The communication network based on the optical system requires more precise and efficient devices and equipment. Optical ring resonator is a versatile device used as a filter, delay line, add/drop multiplexer, switch, sensor and analyzer etc. We modelled the new symmetrical quadrature optical ring resonator (SQORR) and proposed four different topologies architecture for the multiple optical ring resonator. The performance of the proposed symmetrical quadrature optical ring resonator is assessed using regression based machine learning (ML) approach utilizing an Artificial neural network (ANN) with various checks are done to validate the approximations like error histogram, time series response, error autocorrelation, input-error correlation and mean-square-error (MSE). The best validation performance is 8.8486e-9 stopped at 103 for rhombus topology, 8.347e-9 stopped at 242 epochs, 8.8486e-9 stopped at 311 epochs for horizontal topology and 5.8012e-09 stopped at 373 epochs for square topology. The significance of the work is to increase FSR (free spectral range) by adding rings and wide FSR achieved 299.79 THz (maximum) and throughput of 0.027 W/m for rhombus topology and other performance parameters of the optical ring resonator using symmetrical rings. Such structure is small in the size and easy to fabricate. Even by the same number of rings using different topologies structure, we can vary the performance parameters as per the different applications.

Brain tumor data have recently been analyzed using deep learning techniques. Segmentation and classification of brain tumors and distinguishing tumorous and nontumorous cells are fascinating when it comes to distinguishing brain cell with tumorous and without tumorous and differentiate the tumorous cells to find their class label. For this purpose, segmentation is an appropriate method for classifying the brain image and it is commonly employed by researchers. To achieve accurate classification, it is necessary to begin with the extraction of relevant features. In this work, the probabilistic Fuzzy C-means (FCM) algorithm, is utilized to further refine the segmentation process. This analysis makes it possible to distinguish the regions of interest for magnetic resonance imaging (MRI) scan of the brain revealed, which provides a framework for reducing the dimensionality of MRI brain image. Local directional pattern (LDP) is applied to the segments after they have been segmented to extract the significant regions of features that have been identified by the segmentation method. Next to deep belief network, the features are provided, which determines whether the images are normal or abnormal, and whether MRI can be used to detect or rule out the presence of tumors. Experimentation is conducted with the help of the proposed method and brain tumor segmentation database; the accuracy has been assessed in relation to the highest percentage of 95.78% is obtained.