11115 research outputs found
Sort by
Editorial: Artificial Intelligence for Smart Health: Learning, Simulation, and Optimization
Get PDFWith rapid developments in medical sensing and imaging, we now live in an era of data explosion in which large amounts of data are readily available in clinical environments. The fast-growing biomedical and healthcare data provide unprecedented opportunities for data-driven scientific knowledge discovery and clinical decision support. Our Research Topic aims to catalyze synergies among biomedical informatics, machine learning, computer simulation, operations research, systems engineering, and other related fields with three specific goals: (1) develop cutting-edge data-driven models to accelerate scientific knowledge discovery in biomedicine using healthcare data collected from laboratory systems, imaging systems, and medical and sensing devices; (2) develop advanced simulation and calibration algorithms to build personalized digital twins by effectively assimilating patient-specific medical data with population-level computer models, facilitating precision medical planning; (3) develop innovative optimization algorithms for optimal medical decision making in the face of uncertainty factors, conflicting objectives, and complex trade-offs. This Research Topic, containing 10 articles, will offer a timely collection of information to benefit researchers and practitioners working in the broad fields of biomedical informatics, healthcare data analytics, medical image processing, and health-related AI
Improving 6D Localization with Rigid Body Prediction of Off-image Features
No full textExcerpt: Accurately calculating the 6D position and orientation of an object from a single color image is a challenging task due to occlusions and information loss when the object moves out of the image frame. In this research, we present a novel approach to address this issue by training a neural network to detect components of the object that are outside the image boundary
Data-Driven Sparse Sensor Placement Optimization on Wings for Flight-By-Feel: Bioinspired Approach and Application
Get PDFFlight-by-feel (FBF) is an approach to flight control that uses dispersed sensors on the wings of aircraft to detect flight state. While biological FBF systems, such as the wings of insects, often contain hundreds of strain and flow sensors, artificial systems are highly constrained by size, weight, and power (SWaP) considerations, especially for small aircraft. An optimization approach is needed to determine how many sensors are required and where they should be placed on the wing. Airflow fields can be highly nonlinear, and many local minima exist for sensor placement, meaning conventional optimization techniques are unreliable for this application. The Sparse Sensor Placement Optimization for Prediction (SSPOP) algorithm extracts information from a dense array of flow data using singular value decomposition and linear discriminant analysis, thereby identifying the most information-rich sparse subset of sensor locations. In this research, the SSPOP algorithm is evaluated for the placement of artificial hair sensors on a 3D delta wing model with a 45° sweep angle and a blunt leading edge. The sensor placement solution, or design point (DP), is shown to rank within the top one percent of all possible solutions by root mean square error in angle of attack prediction. This research is the first to evaluate SSPOP on a 3D model and the first to include variable length hairs for variable velocity sensitivity. A comparison of SSPOP against conventional greedy search and gradient-based optimization shows that SSPOP DP ranks nearest to optimal in over 90 percent of models and is far more robust to model variation. The successful application of SSPOP in complex 3D flows paves the way for experimental sensor placement optimization for artificial hair-cell airflow sensors and is a major step toward biomimetic flight-by-feel
Epistemic Modeling Uncertainty of Rapid Neural Network Ensembles for Adaptive Learning
No full textEmulator embedded neural networks, which are a type of physics informed neural network, leverage multi-fidelity data sources for efficient design exploration of aerospace engineering systems. Multiple realizations of the neural network models are trained with different random initializations. The ensemble of model realizations is used to assess epistemic modeling uncertainty caused due to lack of training samples. This uncertainty estimation is crucial information for successful goal-oriented adaptive learning in an aerospace system design exploration. However, the costs of training the ensemble models often become prohibitive and pose a computational challenge, especially when the models are not trained in parallel during adaptive learning. In this work, a new type of emulator embedded neural network is presented using the rapid neural network paradigm. Unlike the conventional neural network training that optimizes the weights and biases of all the network layers by using gradient-based backpropagation, rapid neural network training adjusts only the last layer connection weights by applying a linear regression technique. It is found that the proposed emulator embedded neural network trains near-instantaneously, typically without loss of prediction accuracy. The proposed method is demonstrated on multiple analytical examples, as well as an aerospace flight parameter study of a generic hypersonic vehicle
Towards an Efficient Method for F-16 Limit Cycle Oscillation Prediction
