Search NASA⌕ Search

Engineering topics

Li, Youjia

Publications and source records attributed to Li, Youjia.

4 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Simultaneously improving accuracy and computational cost under parametric constraints in materials property prediction tasks

Abstract Modern data mining techniques using machine learning (ML) and deep learning (DL) algorithms have been shown to excel in the regression-based task of materials property prediction using various materials representations. In an attempt to improve the predictive performance of the deep neural network model, researchers have tried to add more layers as well as develop new architectural components to create sophisticated and deep neural network models that can aid in the training process and improve the predictive ability of the final model. However, usually, these modifications require a lot of computational resources, thereby further increasing the already large model training time, which is often not feasible, thereby limiting usage for most researchers. In this paper, we study and propose a deep neural network framework for regression-based problems comprising of fully connected layers that can work with any numerical vector-based materials representations as model input. We present a novel deep regression neural network, iBRNet, with branched skip connections and multiple schedulers, which can reduce the number of parameters used to construct the model, improve the accuracy, and decrease the training time of the predictive model. We perform the model training using composition-based numerical vectors representing the elemental fractions of the respective materials and compare their performance against other traditional ML and several known DL architectures. Using multiple datasets with varying data sizes for training and testing, We show that the proposed iBRNet models outperform the state-of-the-art ML and DL models for all data sizes. We also show that the branched structure and usage of multiple schedulers lead to fewer parameters and faster model training time with better convergence than other neural networks. Scientific contribution: The combination of multiple callback functions in deep neural networks minimizes training time and maximizes accuracy in a controlled computational environment with parametric constraints for the task of materials property prediction.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Effects of residue removal and tillage on greenhouse gas emissions in continuous corn systems as simulated with RZWQM2

Agricultural production is a major source of carbon dioxide (CO 2 ) and nitrous oxide (N 2 O) globally. The effects of conservation practices on soil CO 2 and N 2 O emissions remain a high degree of uncertainty. In this study, soil CO 2 and N 2 O emissions under different residue and tillage practices in an irrigated, continuous corn system, were investigated using the Root Zone Water Quality Model (RZWQM2). Combinations of no/high stover removal (NR and HR, respectively) and no-till/conventional tillage (NT and CT, respectively) field experiments were tested over the four crop-years (Apr. 2011–Apr. 2015). The model was calibrated using the NRCT, and validated with other treatments. The simulation results showed that soil volumetric water content (VWC) in the NR treatments (i.e., NRCT and NRNT) was 1.3%–1.9% higher than that in the HR treatments (i.e., HRCT and HRNT) averaged across the four years. A higher amount of CO 2 and N 2 O emissions were simulated in the NRCT across the four years (annual average: 7034 kg C/ha/yr for CO 2 and 3.8 kg N/ha/yr for N 2 O), and lower emissions were in the HRNT (annual average: 6329 kg C/ha/yr and 3.7 kg N/ha/yr for N 2 O). A long-term simulation (2001–2015) suggested that the CO 2 and N 2 O emissions were closely correlated with the stover removal degree (SRD), tillage, VWC, soil temperature (ST), years in management (Y), and fertilizer application. Stover and tillage practices had cumulative effects on CO 2 emissions. The simulated annual CO 2 emissions in 1st year from NRCT, NRNT, and HRCT were 7.8%, 0.0%, and 7.7% higher than that from HRNT, respectively; then the emissions in 15th year were 63.6%, 47.7%, and 29.1% higher, respectively. Meanwhile, there were no cumulative effects on N 2 O emissions. Furthermore, the results also demonstrated that the RZWQM2 is a promising tool for evaluating the long-term effects of CO 2 and N 2 O emissions on different conservation practices.

54 ENVIRONMENTAL SCIENCES↗