Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Active Learning”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 703 records · Page 39

Hybrid-RL-MPC4CLR (Hybird-Reinforcement-Learning-Model-Predictive-Control-for-Reserve-Policy-Assisted-Critical-Load-Restoration-in-Distribution-Grids)

Hybrid-RL-MPC4CLR was developed as a hybrid controller for active distribution grid critical load restoration, combining deep reinforcement learning (RL) and model predictive control (MPC) aiming at maximizing total restored load following an extreme event. The RL determines a policy for quantifying operating reserve requirements, thereby hedging against uncertainty, while the MPC models grid operations incorporating the RL policy actions (i.e., reserve requirements), renewable (wind and solar) power predictions, and load demand forecasts. The developers formulated the reserve requirement determination problem as a sequential decision-making problem based on the Markov Decision Process (MDP) and design an RL learning environment based on the OpenAI Gym framework and MPC simulation. The RL agent reward and MPC objective function aim to maximize and monotonically increase total restored load and minimize load shedding and renewable power curtailment. The software is developed using various software packages in Python. The MPC's optimal power flow (OPF) model is implemented using the Pyomo package, the RL simulation environment is implemented using the MPC simulation with various scenarios of renewable energy and load demand profiles and power outage beginning times, based on the OpenAI Gym framework. The RL agent training is performed using the RLlib Ray package. The RL algorithm is trained offline using historical forecasts of renewable generation and load demand profiles. Simulation analysis and performance tests are conducted using a modified IEEE 13-bus distribution test feeder containing wind turbine, photovoltaic, microturbine, and battery.

Eseye, Abinet Tesfaye↗

The Effect of Realtime Monitoring on Dose Exposure to Staff Within an Interventional Radiology Setting

PurposeThe purpose of this study is to evaluate a new device providing real-time monitoring on radiation exposure during fluoroscopy procedures intending to reduce radiation in an interventional radiology setting.Materials and MethodsIn one interventional suite, a new system providing a real-time radiation dose display and five individual wireless dosimeters were installed. The five dosimeters were worn by the attending, fellow, nurse, technician, and anesthesiologist for every procedure taking place in that suite. During the first 6-week interval the dose display was off (closed phase) and activated thereafter, for a 6-week learning phase (learning phase) and a 10-week open phase (open phase). During these phases, the staff dose and the individual dose for each procedure were recorded from the wireless dosimeter and correlated with the fluoroscopy time. Further subanalysis for dose exposure included diagnostic versus interventional as well as short (<10 min) versus long (>10 min) procedures.ResultsA total of 252 procedures were performed (n = 88 closed phase, n = 50 learning phase, n = 114 open phase). The overall mean staff dose per fluoroscopic minute was 42.79 versus 19.81 µSv/min (p < 0.05) comparing the closed and open phase. Thereby, anesthesiologists were the only individuals attaining a significant dose reduction during open phase 16.9 versus 8.86 µSv/min (p < 0.05). Furthermore, a significant reduction of total staff dose was observed for short 51 % and interventional procedures 45 % (p < 0.05, for both).ConclusionA real-time qualitative display of radiation exposure may reduce team radiation dose. The process may take a few weeks during the learning phase but appears sustained, thereafter.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Back-to-back high category atmospheric river landfalls occur more often on the west coast of the United States

Abstract The catastrophic December 2022-January 2023 nine atmospheric rivers in California underscore the urgent need to better understand such high-risk weather extremes. Here we applied a machine learning clustering tool to understand the activity of atmospheric river clusters. Reanalysis results show that clusters with high density, that is the time fraction under atmospheric river conditions within a cluster, exhibit more frequent high-category atmospheric rivers, alongside an increased likelihood for extreme precipitation and severe land surface responses. The key circulation patterns of atmospheric river clusters are primarily attributed to subseasonal variability. Furthermore, the occurrence and density of atmospheric river clusters are modulated by the daily variability of the geopotential height field. Climate model projections suggest that atmospheric river clusters with higher density and higher categories will be more frequent as warming level increases. Our findings emphasize the important role of atmospheric river clusters in the development of climate adaptation and resilience strategies.

54 ENVIRONMENTAL SCIENCES↗

Role of Local Structure in the Enhanced Dynamics of Deformed Glasses

External stress can accelerate molecular mobility of amorphous solids by several orders of magnitude. The changes in mobility are commonly interpreted through the Eyring model, which invokes an empirical activation volume. Here, we analyze constant-stress molecular dynamics simulations and propose a structure-dependent Eyring model, connecting activation volume to a machine-learned field, softness. We show that stress has a heterogeneous effect on the mobility that depends on local structure through softness. The barrier impeding relaxation reduces more for well-packed particles, which explains the narrower distribution of relaxation time observed under stress.

