Browsing by Author "Yang, Z"

Now showing 1 - 5 of 5
  • No Thumbnail Available
    Item
    Auto-MatRegressor: liberating machine learning alchemists
    (Elsevier, 2023-06-30) Liu, Y; Wang, S; Yang, Z; Avdeev, M; Shi, S
    Machine learning (ML) is widely used to uncover structure–property relationships of materials due to its ability to quickly find potential data patterns and make accurate predictions. However, like alchemists, materials scientists are plagued by time-consuming and labor-intensive experiments to build high-accuracy ML models. Here, we propose an automatic modeling method based on meta-learning for materials property prediction named Auto-MatRegressor, which automates algorithm selection and hyperparameter optimization by learning from previous modeling experience, i.e., meta-data on historical datasets. The meta-data used in this work consists of 27 meta-features that characterize the datasets and the prediction performances of 18 algorithms commonly used in materials science. To recommend optimal algorithms, a collaborative meta-learning method embedded with domain knowledge quantified by a materials categories tree is designed. Experiments on 60 datasets show that compared with the traditional modeling method from scratch, Auto-MatRegressor automatically selects appropriate algorithms at lower computational cost, which accelerates constructing ML models with good prediction accuracy. Auto-MatRegressor supports dynamic expansion of meta-data with the increase of the number of materials datasets and other required algorithms and can be applied to any ML materials discovery and design task. © 2023 Science China Press. Published by Elsevier B.V. and Science China Press. All rights reserved.
  • No Thumbnail Available
    Item
    An automatic descriptors recognizer customized for materials science literature
    (Elsevier, 2022-10) Liu, Y; Ge, X; Yang, Z; Sun, S; Liu, D; Avdeev, M; Shi, SH
    Materials science literature contains domain knowledge about numerous descriptors, which play a critical role in data-driven materials design. However, automatically extracting descriptors from literature remains challenging. Here, we develop an automatic descriptors recognizer based on natural language processing (NLP) to mine latent descriptors, which consists of a conditional data augmentation model incorporating materials domain knowledge (cDA-DK), coarse- and fine-grained descriptors subrecognizers (CGDR and FGDR). cDA-DK conducts augmenting training data of text mining model, which can significantly reduce the cost of manually labeling and enhance the robustness of its model. On this basis, CGDR recognizes coarse-grained descriptor entities automatically, and FGDR performs screening of fine-grained descriptors related to specific materials design. Following this, the activation energy of NASICON-type solid electrolytes, which is influenced by complicated descriptors, is taken as an example to demonstrate the potential utility of our recognizer. CGDR extracts 106896 descriptor entities from 1808 relevant articles with an accuracy (F1) of 0.87. Furthermore, with features from 408 descriptors screened by FGDR, six activation energy prediction models are constructed to perform experiments, achieving an optimal prediction performance (R2) of 0.96. This work provides important insight towards the understanding of structure-activity relationships, thus promoting materials design and discovery. © 2022 Elsevier B.V.
  • Thumbnail Image
    Item
    Effect of ingestion temperature on the pepsin-induced coagulation and the in vitro gastric digestion behavior of milk
    (Elsevier B. V., 2023-05) Yang, MX; Ye, AQ; Yang, Z; Everett, DW; Gilbert, EP; Singh, H
    Pepsin-induced protein coagulation occurs in the gastric environment when the milk pH is above the isoelectric point of casein proteins. In this study, the effect of milk temperature (4–48 °C) on the hydrolysis of κ-casein by pepsin and the consequent protein coagulation was studied at pH 6.0 for 120 min. Quantitative determination of the released para-κ-casein showed that both the κ-casein hydrolysis reaction rate constant and the pepsin denaturation rate constant increased with an increase in temperature. The temperature coefficient (Q10) of the specific hydrolysis of κ-casein was calculated to be ∼1.95. The coagulation process was investigated by the evolution of the storage modulus (Gʹ). At higher temperature, the milk coagulated faster but had a lower firming rate and Gʹmax with larger aggregates and voids were observed. The digestion behavior of the milk ingested at 4 °C, 37 °C, or 50 °C was investigated for 240 min in a human gastric simulator, in which the milk temperature increased or decreased to 37 °C (body temperature) over ∼ 60 min. The coagulation of the 4 °C milk was slower than for the 37 °C and 50 °C milk. The curd obtained from the 4 °C milk had a looser and softer structure with a significantly higher moisture content at the initial stage of digestion (20 min) which, in turn, facilitated the breakdown and hydrolysis of the caseins by pepsin. During the digestion, the curd structure became more cohesive, along with a decrease in moisture content. The knowledge gained from this study provides insight into the effect of temperature on the kinetics of pepsin-induced milk coagulation and the consequent digestion behavior. © The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync- nd/4.0/).
