DeepVision: Enhanced Drone Detection and Recognition in Visible Imagery through Deep Learning Networks.

Drones are increasingly capturing the world's attention, transcending mere hobbies to revolutionize areas such as engineering, disaster aid, logistics, and airport protection, among myriad other fascinating applications. However, there is growing concern about the risks that they pose to physical infrastructure, particularly at airports, due to potential misuse. In recent times, numerous incidents involving unauthorized drones at airports disrupting flights have been reported. To solve this issue, this article introduces an innovative deep learning method proposed to effectively distinguish between drones and birds. Evaluating the suggested approach with a carefully assembled image dataset demonstrates exceptional performance, surpassing established detection systems previously proposed in the literature. Since drones can appear extremely small compared to other aerial objects, we developed a robust image-tiling technique with overlaps, which showed improved performance in the presence of very small drones. Moreover, drones are frequently mistaken for birds due to their resemblances in appearance and movement patterns. Among the various models tested, including SqueezeNet, MobileNetV2, ResNet18, and ResNet50, the SqueezeNet model exhibited superior performance for medium area ratios, achieving higher average precision (AP) of 0.770. In addition, SqueezeNet's superior AP scores, faster detection times, and more stable precision-recall dynamics make it more suitable for real-time, accurate drone detection than the other existing CNN methods. The proposed approach has the ability to not only detect the presence or absence of drones in a particular area but also to accurately identify and differentiate between drones and birds. The dataset utilized in this research was obtained from a real-world dataset made available by a group of universities and research institutions as part of the 2020 Drone vs. Bird Detection Challenge. We have also tested the performance of the proposed model on an unseen dataset, further validating its better performance.

Medical ultrasound image speckle reduction and resolution enhancement using texture compensated multi-resolution convolution neural network.

Ultrasound (US) imaging is a mature technology that has widespread applications especially in the healthcare sector. Despite its widespread use and popularity, it has an inherent disadvantage that ultrasound images are prone to speckle and other kinds of noise. The image quality in the low-cost ultrasound imaging systems is degraded due to the presence of such noise and low resolution of such ultrasound systems. Herein, we propose a method for image enhancement where, the overall quality of the US images is improved by simultaneous enhancement of US image resolution and noise suppression. To avoid over-smoothing and preserving structural/texture information, we devise texture compensation in our proposed method to retain the useful anatomical features. Moreover, we also utilize US image formation physics knowledge to generate augmentation datasets which can improve the training of our proposed method. Our experimental results showcase the performance of the proposed network as well as the effectiveness of the utilization of US physics knowledge to generate augmentation datasets.

ECM-LSE: Prediction of Extracellular Matrix Proteins Using Deep Latent Space Encoding of k-Spaced Amino Acid Pairs.

Extracelluar matrix (ECM) proteins create complex networks of macromolecules which fill-in the extracellular spaces of living tissues. They provide structural support and play an important role in maintaining cellular functions. Identification of ECM proteins can play a vital role in studying various types of diseases. Conventional wet lab-based methods are reliable; however, they are expensive and time consuming and are, therefore, not scalable. In this research, we propose a sequence-based novel machine learning approach for the prediction of ECM proteins. In the proposed method, composition of k-spaced amino acid pair (CKSAAP) features are encoded into a classifiable latent space (LS) with the help of deep latent space encoding (LSE). A comprehensive ablation analysis is conducted for performance evaluation of the proposed method. Results are compared with other state-of-the-art methods on the benchmark dataset, and the proposed ECM-LSE approach has shown to comprehensively outperform the contemporary methods.

Beamforming Using Exact Evaluation of Leakage and Ergodic Capacity of MU-MIMO System.

Closed-form evaluation of key performance indicators (KPIs) of telecommunication networks help perform mathematical analysis under several network configurations. This paper deals with a recent mathematical approach of indefinite quadratic forms to propose simple albeit exact closed-form expressions of the expectation of two significant logarithmic functions. These functions formulate KPIs which include the ergodic capacity and leakage rate of multi-user multiple-input multiple-output (MU-MIMO) systems in Rayleigh fading channels. Our closed-form expressions are generic in nature and they characterize several network configurations under statistical channel state information availability. As a demonstrative example of the proposed characterization, the derived expressions are used in the statistical transmit beamformer design in a broadcast MU-MIMO system to portray promising diversity gains using standalone or joint maximization techniques of the ergodic capacity and leakage rate. The results presented are validated by Monte Carlo simulations.