74 ATOMIC AND MOLECULAR PHYSICS↗

QuCNN : A Quantum Convolutional Neural Network with Entanglement Based Backpropagation

Quantum Machine Learning continues to be a highly active area of interest within Quantum Computing. Many of these approaches have adapted classical approaches to the quantum settings, such as QuantumFlow, etc. We push forward this trend, and demonstrate an adaption of the Classical Convolutional Neural Networks to quantum systems - namely QuCNN. QuCNN is a parameterised multi-quantum-state based neural network layer computing similarities between each quantum filter state and each quantum data state. With QuCNN, back propagation can be achieved through a single-ancilla qubit quantum routine. QuCNN is validated by applying a convolutional layer with a data state and a filter state over a small subset of MNIST images, comparing the backpropagated gradients, and training a filter state against an ideal target state.

high performance computing, quantum computing↗

LRBOI Tribal Renewable Energy-Final Report

In 2011, a DOE Tribal energy Program first Steps Human Capacity was awarded to the Little River Band of Ottawa Indians (LRBOI). The main purpose of the grant was to increase human capacity of LRBOI, to understand the components of renewable energy and the importance of energy efficiency. This report summarizes the activities, the outcomes, and the lessons learned during this grant.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

ReLU, Sparseness, and the Encoding of Optic Flow in Neural Networks

Accurate self-motion estimation is critical for various navigational tasks in mobile robotics. Optic flow provides a means to estimate self-motion using a camera sensor and is particularly valuable in GPS- and radio-denied environments. The present study investigates the influence of different activation functions—ReLU, leaky ReLU, GELU, and Mish—on the accuracy, robustness, and encoding properties of convolutional neural networks (CNNs) and multi-layer perceptrons (MLPs) trained to estimate self-motion from optic flow. Our results demonstrate that networks with ReLU and leaky ReLU activation functions not only achieved superior accuracy in self-motion estimation from novel optic flow patterns but also exhibited greater robustness under challenging conditions. The advantages offered by ReLU and leaky ReLU may stem from their ability to induce sparser representations than GELU and Mish do. Our work characterizes the encoding of optic flow in neural networks and highlights how the sparseness induced by ReLU may enhance robust and accurate self-motion estimation from optic flow.

97 MATHEMATICS AND COMPUTING↗

Next Steps for the FCEV Learning Demonstration Project (Presentation)

This presentation summarizes project goals; vehicle and H2 station deployment status, critical performance compared to targets; highlights of latest vehicle and infrastructure analysis results and progress; learning demo next steps; highlights of partner activities and summary.

08 HYDROGEN↗

Transfer learning for smart buildings: A critical review of algorithms, applications, and future perspectives

Smart buildings play a crucial role toward decarbonizing society, as globally buildings emit about one-third of greenhouse gases. In the last few years, machine learning has achieved a notable momentum that, if properly harnessed, may unleash its potential for advanced analytics and control of smart buildings, enabling the technique to scale up for supporting the decarbonization of the building sector. In this perspective, transfer learning aims to improve the performance of a target learner exploiting knowledge in related environments. The present work provides a comprehensive overview of transfer learning applications in smart buildings, classifying and analyzing 77 papers according to their applications, algorithms, and adopted metrics. The study identified four main application areas of transfer learning: (1) building load prediction, (2) occupancy detection and activity recognition, (3) building dynamics modeling, and (4) energy systems control. Furthermore, the review highlighted the role of deep learning in transfer learning applications that has been used in more than half of the analyzed studies. The paper also discusses how to integrate transfer learning in a smart building's ecosystem, identifying, for each application area, the research gaps and guidelines for future research directions.

Pinto, G↗

Machine learning properties of binary wurtzite superlattices

The burgeoning paradigm of high-throughput computations and materials informatics brings new opportunities in terms of targeted materials design and discovery. The discovery process can be significantly accelerated and streamlined if one can learn effectively from available knowledge and past data to predict materials properties efficiently. Indeed, a very active area in materials science research is to develop machine learning based methods that can deliver automated and cross-validated predictive models using either already available materials data or new data generated in a targeted manner. In the present paper, we show that fast and accurate predictions of a wide range of properties of binary wurtzite superlattices, formed by a diverse set of chemistries, can be made by employing state-of-the-art statistical learning methods trained on quantum mechanical computations in combination with a judiciously chosen numerical representation to encode materials’ similarity. These surrogate learning models then allow for efficient screening of vast chemical spaces by providing instant predictions of the targeted properties. Moreover, the models can be systematically improved in an adaptive manner, incorporate properties computed at different levels of fidelities and are naturally amenable to inverse materials design strategies. Finally, while the learning approach to make predictions for a wide range of properties (including structural, elastic and electronic properties) is demonstrated here for a specific example set containing more than 1200 binary wurtzite superlattices, the adopted framework is equally applicable to other classes of materials as well.

