RESUMEN
Convolutional Neural Networks (CNNs) have been widely applied in various edge computing devices based on intelligent sensors. However, due to the high computational demands of CNN tasks, the limited computing resources of edge intelligent terminal devices, and significant architectural differences among these devices, it is challenging for edge devices to independently execute inference tasks locally. Collaborative inference among edge terminal devices can effectively utilize idle computing and storage resources and optimize latency characteristics, thus significantly addressing the challenges posed by the computational intensity of CNNs. This paper targets efficient collaborative execution of CNN inference tasks among heterogeneous and resource-constrained edge terminal devices. We propose a pre-partitioning deployment method for CNNs based on critical operator layers, and optimize the system bottleneck latency during pipeline parallelism using data compression, queuing, and "micro-shifting" techniques. Experimental results demonstrate that our method achieves significant acceleration in CNN inference within heterogeneous environments, improving performance by 71.6% compared to existing popular frameworks.
RESUMEN
Due to the potential applications in next-generation micro/nano electronic devices and functional materials, magnetic germanium (Ge)-based clusters are receiving increasing attention. In this work, we reported the structures, electronic and magnetic properties of CrMnGen with sizes n = 3-20. Transition metals (TMs) of Cr and Mn tend to stay together and be surrounded by Ge atoms. Small sized clusters with n ≤ 8 prefer to adopt bipyramid-based structures as the motifs with the excess Ge atoms absorbed on the surface. Starting from n = 9, the structure with one TM atom interior appears and persists until n = 16, and for larger sizes n = 17-20, the two TM atoms are full-encapsulated by Ge atoms to form endohedral structures. The Hirshfeld population analyses show that Cr atom always acts as the electron donor, while Mn atom is always the acceptor except for sizes 3 and 6. The average binding energies of these clusters increase with cluster size n, sharing a very similar trend as that of CrMnSin (n = 4-20) clusters, which indicates that it is favorable to form large-sized clusters. CrMnGen (n = 6, 13, 16, 19, and 20) clusters prefer to exhibit ferromagnetic Cr-Mn coupling, while the remaining clusters are ferrimagnetic.
RESUMEN
Epigallocatechin gallate (EGCG), an active constituent of tea, is recognized for its anticancer and anti-inflammatory properties. However, the specific mechanism by which EGCG protects osteoblasts from cadmium-induced damage remains incompletely understood. Here, the action of EGCG was investigated by exposing MC3T3-E1 osteoblasts to EGCG and CdCl2 and examining their growth, apoptosis, and differentiation. It was found that EGCG promoted the viability of cadmium-exposed MC3T3-E1 cells, mitigated apoptosis, and promoted both maturation and mineralization. Additionally, CdCl2 has been reported to inhibit both the phosphoinositide 3-kinase/protein kinase B/mammalian target of rapamycin (PI3K/AKT/mTOR) and nuclear factor erythroid 2-related factor 2/heme oxygenase-1(Nrf2/HO-1) signaling pathways. EGCG treatment attenuated cadmium-induced apoptosis in osteoblasts and restored their function by upregulating both signaling pathways. The findings provide compelling evidence for EGCG's role in attenuating cadmium-induced osteoblast apoptosis and dysfunction through activating the PI3K/AKT/mTOR and Nrf2/HO-1 pathways. This suggests the potential of using EGCG for treating cadmium-induced osteoblast dysfunction.
Asunto(s)
Apoptosis , Catequina , Hemo-Oxigenasa 1 , Factor 2 Relacionado con NF-E2 , Osteoblastos , Fosfatidilinositol 3-Quinasas , Proteínas Proto-Oncogénicas c-akt , Transducción de Señal , Serina-Treonina Quinasas TOR , Catequina/análogos & derivados , Catequina/farmacología , Apoptosis/efectos de los fármacos , Factor 2 Relacionado con NF-E2/metabolismo , Animales , Ratones , Serina-Treonina Quinasas TOR/metabolismo , Transducción de Señal/efectos de los fármacos , Hemo-Oxigenasa 1/metabolismo , Proteínas Proto-Oncogénicas c-akt/metabolismo , Fosfatidilinositol 3-Quinasas/metabolismo , Osteoblastos/efectos de los fármacos , Osteoblastos/metabolismo , Cadmio/toxicidad , Diferenciación Celular/efectos de los fármacos , Línea Celular , Proteínas de la MembranaRESUMEN
BACKGROUND: Eucommia ulmoides Oliver is a unique high-quality natural rubber tree species and rare medicinal tree species in China. The rapid characterization of E. ulmoides gene function has been severely hampered by the limitations of genetic transformation methods and breeding cycles. The polyethylene glycol (PEG)-mediated protoplast transformation system is a multifunctional and rapid tool for the analysis of functional genes in vivo, but it has not been established in E. ulmoides. METHODS: In this study, a large number of highly active protoplasts were isolated from the stems of E. ulmoides seedlings by enzymatic digestion, and green fluorescent protein expression was facilitated using a PEG-mediated method. RESULTS: Optimal enzymatic digestion occurred when the enzyme was digested for 10 h in an enzymatic solution containing 2.5% Cellulase R-10 (w/v), 0.6% Macerozyme R-10 (w/v), 2.5% pectinase (w/v), 0.5% hemicellulase (w/v), and 0.6 mol/L mannitol. The active protoplast yield under this condition was 1.13 × 106 protoplasts/g fresh weight, and the protoplast activity was as high as 94.84%. CONCLUSIONS: This study established the first protoplasm isolation and transient transformation system in hard rubber wood, which lays the foundation for subsequent functional studies of E. ulmoides genes to achieve high-throughput analysis, and provides a reference for future gene function studies of medicinal and woody plants.
