International Conference on Technology, Engineering, Science and Maths – Conference Proceedings

Conference Proceedings – Online:

1] CERVICAL CANCER DIAGNOSIS THROUGH ARTIFICIAL INTELLIGENCE

Author: Ajni K Ajai – Trivandrum, Kerala, India

Abstract:

Cervical cancer is the second most common cancer among the females after breast cancer. It is estimated that over a million women worldwide currently affected with cervical cancer. World Health Organization estimates every year 1,22,844 women are diagnosed with cervical cancer and 67,477 women die from the disease. There are several tests that can effectively detect Pre-cancer. Artificial intelligence (AI) concepts, techniques, tools have been utilized in medical applications in improving their effectiveness, productivity and consistency. The precision in distinguishing a cancerous structure from a benign structure has the potential to immediately improve health outcomes in one of our more pressing diseases. Automating the cancer diagnosis process can play a very significant role in reducing the number of cancer diagnosis is semiautomatic and is prone to human error and time consuming. A computer system that performs automatic grading can assist pathologists by providing second opinions, reducing their workload, and altering them to cases that requires closer attention, allowing them to focus on diagnosis and prognosis. This paper discussed the recent advances and future perspectives in relation to cervical cancer detection.

Keywords- Cervical cancer, Machine learning, Deep learning, Natural Language Processing.

2] AN EFFICIENT SECURED ID-BASED AGGREGATE SIGNATURE SCHEME FOR WIRELESS SENSOR NETWORKS

Author(s): S. Ninisha Nels and J. Amar Pratap Singh – Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Kumaracoil, Thuckalay, Tamil Nadu, India

Abstract:

Data Aggregation is a suitable technique of reducing the energy consumption of sensor nodes in wireless sensor networks (WSN’s) for affording secure and efficient big data aggregation. Data aggregation technique is considered as a holy grail to reduce energy conception for WSN. Due to the limited resources of sensor nodes, in terms of computation, memory and battery power an energy saved data aggregation methods should be designed in WSNs. To reduce the energy cost of data collection, data processing and data transmission an ID-based aggregate signature scheme is used. The ultimate goal of this paper is to design and model a security-based data aggregation model in Wireless sensor networks (WSNs), which will be simulated through deploying the nodes in the environment for sensing and collecting the data. Novel prediction-based mechanism is modelled using the tri-model, least mean-square (LMS)-digital filter is devised for prediction and the subsequent data aggregation. An ID-based aggregate signature scheme for WSNs, Aggregation based on mean-square (LMS) filter is proposed for predicting the subsequent data aggregation.

Keywords: WSN, data aggregation, sensors, network, nodes

3] Survey of Acute Lymphoblastic Leukemia Classification methods using Blood Smear Microscopic Images

Author(s): G. Mercy Bai – Research Scholar, P.Venkadesh –  Assistant Professor, Department of Computer Science and Engineering Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Kumarcoil, TamilNadu.

Abstract:

Rapid increase in the immature lymphocytic cells leads to Acute Lymphoblastic Leukemia (ALL), which is a type of blood cancer. The challenging task in the classification of ALL is the effective segmentation and the classification of the leukocytes using the Blood Smear Microscopic Images. This survey reviews the research works on the ALL Classification methods, research gaps and the future scope. For the literature review, 20 research papers based on the ALL classification are taken into consideration. The research papers are categorized into Machine learning classifiers, Ensemble classifiers, Deep learning classifiers and so on. The challenges and the research gaps faced during the classification of ALL are elaborated. The result and analysis of the ALL Classification methods are done based on the performance metrics, year of publication and the accuracy range. From the analysis, it is concluded that most of the research works are published in the year 2018. The most commonly used performance metrics is accuracy and the accuracy range for most of the ALL Classification methods ranges from 90% to 94%.

Keywords: Acute Lymphoblastic Leukemia, Blood Smear Microscopic Images, leukocytes, image processing techniques, segmentation.