36 MATERIALS SCIENCE↗

Graphene-supported single atom catalysts for high performance lithium-oxygen batteries

The optimal choice of d-block metals in single atom catalysts (SACs) is crucial for designing efficient electrocatalysts for activating the Oxygen reduction reaction (ORR)/ Oxygen evolution reaction (OER) in lithiumoxygen batteries (LOBs). Herein, we used the Quantum Mechanics methods to understand the origin of reactivity for a series of 16 d-block metals supported on nitrogen-doped graphene as SACs for ORR and OER in LOBs. Based on the Gibbs free energy calculations, we found that among the 16 SACs investigated, Zn-SAC exhibits the highest electrochemical activity with the lowest overpotential of 0.17 V. Here we then used machine learning (ML) to develop an intrinsic descriptor, phi, that correlates the catalytic activity with electronic and chemical properties of the catalytic centers at the M-N 4 active site on graphene surface. We established a linear relationship between phi and the catalytic activity that provides guidance for designing efficient SACs for electrocatalysis in LOBs. To validate these predictions, we report electrochemical measurements showing that Zn-SAC exhibits an ultra-stable cyclability with reduced overpotentials over Mo-SAC and nitrogen-doped graphene (NG), confirming our theoretical prediction. This fundamental work provides a deep understanding on the rational design of efficient SACs for OER/ ORR in LOBs.

25 ENERGY STORAGE↗

Cartesian equivariant representations for learning and understanding molecular orbitals

Qualitative and quantitative orbital properties such as bonding/antibonding character, localization, and orbital energies are critical to how chemists understand reactivity, catalysis, and excited-state behavior. Despite this, representations of orbitals in deep learning models have been very underdeveloped relative to representations of molecular geometries and Hamiltonians. Here, we apply state-of-the-art equivariant deep learning architectures to the task of assigning global labels to orbitals, namely energies characterizations, given the molecular coefficients from Hartree–Fock or density functional theory. The architecture we have developed, the Cartesian Equivariant Orbital Network (CEONET), shows how molecular orbital coefficients are readily featurized as equivariant node features common to all graph-based machine-learned potentials. We find that CEONET performs well at predicting difficult quantitative labels such as the orbital energy and orbital entropy. Furthermore, we find that the CEONET representation provides an intuitive latent space for differentiating orbital character for the qualitative assignment of e.g. bonding or antibonding character. In addition to providing a useful representation for further integrating deep learning with electronic structure theory, we expect CEONET to be useful for automatizing and interpreting the results of advanced electronic structure methods such as complete active space self-consistent field theory. In particular, the ability of CEONET to infer multireference character via the orbital entropy paves the way toward the machine-learned selection of active spaces.

chemical reactions↗

Stability of heterogeneous single-atom catalysts: a scaling law mapping thermodynamics to kinetics

Heterogeneous single-atom catalysts (SACs) hold the promise of combining high catalytic performance with maximum utilization of often precious metals. We extend the current thermodynamic view of SAC stability in terms of the binding energy (E bind ) of single-metal atoms on a support to a kinetic (transport) one by considering the activation barrier for metal atom diffusion. A rapid computational screening approach allows predicting diffusion barriers for metal–support pairs based on E bind of a metal atom to the support and the cohesive energy of the bulk metal (E c ). Metal–support combinations relevant to contemporary catalysis are explored by density functional theory. Assisted by machine-learning methods, we find that the diffusion activation barrier correlates with (E bind ) 2 /E c in the physical descriptor space. This diffusion scaling-law provides a simple model for screening thermodynamics to kinetics of metal adatom on a support.

36 MATERIALS SCIENCE↗

MPTP-meditated hippocampal dopamine deprivation modulates synaptic transmission and activity-dependent synaptic plasticity

