MRI Brain Cancer Diagnosis Approach Using Gabor Filter and Support Vector Machine

Cervical cancer is the fourth most familiar root of cancer correlative ruination. Cervical cancer is the basis of HPV (Human Papilloma Virus) which generate irregular growth of cells in the cervix part. Cervical cancer is one of the primary stimulation of women's mortality globally and it is very challenging because this cancer occurs without any symptoms. So, Accurate and timed detection of cancer cells can save lives. Here we are using deep learning architecture for classification. For segmentation, we use Mask R-CNN and classify using VGG-Network because it gives high accuracy. ResNet-10 makes total actuate of structural region with previous learning and accustomed to be a determination for Mask R-CNN. Here Pap test datasets are taken. In the bisection stage, when Mask R-CNN is practiced on the total cervical unit, it segments each cell with high accuracy. In the classifying step, Visual Geometry Graph Network is applied and yields an accuracy of 89% with low standard deviation.

Section: Literature Reviewmentioning

confidence: 67%

Section: Introductionmentioning

confidence: 99%

Cervical Cancer Classification

Deepa¹

2020

“…Tree type structure is shaped in progressive put together strategies based on respect to the chain of command. Various leveled bundling starts by viewing each recognition as an alternate gathering [11]. By then, it, again and again, executes the going with two phases: a) Recognizing the two packs that are closest together.…”

Section: Hierarchical Clusteringmentioning

confidence: 99%

Load Balancing Analyzer: A Recommendation System using Machine Learning

Tata¹

2020

The Parallel and distributed systems focus on the concept of load balancing. A tangle is split into a hard and fast number of processors that are to be executed in parallel. However, there may be a state that some processors will complete their tasks before other processes and reach idle state as the work is unevenly divided or perhaps some processors complete before the others. Ideally, we like all the processors to have the minimum wait time. Achieving the above goal by spreading the tasks is eventually named as load balancing. In the paper, we stress on clustering techniques. The basis of the join-idle-queue algorithm is seen by using clustering. The technique of load balancing uses Support Vector Machine (SVM) and clustering techniques (K-means, Hierarchical). A comparative study of the above techniques is done utilizing load balancing.

“…To carry out classification, the outputs can be used as inputs to a SVM or RF or, The SVM classification model is generated from the training process with the training data. The main concept of SVM [35] is using hyper-planes for defining decision boundaries that separate between data points of different classes. Optimal hyper-plane is the hyper-plane with the maximum if the classifier needs input quantity, which behaves like probabilities, the output phase can be a soft max function as given in Eq.…”

Section: 1deep Convolutional Neural Networkmentioning

confidence: 99%

Meta-Heuristic based Trained Deep Convolutional Neural Network for Crop/Weed Classification

Gomathi¹

2020

Computer vision and camera sensors are hopeful technologies for capturing information and processing to facilitate autonomous cultivation with machine learning. Nowadays, field robots are widely utilized, which autonomously navigates in fields tasks for advanced developments. However, manual activities are also required for certain needs, for instance, controlling weed in organic carrot farming is carried out manually and it is essential to evade considerable crop yield loss. This paper makes an attempt to introduce a new automatic crop/weed classification under three major phases: (i) Pre-processing (ii) Feature Extraction and (iii) Classification. Initially, the images are converted to greyscale images under pre-processing stage. Further, from the pre-processed image, the features like "Grey Level Co-occurrence Matrix (GLCM) and Grey Level Runlength Matrix (GLRM) based texture features" are extracted. Finally, the classification is done by the hybrid classifiers, where both the Deep Convolutional Neural Network (DCNN) and Neural Network (NN) is incorporated. Finally, both the classified outputs are ORed to get the final classification output. Moreover, in order to enhance the performance of proposed work, it is planned to tune the hidden neurons of NN optimally via a new improved Moth Search Algorithm (MSA) and is named as Moth Search with newStep Length evaluation (MS-SL). Finally, the performance of proposed work is evaluated over other state-of-the-art models with respect to certain performance measures.