’s behavior, hence separating the responses on the basis of marking patterns is an effective approach to identify the dominant behaviors. Thus, in this paper, we propose a specialized clustering method for quantitative survey data that combines the features of both, value based as well as pattern based approaches in order to obtain meaningful results. The proposed method does not require presetting of the clustering parameters while it makes use of group labels for selecting features and guiding the centroids at positions, which best describe divergent marking habits. We apply the proposed method over an educational survey dataset and compare its results with K-means clustering with respect to the benchmark stakeholder theory. Comparison results show that the proposed method is more appropriate for quantitative survey data.
