Alzheimer's disease (AD) is the most prevalent type of dementia, resulting in gradual memory and cognitive impairment. Radiomic characteristics acquired from brain MRI have shown significant promise as non-invasive indicators for this illness. However, their use for particular brain areas has not yet been investigated. To study Alzheimer's disease (AD), conventional machine learning approaches have evolved from image decomposition methods like principal component analysis to more complicated, non-linear algorithms. Now that the deep learning paradigm has arrived, it's feasible to extract high-level abstract features directly from MRI scans that characterize how data is distributed in low-dimensional manifolds internally. In this paper, we proposed a new Convolutional Neural Network based architecture for classifying Alzheimer's disease. The model is also evaluated using performance measures like precision, recall, f1 score, confusion matrix, etc. The proposed model was evaluated and achieved a classification accuracy of 97% for the classification of 4 different classes, which surpassed almost all cutting-edge technologies.
Today, designing a reliable and efficient step-up converter with high voltage gain for Photovoltaic applications has been a major concern for designers. This work proposes a fully soft switched high step-up converter that combines synchronous Cuk and Boost converters utilizing the shared component approach. The advantages of this converter compared with traditional Cuk and Boost converters are higher voltage gain and zero switching loss. A PSpice simulation was done at 225w power, 24 V input voltage, 150 V output voltage, and 100k Hz switching frequency to verify the proposed converter's functioning and the efficiency 96.56% has been achieved at mentioned operating power. Additionally, a bond graph model is designed to examine the dynamic behavior of the converter as it is presented, taking into account parasitic parameters and the specifications of realistic components. The bond graph model's simulation results have been explored. In the final section of this study, a PID controller employing the state space averaging approach is designed using SISOTOOL in MATLAB. Additionally, a Simulink model of the closed loop proposed converter is designed, and simulation results are reported.
This paper discusses interference cancellation in an Uplink Multi-User (MU) Multiple Input Multiple Output (MIMO) system, in which a Full Duplex (FD) Decode and Forward (DF) relay is used. In order to decode the symbols at the relay, Block Diagonalization (BD) based processors are applied for MU interference cancellation. Moreover, the self-interference signal at the relay is suppressed by using the equivalent relay concept. Minimum Mean Squared Error (MMSE) technique is then adopted at the relay for information extraction. For MU interference cancellation and information extraction at the Base Station (BS), a similar procedure is applied. Simulation results show the superiority of the proposed system to the Amplify and Forward (AF) relaying system. In addition, we provide the appropriate number of antennas and transmit powers for each setup in order to achieve a relatively high sum rate.
Next word prediction for technological devices has become a contentious issue in predictive analytics. Word pre-diction requires a huge amount of data from a vast number of different users, which introduces issues like heterogeneous dataset management and privacy concerns, thus raising the importance of Federated Learning (FL) as it can learn from a large amount of sparsely distributed data while ensuring privacy. Further investigations on related studies have led us to contribute in this area based on the Bangla language due to the minimal appearance of this language in the domain of word prediction based on FL. In consideration of implementing our work, some of the prime issues have been noted where avoiding data poisoning attacks, prioritizing user preferences during the global update, and focusing on fast convergence to reach a high-quality model in a few training rounds have become unavoidable matters of contention. Besides this, generating initial models which can be easily adapted by the current or new users is also an important research problem. Based on these, we have proposed a structure that combines semantic analysis and personalized FL using Model-Agnostic Meta-Learning (MAML) to obtain the desired goals. Extensive evaluations have been done on Non-IID data for both FL and personalized FL with MAML on the next word prediction model, which justifiably achieved 83.05% and 86.23% test accuracy respectively on the server model after ten communication rounds across ten clients. Apart from that, the effective results of semantic approaches show an impressive outcome as intended. Thus, the limitations and the challenges of FL on next word prediction conceptually allowed us to use a variety of algorithms to turn them into advanced ones for further usage.
This study presents a design for a 5G NR (New Radio) Microstrip Patch antenna (MSP A) operating in the n257 band, which is more commonly known as the 28 GHz band. The simulations were used to determine its free space properties. The initial findings were optimistic. The antenna produced a 4.26 GHz wide bandwidth with a substantial amount of gain above 6.8 dBi and 74% efficiency. To analyze its on-body performance, a virtual human torso phantom was created, and simulations were done again by placing the antenna over the torso with a 4 mm gap. The antenna sustained its wide bandwidth with a high gain of over 8 dBi. Even in the on-body scenarios, the antenna's efficiency remained close to 70%. The antenna is designed for the 5G-enabled Body Centric Network (BCN), which may be used in a wide range of smart medical applications.
