Interpreting electroencephalography signals and the abnormality of the signals can help to find the specific pattern for specific diseases like depression. A Spiking Neural Network is a machine learning approach that emphasizes the data value and manipulates the value to find the particular signal feature. Finding the specific abnormal features of electroencephalography signals can help to detect depression patients. Since a vast number of individuals are suffering from depression and the treatment of depression is possible by detecting depression patients earlier, different deep learning and conventional machine learning approaches were proposed. But speed, accuracy, and reality with less time and space complexity are essential factors in detecting depression patients in our society. We have proposed a leaky integrate and fire spiking neural network model for interpreting the electroencephalography signals of depression patients. The electroencephalography signals of a sixty-channel dataset of 121 subjects are taken for the experiment where frequency for each channel of a subject is recorded for 2 mins in 2-second time intervals, and the dataset contains 4,35,600 data with 121 instances and 3600 attributes. A leaky integrate and fire model is applied to the electroencephalography signals to find the spike sequences and potentials. Then, a three-layered neural network approach is stacked to generate a classifier. The performance of the classifier is shown to be approximately 98% accuracy. Generating a noble classifier and implementing it with a mask of metal disk benefited society for easily and quickly detecting a depression patient, and corresponding treatment can be started. Besides, more experiments are needed on different and more depression datasets with spiking neural network models to identify depression patients and finalize a robotic classifier.
In this work, a plasmonic sensor using copper (Cu) with palladium (Pd) and hexagonal boron nitride (h-BN) (structure: SF11/Cu/Pd/h-BN/SM) based on surface plasmon resonance (SPR) for the detection of female reproductive hormones (progesterone and estradiol). A comparative study is performed between various layers of Cu and Cu/h-BN, which obtained performance parameters for the proposed structure. The reproductive hormones in females have a long refractive index (RI) between 1.3333 and 1.59606. The proposed sensor for progesterone and estradiol of reproductive hormones in females attains the maximal sensitivity of 97.24°/RIU and 79.4°/RIU. After an analysis, the proposed sensor is found to have a figure of merit (FoM) of 192.9/RIU and 189.29/RIU for the same analyte. Moreover, the penetration depth (PD) of 142.55nm for progesterone was measured. Using Cu/Pd/h-BN/SM as a new type of supporting material with enhanced biosensing activity for the detection of long-range RI of sensing medium (SM), we propose a plasmonic sensor with potential applications.
Purpose: Teaching a robotic arm can be unintuitive and challenging due to the disparity between natural human limb movements and robotic manipulator control. Substantial training is often required to effectively operate such systems. This work presents the development of an intuitive, easy-to-use interface that maps human palm gestures to control the movement of a 5-axis robot, the Dexter ER2. Methods: The proposed vision-based system enables the robotic arms to autonomously follow hand gestures through a three-stage methodology: gesture detection, gesture tracking, and robotic manipulator control. A deep learning-based model is employed for accurate gesture recognition, allowing the robot to mimic the controller’s limb movements with minimal cognitive load. For trajectory computation and joint movement generation, an adaptive gradient descent-based inverse kinematic solver is implemented, enabling efficient and smooth convergence of joint angles corresponding to the desired end-effector positions. Results and Conclusion: The interface was successfully implemented, enabling the robot to follow gesture commands and reach target positions. The proposed vision-based gesture-driven control scheme of the Dexter ER2 robot arm is validated through both simulation and experimental studies. The system employed serial communication, ensuring a continuous and lossless data stream between the controller and the computer. A detailed comparative analysis is also presented to highlight the advantages of the proposed approach. Future work includes enhancing visual appeal with a graphical interface and addressing the latency issue in real-time control.
This paper presents the mechanical design of a 70 kW Switched Reluctance Motor prototype with a nominal airgap length of 0.4 mm with a relatively large stator outer diameter of 280 mm. It presents the mechanical design considerations and manufacturing tolerances that are vital for an electric motor to maintain component integrity and meet performance requirements. Multi-stage design of the mechanical components and fittings for the complete mechanical system are presented and justified with finite element analysis results. The radial and axial stack-up analysis are presented which account for the dimensional variations during the manufacturing of motor components. The assembly of the motor prototype is described, which demonstrates that the axial and radial alignment of the motor components are achieved while maintaining the small airgap length required. The motor assembly is then verified with static end-of-line tests, such as winding insulation and housing leakage tests. Additionally, experimental modal analysis of the housing assembly is performed to examine the influence of the winding, housing, and potting on its modal frequencies and damping characteristics.
Recent advances in adversarial machine learning have shown that defenses previously considered robust are actually susceptible to adversarial attacks which are specifically customized to target their weaknesses. However, whether the adversarial examples generated by customized attacks, are effective on other defenses, is an open question. In this work we seek to explore three important security questions: First, do different defense strategies exhibit the same low transferability properties as different model architectures and, if so, how can this low transferability be utilized to improve robustness? Second, how can a white-box adversary design attacks to specifically thwart multi-defense based setups? Last, how can game theoretic analysis further improve the robustness against an adversary capable of implementing multiple state-of-the-art attacks? To this end we provide multiple contributions, including the first transferability study between multiple defense strategies, three new attack algorithms designed to break random transform and ensemble defenses, and two game theoretic frameworks for analyzing and optimizing robustness over a combination of adversarial attacks and defenses. Empirically, we show our framework is 18% more robust on CIFAR-10 and is 27% more robust on Tiny-ImageNet than the best single state-of-the-art defense that we analzye.
This article proposes a back-to-back converter-based electric spring for the voltage and frequency regulation in a self-excited induction generator-based isolated generation system. The proposed electric spring topology gives the system seamless performance with linear, non-linear, and unbalanced load conditions. The electric spring in this work is equipped with a modified complex coefficient filter, which eliminates the harmonics and the unbalanced negative sequence component from the system currents caused by unbalanced loads. The performance of the proposed electric spring was evaluated with non-linear loads, unbalanced critical load, and unbalanced non-critical load. The proposed electric spring works effectively to regulate the system voltage and frequency within 2% tolerance limits. In the presence of non-linear loads, the THD of the generator current could be reduced to 1.97%, proving the electric spring’s efficacy as an active filter. The modified complex coefficient filter-based shunt side control strategy effectively eliminates the unbalanced current components from the system currents, which keeps the self-excited induction generator unaffected by load imbalances caused by both the critical load and non-critical load sides. The shunt side control strategy also effectively maintained the DC link voltage fluctuation within 16V, which is 6.4% of the rated value of 250V. The proposed minimum voltage injection control strategy to control the series converter was highly effective in maintaining a constant voltage of 230V during unbalanced non-critical load conditions. The real-time validation of the proposed work is done with OPAL-RT 4510 and 4520 platforms, and the detailed results and their analysis are presented in this work.