Underwater Vehicle Positioning by Correntropy-Based Fuzzy Multi-Sensor Fusion.

The ability of the underwater vehicle to determine its precise position is vital to completing a mission successfully. Multi-sensor fusion methods for underwater vehicle positioning are commonly based on Kalman filtering, which requires the knowledge of process and measurement noise covariance. As the underwater conditions are continuously changing, incorrect process and measurement noise covariance affect the accuracy of position estimation and sometimes cause divergence. Furthermore, the underwater multi-path effect and nonlinearity cause outliers that have a significant impact on positional accuracy. These non-Gaussian outliers are difficult to handle with conventional Kalman-based methods and their fuzzy variants. To address these issues, this paper presents a new and improved adaptive multi-sensor fusion method by using information-theoretic, learning-based fuzzy rules for Kalman filter covariance adaptation in the presence of outliers. Two novel metrics are proposed by utilizing correntropy Gaussian and Versoria kernels for matching theoretical and actual covariance. Using correntropy-based metrics and fuzzy logic together makes the algorithm robust against outliers in nonlinear dynamic underwater conditions. The performance of the proposed sensor fusion technique is compared and evaluated using Monte-Carlo simulations, and substantial improvements in underwater position estimation are obtained.

Multi-Sensor Fusion for Underwater Vehicle Localization by Augmentation of RBF Neural Network and Error-State Kalman Filter.

The Kalman filter variants extended Kalman filter (EKF) and error-state Kalman filter (ESKF) are widely used in underwater multi-sensor fusion applications for localization and navigation. Since these filters are designed by employing first-order Taylor series approximation in the error covariance matrix, they result in a decrease in estimation accuracy under high nonlinearity. In order to address this problem, we proposed a novel multi-sensor fusion algorithm for underwater vehicle localization that improves state estimation by augmentation of the radial basis function (RBF) neural network with ESKF. In the proposed algorithm, the RBF neural network is utilized to compensate the lack of ESKF performance by improving the innovation error term. The weights and centers of the RBF neural network are designed by minimizing the estimation mean square error (MSE) using the steepest descent optimization approach. To test the performance, the proposed RBF-augmented ESKF multi-sensor fusion was compared with the conventional ESKF under three different realistic scenarios using Monte Carlo simulations. We found that our proposed method provides better navigation and localization results despite high nonlinearity, modeling uncertainty, and external disturbances.

Performance Evaluation of EEG Based Mental Stress Assessment Approaches for Wearable Devices.

Mental stress has been identified as the root cause of various physical and psychological disorders. Therefore, it is crucial to conduct timely diagnosis and assessment considering the severe effects of mental stress. In contrast to other health-related wearable devices, wearable or portable devices for stress assessment have not been developed yet. A major requirement for the development of such a device is a time-efficient algorithm. This study investigates the performance of computer-aided approaches for mental stress assessment. Machine learning (ML) approaches are compared in terms of the time required for feature extraction and classification. After conducting tests on data for real-time experiments, it was observed that conventional ML approaches are time-consuming due to the computations required for feature extraction, whereas a deep learning (DL) approach results in a time-efficient classification due to automated unsupervised feature extraction. This study emphasizes that DL approaches can be used in wearable devices for real-time mental stress assessment.

Distributed Extended Kalman Filtering Based Techniques for 3-D UAV Jamming Localization.

Wireless networks are vulnerable to jamming attacks. Jamming in wireless communication becomes a major research problem due to ease in Unmanned Aerial Vehicle (UAV) launching and blocking of communication channels. Jamming is a subset of Denial of Service Attack (DoS) and an intentional interference where the malicious node disrupts the wireless communication by increasing the noise at the receiver node through transmission interference signal towards the target channel. In this work, the considered jammer is a UAV hovering around the target area to block the communication channel between two transceivers. We proposed a three-dimensional (3-D) UAV jamming localization scheme to track and detect the jammer position at each time step by employing a single boundary node observer. For this purpose, we developed two distributed Extended Kalman Filter (EKF) based schemes: (1) the Distributed EKF (DEKF) scheme using the information of the received power from the jammer at a single nearby boundary node only and (2) Distance Ratio aided Distributed EKF (DEKF-DR) based scheme utilizing an edge node in addition to a single boundary node. Extensive simulations are conducted in order to evaluate the performance of the proposed distributed algorithms for a 3-D trajectory and compared with that of the conventional Centralized EKF (EKF-Centr) based method (which is also modified for the 3-D scenario). The results show the clear supremacy of the proposed distributed algorithms with much lesser complexity in contrast to the conventional EKF-Centr technique.