No full textThis study presents the development and validation of a computationally efficient prediction framework for the well-known nonlinear F-16 Limit Cycle Oscillations (LCO) phenomenon. The framework relies on a simple physical working model which has been suggested and demonstrated in the past according to which LCO is primarily a flutter instability that is bounded by the existence of nonlinear structural damping, although potentially affected by nonlinear aerodynamic effects as well. In the framework developed herein, the nonlinear structural damping (NSD) model is derived and calibrated using a novel method which simplifies the process and allows applicability of the derived NSD models for multiple aircraft download cases. Good LCO prediction capabilities are obtained using the suggested method in terms of LCO levels and trends with flight conditions, as demonstrated using four F-16 test configurations. This framework also allows several practical benefits which makes it particularly suitable for industrial-level applications
The Effects of a Non-Uniform Magnetic Field on Solar Cell Efficiency
Get PDFCommercial-grade silicon-based solar cells have an efficiency in the 20-30% range. The addition of a non-uniform magnetic field manipulates the movement of charge carriers within the silicon of a solar cell. With this manipulation, one hypothesis is that the magnetic field increases the current produced by the solar cell which helps to increase the power and efficiency of the solar cell. Measuring a solar cell’s output current and voltage both with and without the presence of a non-uniform magnetic field tests this theory. Voltage multiplied by current gives output power, and to be calculated, efficiency needs maximum output power. The experiment shows slight increases in efficiency when factoring in the presence of a non-uniform magnetic field, and the results indicate further testing should be done
Improving Rogue Radio Emitter Detection Using Siamese Networks
Get PDFRadio Frequency Fingerprinting (RFF) is the process of creating discerning signatures of emitted radio signals, most often with the goal of identifying specific devices again in the future. The security benefits of this task are intended to build upon current software-based authentication by making use of multi-factor authentication (MFA), but the related task of being able to reject unwanted emitters is limited. This paper presents a Siamese network trained on two different extracted fingerprints of raw Wi-Fi signals, along with a verifier to perform classification and rogue device detection. It was found that fingerprints using the Distortion Reconstruction (DR) technique outperform the popular Time-Domain Distinct Native Attributes (TD-DNA) method with respective classification accuracies of 99.15% and 85.88% when trained on 190 classes, and using a modified triplet loss with the Siamese network could create embeddings of the fingerprints capable of comparable classification while being able to reject rogue devices better than a straightforward classifier, with respective rejection rates of 98.12% and 86.88%, and while using one-third fewer parameters
A Random Forest-Based Q-Learning Algorithm: Toward Interpretable Artificial Intelligence
Get PDFA growing demand exists for interpretable artificial intelligence models, leading to extensive research efforts to enhance the explainability and transparency of policies generated by reinforcement learning (RL) methods. This research develops random forest-based RL algorithms as a logical progression in this academic pursuit. The algorithms are evaluated using three standard benchmark environments from OpenAI gym — CartPole, MountainCar, and LunarLander — and compared to implementations of the Deep Q-learning Network (DQN) and Double DQN (DDQN) algorithms for various metrics, including performance, robustness, efficiency, and interpretability. The random forest-based algorithms exhibit superior performance to both neural network-based algorithms in two out of three environments while additionally providing easily interpretable decision tree policies. However, the proffered approach faces challenges in solving the LunarLander environment, indicating limitations in its current ability to scale to larger environments
An Introduction of Adaptive Training Aid Concepts and Its Application to Accelerated Training for Air Battle Managers
Get PDFOver the years, the integration of artificial intelligence (AI) to enable autonomous systems has undergone transformative shifts in the Department of Defense (DoD), revolutionizing capabilities and strategic approaches. To further optimize these advancements, varying levels of autonomy have been introduced across critical military applications, spanning intelligence, surveillance, reconnaissance (ISR), air battle management, and offensive/defensive air operations. As the technological landscape expands, so do the opportunities for autonomy to augment operations through human-agent teaming. Within the Air Force, one notably cognitively demanding role that stands to benefit from these strides is that of the Air Battle Manager (ABM). In support of the autonomy evolution, the DoD devised the Unmanned Systems Integrated Roadmap (USIR), offering comprehensive strategic guidance aligning unmanned systems with the broader DoD vision (USIR, 2017). While the USIR addresses various themes, this research zeroes in on autonomy and its enablers, emphasizing training to enhance the efficiency and effectiveness of DoD systems. Notably, this research extends the roadmap\u27s general framework, delving into real-world implications for ABMs and how the initial training for this critical career field could be influenced