Browse Articles
Discover research articles across all indexed journals
The inner ear is a barometric pressure sensor—change in barometric pressure induces vestibular ganglion cell activation in mice
Alkane Coordination by a Neutral, Lewis Acidic Magnesium Complex
Optimization of technological parameters for durum wheat pasta production with carrot powder and ion-ozonated water
Abstract The quality of food products and a balanced diet play a key role in maintaining health and improving the quality of life. Due to the high popularity of pasta, a promising direction is the development of recipes that include whole grain flour, carrot powder and ion-ozonated water, which increase the nutritional value of the product and improve its chemical composition. The aim of this study was to optimise the technological modes of production of durum wheat pasta using carrot powder and ion-ozonated water to minimise the loss of dry matter (DM) during cooking and increase the protein content in the finished product. We used multifactorial experiment planning with the sequential regression analysis program PLAN. The variable factors considered were the concentration of ion-ozone in the water (C io ), the carrot powder content (C cp. ), the temperature of the water (t w ) used in kneading the dough and the drying temperature (t d ). We found that a minimal DM loss into the cooking water (5.86%) was ensured using a C cp. of 3.0% and a t d of 50 °C. Neither the C io nor the t w had any significant effect on this formula. To achieve the maximum protein content in the finished pasta (14.25%), the optimal parameters used were C cp. = 1.0% and t d = 50 °C. Similarly, the C io and t w factors had no significant effect on the protein composition. These results demonstrate the possibility of controlling the quality indicators of pasta by regulating temperature parameters and the concentration of plant additives. To increase the nutritional value of pasta products without deteriorating the technological properties, it is recommended to use carrot powder at a concentration of 1.0–3.0% and dry it at 50 °C. The use of ion-ozonated water is justified from the point of view of safety and microbiological stability, but it did not have a significant impact on the quality indicators studied.
Bidirectional reinforcement learning neural network for constrained molecular design
Abstract We present BiRLNN, a bidirectional molecular design framework that combines recurrent neural networks with reinforcement learning to optimize drug-like properties of generated compounds. We examined the use of Self-Referencing Embedded Strings representations, which ensures 100% syntactic validity of generated molecules. By generating molecular sequences in both forward and backward directions, we enabled more balanced exploration of chemical space while maintaining constraint requirements during molecular design. To guide generation towards desirable pharmacological targets, we implement a multi-objective reward function based on quantitative estimate of drug-likeness and synthetic accessibility, and apply policy gradient-based reinforcement learning for fine-tuning. We demonstrate that our bidirectional model covers the full constrained chemical space compared to unidirectional ones using pharmaceutically relevant fragments, allowing it to explore regions containing molecules unreachable by the latter. Moreover, the reinforcement learning process successfully steers the constrained generation process toward desirable compound classes with improved reward metrics. Our results demonstrate that BiRLNN offers a robust and flexible strategy for navigating chemical space in multi-objective drug design tasks.
Performance optimization of rigid pavement concrete using metakaolin treated RCA and silica fume with an experimental and machine learning based approach
Abstract Concrete production has a drastic effect on the environment with ordinary portland cement (OPC) contributing approximately 6–8% to the world CO 2 emissions. The best solutions to reduce these effects is the use of recycled concrete aggregate (RCA) and secondary cementitious materials (SCM) as an alternative to natural aggregates and OPC. Nevertheless, RCA based on low-strength parent concrete is normally characterized by high porosity, low-bond mortar and low mechanical performance which restricts the scope of its structural use. This paper examines the improvement of RCA-based concrete using metakaolin (MK) slurry treatment and addition of silica fume (SF) at different dosages (2.5–10%) with the replacement contents of RCA being 0, 50, 75, and 100%. It has been experimentally found that the addition of MK and SF can significantly enhance mechanical strength and durability and adequately address the intrinsic weaknesses of low-grade RCA. The statistical validation with one-way ANOVA showed that all the P -values were less than 0.05 and proved that the improvements due to the addition of SCM and the adjustment of RCA were significant. In addition, 96 experimental and 48 literature-based datasets were used to predict and optimize compressive strength with the use of machine learning (ML) models. K-fold cross-validation was used to fine-tune hyperparameters and the Grey Wolf Optimizer (GWO) was used to optimize them. Extreme Gradient Boosting (XGB) gave the best accuracy with the highest R 2 of 0.949 (training) and 0.899 (testing) and low RMSE of 1.490 and 1.845 respectively. AdaBoost (ADB) also provided satisfactory results after XGB (R 2 = 0.929 training, 0.878 testing). In general, the findings substantiate the claim that the ensemble learning frameworks especially XGB are quite effective to derive complex relationships in RCA-based concrete data. RCA, SCMs (MK and SF) and predictive ML modeling can provide a sustainable mix design optimization route and structural life enhancement of RCA concrete in rigid pavement applications.