Real-Time Stress Assessment Using Sliding Window Based Convolutional Neural Network.

Mental stress has been identified as a significant cause of several bodily disorders, such as depression, hypertension, neural and cardiovascular abnormalities. Conventional stress assessment methods are highly subjective and tedious and tend to lack accuracy. Machine-learning (ML)-based computer-aided diagnosis systems can be used to assess the mental state with reasonable accuracy, but they require offline processing and feature extraction, rendering them unsuitable for real-time applications. This paper presents a real-time mental stress assessment approach based on convolutional neural networks (CNNs). The CNN-based approach afforded real-time mental stress assessment with an accuracy as high as 96%, the sensitivity of 95%, and specificity of 97%. The proposed approach is compared with state-of-the-art ML techniques in terms of accuracy, time utilisation, and quality of features.

Differences in photosynthetic syndromes of four halophytic marsh grasses in Pakistan.

Salt-tolerant grasses of warm sub-tropical ecosystems differ in their distribution patterns with respect to salinity and moisture regimes. Experiments were conducted on CO2 fixation and light harvesting processes of four halophytic C4 grasses grown under different levels of salinity (0, 200 and 400 mM NaCl) under ambient environmental conditions. Two species were from a high saline coastal marsh (Aeluropus lagopoides and Sporobolus tremulus) and two were from a moderate saline sub-coastal draw-down tidal marsh (Paspalum paspalodes and Paspalidium geminatum). Analyses of the carbon isotope ratios of leaf biomass in plants indicated that carbon assimilation was occurring by C4 photosynthesis in all species during growth under varying levels of salinity. In the coastal species, with increasing salinity, there was a parallel decrease in rates of CO2 fixation (A), transpiration (E) and stomatal conductance (g s), with no effect on water use efficiency (WUE). These species were adapted for photoprotection by an increase in the Mehler reaction with an increase in activity of PSII/CO2 fixed accompanied by high levels of antioxidant enzymes, superoxide dismutase and ascorbate peroxidase. The sub-coastal species P. paspalodes and P. geminatum had high levels of carotenoid pigments and non-photochemical quenching by the xanthophyll cycle.

Excreting and non-excreting grasses exhibit different salt resistance strategies.

The combination of traits that makes a plant successful under saline conditions varies with the type of plant and its interaction with the environmental conditions. Knowledge about the contribution of these traits towards salt resistance in grasses has great potential for improving the salt resistance of conventional crops. We attempted to identify differential adaptive response patterns of salt-excreting versus non-excreting grasses. More specifically, we studied the growth, osmotic, ionic and nutrient (carbon/nitrogen) relations of two salt-excreting (Aeluropus lagopoides and Sporobolus tremulus) and two non-excreting (Paspalum paspalodes and Paspalidium geminatum) perennial C4 grasses under non-saline and saline (0, 200 and 400 mM NaCl) conditions. Growth and relative growth rate decreased under saline conditions in the order P. geminatum > S. tremulus = A. lagopoides > P. paspalodes. The root-to-shoot biomass allocation was unaffected in salt-excreting grasses, increased in P. paspalodes but decreased in P. geminatum. Salt-excreting grasses had a higher shoot/root Na(+) ratio than non-excreting grasses. K(+), Ca(2+) and Mg(2+) homoeostasis remained undisturbed among test grasses possibly through improved ion selectivity with rising substrate salinity. Salt-excreting grasses increased leaf succulence, decreased ψs and xylem pressure potential, and accumulated proline and glycinebetaine with increasing salinity. Higher salt resistance of P. paspalodes could be attributed to lower Na(+) uptake, higher nitrogen-use efficiency and higher water-use efficiency among the test species. However, P. geminatum was unable to cope with salt-induced physiological drought. More information is required to adequately document the differential strategies of salt resistance in salt-excreting and non-excreting grasses.

An adaptive learning rate for RBFNN using time-domain feedback analysis.

Radial basis function neural networks are used in a variety of applications such as pattern recognition, nonlinear identification, control and time series prediction. In this paper, the learning algorithm of radial basis function neural networks is analyzed in a feedback structure. The robustness of the learning algorithm is discussed in the presence of uncertainties that might be due to noisy perturbations at the input or to modeling mismatch. An intelligent adaptation rule is developed for the learning rate of RBFNN which gives faster convergence via an estimate of error energy while giving guarantee to the l 2 stability governed by the upper bounding via small gain theorem. Simulation results are presented to support our theoretical development.