Adverse drug-drug interactions (DDIs) remain a leading cause of morbidity and mortality. Identifying potential DDIs during the drug design process is critical for patients and society. Although several computational models have been proposed for DDI prediction, there are still limitations: (1) specialized design of drug representation for DDI predictions is lacking; (2) predictions are based on limited labelled data and do not generalize well to unseen drugs or DDIs; and (3) models are characterized by a large number of parameters, thus are hard to interpret. In this work, we develop a ChemicAl SubstrucTurE Representation (CASTER) framework that predicts DDIs given chemical structures of drugs. CASTER aims to mitigate these limitations via (1) a sequential pattern mining module rooted in the DDI mechanism to efficiently characterize functional sub-structures of drugs; (2) an auto-encoding module that leverages both labelled and unlabelled chemical structure data to improve predictive accuracy and generalizability; and (3) a dictionary learning module that explains the prediction via a small set of coefficients which measure the relevance of each input sub-structures to the DDI outcome. We evaluated CASTER on two real-world DDI datasets and showed that it performed better than state-of-the-art baselines and provided interpretable predictions.
Abstract. A fundamental issue in active learning of Gaussian processes is that of the exploration-exploitation trade-off. This paper presents a novel nonmyopic -Bayes-optimal active learning ( -BAL) approach [4] that jointly optimizes the trade-off. In contrast, existing works have primarily developed greedy algorithms or performed exploration and exploitation separately. To perform active learning in real time, we then propose an anytime algorithm [4] based on -BAL with performance guarantee and empirically demonstrate using a real-world dataset that, with limited budget, it outperforms the state-of-the-art algorithms.
Vietnam is the world's fifth largest rice producing country. Since methane (CH 4 ), a potent greenhouse gas (GHG), emission from the rice cultivation accounts for 14.6% of the national anthropogenic GHG emission, developing and disseminating mitigation options are the urgent need. Alternate wetting and drying (AWD) is the irrigation technique, in which a paddy field encompasses several soil-drying phases during the growth period, thereby reducing the CH 4 emission. However, field trials of the AWD's feasibility in Central Vietnam are limited so far. We therefore carried out a 3-year experiment in a farmer's field both in winter-spring season and summer-autumn season. CH 4 and nitrous oxide (N 2 O) emissions were compared among the three treatments of water management: continuous flooding (CF), AWD, and site-specific AWD (AWDS) that changed the degree of soil drying depending on the growth stage. The total water use including irrigation and rainfall was significantly (p < 0.05) reduced by AWD (by 15%) and AWDS (by 14%) compared to CF, but rice grain yield did not differ among the three treatments. The seasonal cumulative CH 4 emission was significantly reduced by AWD (26%) and AWDS (26%) compared to CF, whereas the seasonal cumulative N 2 O emission did not differ among treatments. The resultant global warming potentials (GWPs) of CH 4 and N 2 O under AWD and AWDS were 26% and 29% smaller than that under CF, respectively. The GWP of N 2 O was only 0.8% of the total GWP of CH 4 + N 2 O. The yield-scaled GWP and water productivity (i.e., the ratio of grain yield to water use) were also improved by AWD and AWDS. No significant differences in the measured items between AWD and AWDS were attributed to similar variation patterns in the surface water level. The results confirm the AWD's performance as a mitigation option for paddy GHG emission in Central Vietnam. ARTICLE HISTORY
This study comprises field experiments on methane emissions from rice fields conducted with an Eddy-Covariance (EC) system as well as test runs for a modified closed chamber approach based on measurements at nighttime. The EC data set covers 4 cropping seasons with highly resolved emission rates (raw data in 10 Hz frequency have been aggregated to 30-min records). The diel patterns were very pronounced in the two dry seasons with peak emissions at early afternoon and low emissions at nighttime. These diel patterns were observed at all growing stages of the dry seasons. In the two wet seasons, the diel patterns were only visible during the vegetative stages while emission rates during reproductive and ripening stages remained within a fairly steady range and did not show any diel patterns. In totality, however, the data set revealed a very strong linear relationship between nocturnal emissions (12-h periods) and the full 24-h periods resulting in an R2-value of 0.8419 for all data points. In the second experiment, we conducted test runs for chamber measurements at nighttime with much longer deployment times (6 h) as compared to measurements at daylight (typically for 30 min). Conducting chamber measurements at nighttime excluded drastic changes of temperatures and CO2 concentrations. The data also shows that increases in CH4 concentrations remained on linear trajectory over a 6h period at night. While end CH4 concentrations were consistently >3.5 ppm, this long-term enclosure represents a very robust approach to quantify emissions as compared to assessing short-term concentration increases over time near the analytical detection limit. Finally, we have discussed the potential applications of this new approach that would allow emission measurements even when conventional (daytime) measurements will not be suitable. Nighttime chamber measurements offer an alternative to conventional (daytime) measurements if either (i) baseline emissions are at a very low level, (ii) differences of tested crop treatments or varieties are very small or (iii) the objective is to screen a large number of rice varieties for taking advantage of progress in genome sequencing.
This paper presents an overview of our novel decision-theoretic multi-agent approach for controlling and coordinating multiple active cameras in surveillance. In this approach, a surveillance task is modeled as a stochastic optimization problem, where the active cameras are controlled and coordinated to achieve the desired surveillance goal in presence of uncertainties. We enumerate the practical issues in active camera surveillance and discuss how these issues are addressed in our decision-theoretic approach. We focus on two novel surveillance tasks: maximize the number of targets observed in active cameras with guaranteed image resolution and to improve the fairness in observation of multiple targets. We discuss the overview of our novel decision-theoretic frameworks: Markov Decision Process and Partially Observable Markov Decision Process frameworks for coordinating active cameras in uncertain and partially occluded environments.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.