Ghani Khan Choudhury Institute of Engineering & Technology

This paper deals with Digital Learning and Training Methodology (DLTM) or Pedagogy (DLTP) in Study of Health Monitoring in RESCO. In a closed room teaching learning process its practically difficult to establish set up for grid connected power system and setup for condition monitoring in all academic institute which can be accessed for all time even from remote end. To overcome this limitation, RESCO model of solar power system connected with grid has been considered. Various health parameters of the system have been captured and stored in cloud from data capturing and transmission unit. Data are made accessible for study by remote end learners as well as data may be accessed from research laboratory. All information is made available for use at all time and can be accessed by computers and smart phone. If needed, researcher can interact with the system. This type of digital learning and training methodology is helpful to provide information, skill and knowledge to big number of learners of any age group as required.

We establish several numerical radius inequalities for bounded linear operators and the product of operators through Orlicz functions. Many inequalities are consequences of using specific Orlicz functions. In particular, it is shown that omega(T) <= n root log [1/2(n-1)e(w)(T-n)+ (1-1/2(n-1))e(parallel to T parallel to n)], <= parallel to T parallel to, for every positive integer n > 1.

The enhancement of energy charging has become a necessity with the increasing popularity of electric vehicles. A solar photovoltaic-based energy system was designed using a incremental conductance based maximum power point tracking technique. The purpose of this project is to design and model a photovoltaic-based energy charger for an electric vehicle that ensures a high rate of performance. The DC-DC converter was used to step down the voltage from the photovoltaic panel to the battery with set ratings connected to an electric vehicle. Solar PV panel and DC-DC converter components were sized according to the required power from the battery for charging. The simulation of the circuit was done using MATLAB/Simulink in which the MPPT algorithm was implemented to extract maximum power from the PV under different irradiation levels. The waveforms of voltage, current, and power were obtained from the results which proved effective performance with enhanced green mobility.

We obtain several Berezin number inequalities for the Kronecker product A⊗ B , where A is an n× n complex matrix and B is a bounded linear operator on a reproducing kernel Hilbert space. These results generalize and improve some existing results.

Early detection of oral cancer is very crucial to increase the survival rate. Deep learning models like Convolutional Neural Networks (CNN) have shown promising results in analysing medical images. However, their high number of connections and floating-point operations make them expensive to run. On the other hand, Spiking Neural Networks (SNN) are a new way of doing neuromorphic computing that mimic the brain's communication mechanisms through event-driven spike communication. This sparse computing approach has shown improved computational efficiency compared to traditional neural networks. In this work, a comparative study is proposed between conventional CNN models and neuromorphic SNN models in classifying oral cancer. Four CNN models, including VGG16, DenseNet121, InceptionV3, and MobileNetV2, are evaluated. In addition, two SNNs, including one based on the Leaky Integrate and Fire neuron and another one based on Hodgkin Huxley model are also inspected. These models are evaluated on three independent oral cancer datasets containing clinical and histopathological images. The results show that CNN models have high peak accuracy such as a complexity of 15.3 billion FLOPs for VGG16. On the other hand, SNN models have competitive results those reduce unnecessary executions and thus results in less floating-point operations such as 77 million FLOPs for Leaky Integrate and Fire Neuron based approaches. The proposed work is conducted based on software based simulated environment. Actual deployment on neuromorphic based hardware is considered as a significant future work.