Parkinson's disease (PD)-like symptoms including learning deficits are inducible by 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP). Therefore, it is possible that MPTP may disturb hippocampal memory processing by modulation of dopamine (DA)- and activity-dependent synaptic plasticity. We demonstrate here that intraperitoneal (i.p.) MPTP injection reduces the number of tyrosine hydroxylase (TH)-positive neurons in the substantia nigra (SN) within 7 days. Subsequently, the TH expression level in SN and hippocampus and the amount of DA and its metabolite DOPAC in striatum and hippocampus decrease. DA depletion does not alter basal synaptic transmission and changes pair-pulse facilitation (PPF) of field excitatory postsynaptic potentials (fEPSPs) only at the 30 ms inter-pulse interval. In addition, the induction of long-term potentiation (LTP) is impaired whereas the duration of long-term depression (LTD) becomes prolonged. Since both LTP and LTD depend critically on activation of NMDA and DA receptors, we also tested the effect of DA depletion on NMDA receptor-mediated synaptic transmission. Seven days after MPTP injection, the NMDA receptor-mediated fEPSPs are decreased by about 23%. Blocking the NMDA receptor-mediated fEPSP does not mimic the MPTP-LTP. Only co-application of D1/D5 and NMDA receptor antagonists during tetanization resembled the time course of fEPSP potentiation as observed 7 days after i.p. MPTP injection. Together, our data demonstrate that MPTP-induced degeneration of DA neurons and the subsequent hippocampal DA depletion alter NMDA receptor-mediated synaptic transmission and activity-dependent synaptic plasticity. - Highlights: > I.p. MPTP-injection mediates death of dopaminergic neurons. > I.p. MPTP-injection depletes DA and DOPAC in striatum and hippocampus. > I.p. MPTP-injection does not alter basal synaptic transmission. > Reduction of LTP and enhancement of LTD after i.p. MPTP-injection. > Attenuation of NMDA-receptors mediated fEPSPs after i.p. MPTP-injection.

60 APPLIED LIFE SCIENCES↗

Recent Progress on Surface Water Quality Models Utilizing Machine Learning Techniques

Surface waterbodies are heavily exposed to pollutants caused by natural disasters and human activities. Empowering sensor technologies in water quality monitoring, sufficient measurements have become available to develop machine learning (ML) models. Numerous ML models have quickly been adopted to predict water quality indicators in various surface waterbodies. This paper reviews 78 recent articles from 2022 to October 2024, categorizing water quality models utilizing ML into three groups: Point-to-Point (P2P), which estimates the current target value based on other measurements at the same time point; Sequence-to-Point (S2P), which utilizes previous time series data to predict the target value at one time point ahead; and Sequence-to-Sequence (S2S), which uses previous time series data to forecast sequential target values in the future. The ML models used in each group are classified and compared according to water quality indicators, data availability, and model performance. Widely used strategies for improving performance, including feature engineering, hyperparameter tuning, and transfer learning, are recognized and described to enhance model effectiveness. The interpretability limitations of ML applications are discussed. This review provides a perspective on emerging ML for surface water quality models.

machine learning (ML)↗

Low responsiveness of machine learning models to critical or deteriorating health conditions

Machine learning (ML) based mortality prediction models can be immensely useful in intensive care units. Such a model should generate warnings to alert physicians when a patient’s condition rapidly deteriorates, or their vitals are in highly abnormal ranges. Before clinical deployment, it is important to comprehensively assess a model’s ability to recognize critical patient conditions. We develop multiple medical ML testing approaches, including a gradient ascent method and neural activation map. We systematically assess these machine learning models’ ability to respond to serious medical conditions using additional test cases, some of which are time series. Guided by medical doctors, our evaluation involves multiple machine learning models, resampling techniques, and four datasets for two clinical prediction tasks. We identify serious deficiencies in the models’ responsiveness, with the models being unable to recognize severely impaired medical conditions or rapidly deteriorating health. For in-hospital mortality prediction, the models tested using our synthesized cases fail to recognize 66% of the injuries. In some instances, the models fail to generate adequate mortality risk scores for all test cases. Our study identifies similar kinds of deficiencies in the responsiveness of 5-year breast and lung cancer prediction models. Using generated test cases, we find that statistical machine-learning models trained solely from patient data are grossly insufficient and have many dangerous blind spots. Most of the ML models tested fail to respond adequately to critically ill patients. How to incorporate medical knowledge into clinical machine learning models is an important future research direction.

60 APPLIED LIFE SCIENCES↗

Deep Learning without Global Optimization by Random Fourier Neural Networks

Here we introduce a new training algorithm for deep neural networks that utilize random complex exponential activation functions. Our approach employs a Markov chain Monte Carlo sampling procedure to iteratively train network layers, avoiding global and gradient-based optimization while maintaining error control. It consistently attains the theoretical approximation rate for residual networks with complex exponential activation functions, determined by network complexity. Additionally, it enables efficient learning of multiscale and high-frequency features, producing interpretable parameter distributions. Despite using sinusoidal basis functions, we do not observe Gibbs phenomena in approximating discontinuous target functions.

97 MATHEMATICS AND COMPUTING↗