  • No Thumbnail Available
    Item
    Small-angle X-ray scattering (SAXS) and small-angle neutron scattering (SANS) study on the structure of sodium caseinate in dispersions and at the oil-water interface: effect of calcium ions
    (Elsevier B. V., 2022-04) Cheng, LR; Ye, AQ; Yang, Z; Gilbert, EP; Knott, RB; de Campo, L; Storer, B; Hemar, Y; Singh, H
    The structure of sodium caseinate particles, as affected by the presence of calcium ions (Ca2+), in aqueous solution and in oil (toluene)-in-water emulsions, was investigated by small-angle X-ray and neutron scattering (SAXS and SANS). SAXS analyses indicated that the sodium caseinate dispersed in water as small particles with electrostatic interactions, which has a radius of gyration (Rg) of ~5 nm and an effective radius (Reff) of ~ 10 nm with an assuming spherical shape. In the presence of Ca2+, the caseinate particles aggregated as large particles with a hydrodynamic diameter > 100 nm as determined by dynamic light scattering. The networks within the large particles were self-assembled from the small Ca2+-cross-linked particles (Rg ~ 6.5–8.0 nm), as probed by SAXS. The fractal-like dimension increased from 2.5 to 3.4 with increasing protein and CaCl2 concentrations, suggesting a denser structure. The integrity of the caseinate particles at the oil-water interface was enhanced by Ca2+ cross-linking, as observed by transmission electron microscopy. The oilsingle bondwater interface stabilised by Ca2+-cross-linked caseinate particles was ~ 30 nm thick, six times thicker than that stabilised by sodium caseinate (~ 5 nm) as analysed by SANS with contrast variation technique. Quantifying the structure of sodium caseinate in an aqueous solution and at the oil-water interface provides valuable insights for designing new casein-based functional materials. © 2022 Elsevier Ltd
  • No Thumbnail Available
    Item
    Structural and electrochemical impacts of Mg/Mn dual dopants on the LiNiO2 cathode in Li-metal batteries
    (American Chemical Society, 2020-03-04) Mu, L; Kan, WH; Kuai, C; Yang, Z; Li, LX; Sun, CJ; Sainio, S; Avdeev, M; Nordlund, D; Lin, F
    Doping chemistry has been regarded as an efficient strategy to overcome some fundamental challenges facing the “no-cobalt” LiNiO2 cathode materials. By utilizing the doping chemistry, we evaluate the battery performance and structural/chemical reversibility of a new no-cobalt cathode material (Mg/Mn-LiNiO2). The unique dual dopants drive Mg and Mn to occupy the Li site and Ni site, respectively. The Mg/Mn-LiNiO2 cathode delivers smooth voltage profiles, enhanced structural stability, elevated self-discharge resistance, and inhibited nickel dissolution. As a result, the Mg/Mn-LiNiO2 cathode enables improved cycling stability in lithium metal batteries with the conventional carbonate electrolyte: 80% capacity retention after 350 cycles at C/3, and 67% capacity retention after 500 cycles at 2C (22 °C). We then take the Mg/Mn-LiNiO2 as the platform to investigate the local structural and chemical reversibility, where we identify that the irreversibility takes place starting from the very first cycle. The highly reactive surface induces the surface oxygen loss, metal reduction reaching the subsurface, and metal dissolution. Our data demonstrate that the dual dopants can, to some degree, mitigate the irreversibility and improve the cycling stability of LiNiO2, but more efforts are needed to eliminate the key challenges of these materials for battery operation in the conventional carbonate electrolyte. © 2020 American Chemical Society