4] PERFORMANCE ANALYSIS OF DIMENSIONALITY REDUCTION BASED ON FEATURE SELECTION

Author(s): Anju A J, Research Scholar : J E Judith, Associate Professor, Department of Computer Science and Engineering, Noorul Islam Centre for Higher Education, Kumaracoil, India.

Abstract:

In software defect classification problems, there are often too many factors on the basis of which the final classification is done. These factors are basically variables called features. The higher the number of features, the harder it gets to visualize the training set and then work on it. Sometimes, most of these features are correlated, and hence redundant. This is where dimensionality reduction algorithms come into play. When dealing with high dimensional data, it is often useful to reduce the dimensionality by projecting the data to a lower dimensional subspace which captures the “essence” of the data. Dimensionality reduction is the process of reducing the number of random variables under consideration, by obtaining a set of principal variables. it can be divided into feature selection and feature extraction. the result and analysis shows that the framework has reduced total dimensionality of dataset. To select the optimal set of feature space, a novel Quantum Theory based PSO technique is utilized. The reason behind improving the default PSO is that to improve the performance of low convergence rate. And the advancement of PSO is done by improving performance analysis of traditional PSOs is the application of quantum theory to detect particle behaviour. the focus of this paper is towards to dimensionality reduction using the combination of PCA and QPSO.

Keywords: Feature selection, PCA,LDA,PSO,QPSO

5] AUTOMATIC SEGMENTATION AND CLASSIFICATION OF BRAIN TUMORS ON PRE-OPERATIVE AND POST-OPERATIVE MRI USING DEEP LEARNING

Author(s): K.V. Shiny, Research Scholar: N.Sugitha, Associate Professor, Department of Information Technology, Noorul Islam Centre for Higher Education, Kumaracoil, India.

Abstract:

The main objective of this research is to detect the brain tumor from MRI and then have to segment and to classify all the abnormalities in the brain. It is a challenging task to detect and segment the tumor tissues and other tissues from brain. The MRI is initially fed into the pre-processing system and is then segmented using Region Growing segmentation algorithm. This will produce the segmented area and is then forwarded for classification. In the classification step the newly implemented Bir-Cat optimization algorithm is applied and is a deep learning concept that uses Deep Belief Network for training the neural network. The Bir-Cat algorithm is the combination of Bird-Swarm algorithm and Cat-Swarm algorithm. This will give the classified tumor tissues and also classify the different types of tissues or abnormalities in brain tumor. The extended concept is the post-operative brain tumor segmentation and classification which includes all the above image processing steps that was done for pre-operative MRI. Finally both the segmented output of pre-operative and post-operative MRI was compared to find out the pixel changes and so that it helps in finding the emerging tumor after surgery and also the success rate of surgery.

Keywords— Brain tumor segmentation, post-operative MRI, MR Image, region growing, necrotic tissue segmentation, enhancing cell, radio surgery, radiotherapy.

6] SYBIL ATTACK AND CORRUPTED RSU DETECTION USING EWCA AND WOA-AODV TECHNIQUES

Abstract:

Vehicular Ad-Hoc Networks (VANET) is part of intelligent networks which enables transferring data among the moving vehicles to avoid traffic disasters and also offers a comfortable journey along with traffic safety. However, the network can be attacked at any time with the Sybil Attacks (SA). Though existing methodologies have conducted research on detecting the SA, it doesn’t offer an effectual outcome. In order to trounce such issues, this research methodology detects SA as well as Road Side Unit (RSU) corruption detection on the VANET. Primarily, the entire vehicles are registered into the Trusted Authority (TA). After that, TA renders the location certificate utilizing RSU. Next, RSU checks whether the above ‘1’ location is found for a node, if yes, then the node is isolated, or else, it is considered for the communication. Then, the Enhanced weight based clustering (EWCA) algorithm takes care of the cluster formation and Cluster Head (CH) selection. Next, the whale Optimization-Ad Hoc On-Demands Distances Vector (WOA-AODV) algorithm initializes the communication process. After that, the direct trust is computed; if the reply count is above 1, then the indirect trust is computed and the node is amassed on the blacklist, or else, the node is incorporated in the communication procedure.