Asunto(s)
Eucommiaceae , Protoplastos , Transfección , Protoplastos/metabolismo , Eucommiaceae/genética , Eucommiaceae/metabolismo , Transfección/métodos , Proteínas Fluorescentes Verdes/genética , Proteínas Fluorescentes Verdes/metabolismo , PolietilenglicolesRESUMEN
The trend toward further miniaturization of micronano antiferromagnetic (AFM) spintronic devices has led to a strong demand for low-dimensional materials. The assembly of AFM clusters to produce such materials is a potential pathway that promotes studies on such clusters. In this work, we report on the discovery of the AFM Cr2Snx (x = 3-20) clusters with a stepwise growth at the density functional theory (DFT) level. In comparison, the two Cr atoms tend to stay together and be buried by Sn atoms, forming endohedral structures with one Cr atom encapsulated at size 9 and finally forming a full-encapsulated structure at size 17. Each successive cluster size is composed of its predecessor with an extra Sn atom adsorbed onto the face, giving evidence of stepwise growth. All these Cr2Snx (x = 3-20) clusters are antiferromagnets, except for the triplet-state ferrimagnetic Cr2Sn11, and all their singly negatively and positively charged ions are ferromagnets. The found stable Cr2Sn17 cluster can dimerize, yielding dimers and trimers without noticeably distorting the geometrical structure and magnetic properties of each of its constituent cluster monomers, making it possible as a building block for AFM materials.
RESUMEN
Magnesium-aluminum (Mg-Al) alloys are widely used in aerospace, automobile and medical equipment owing to their advantages of easy casting, high strength-to-mass ratio and good biocompatibility. The structural, mechanical, electronic and thermodynamic properties of MgxAly alloys (x + y = 16, x = 1, 2, , 15) with varying Al-doping contents were studied using the first-principles method. In this work, the structures of MgxAly alloys were constructed by replacing Mg atoms in a supercell with Al atoms. The lattice parameters of the Al-doped MgxAly alloys decrease with an increasing Al content because of the smaller atomic size of Al than that of Mg. The calculated formation energies show that Mg11Al5, Mg5Al3 and Mg9Al7 have prominent structural stability. The analyses of the mechanical properties reveal that the doping of Al improves the ductility of MgxAly alloys. The elastic moduli increase with an increasing Al content, and Mg9Al7 has a notable ability to resist deformation, while Mg11Al5 and Mg5Al3 have better plasticity. The calculated results of their electronic properties reveal that Mg11Al5, Mg5Al3 and Mg9Al7 are good conductors without magnetism. Furthermore, CDD analyses show that the inner layer charges of Al atoms migrated to the outer layer, and the charges of Mg atoms accumulated significantly in the outer region of Al atoms. The Debye temperature of Mg9Al7 is higher than that of Mg11Al5 and Mg5Al3, indicating that it has better thermodynamic stability. Our findings would be helpful for the design of Mg-Al alloys with excellent mechanical and thermodynamic performances.
RESUMEN
Herein, the structural evolution, electronic and magnetic properties of silicon clusters with two different dopants, CrMnSin (n = 4-20) clusters were investigated at density functional theory (DFT) level. Small-sized CrMnSin (n = 4-9) clusters tend to adopt bipyramid-based geometries, while clusters with sizes n = 10 and 11 prefer to opening cage-like structures. For sizes n = 12 to 14, the half-encapsulated structures gradually transform into closed-cage Cr@Sin structures, with the Mn atom exposed outside. Starting from size 15, both the Cr and Mn atoms are completely encapsulated by silicon atoms. Meanwhile, the Cr and Mn atoms in smaller-sized CrMnSin (n = 4-7) clusters tend to be separated, while they prefer to stay together for larger sizes. Cr atom always acts as electron donor, but not for Mn atom. From the average binding energies, one can conclude that it is easier to form larger size clusters. Smaller and larger sized CrMnSin (n = 4-9 and 19-20) clusters prefer to exhibit ferromagnetic Cr-Mn coupling, while sizes n = 10-18 always exhibit ferrimagnetic state. To our knowledge, the CrMnSin clusters is the first kind of neutral transition-metal doped semiconductor clusters that show ferrimagnetic state within a wide size range.
RESUMEN
In this work, an unbiased global search with a homemade genetic algorithm was performed to investigate the structural evolution and electronic properties of Snx - (x=21-35) clusters with density functional theory (DFT) calculations. All the ground-state structures for all these Snx - (x=21-35) clusters have been confirmed by the comparison of the experimental and simulated photoelectron spectra (PESs). It has been revealed that all Snx - (x=21-35) clusters are tricapped trigonal prism (TTP)-based structures consisting of two (for sizes x=21-28) or three (for x=29-35) TTP units, with the remaining atoms adsorbed on the surface or inserted between TTP units. The gradually decreasing HOMO-LUMO gaps indicate that these clusters are undergoing semiconductor-to-metal transformation. The average binding energies show that the structural stabilities of Snx - clusters are not as good as that of silicon and germanium clusters. It found that sizes x=23, 25, 29, 33 show high relative stability.
RESUMEN
Nowadays, global public health crises are occurring more frequently, and accurate prediction of these diseases can reduce the burden on the healthcare system. Taking COVID-19 as an example, accurate prediction of infection can assist experts in effectively allocating medical resources and diagnosing diseases. Currently, scholars worldwide use single model approaches or epidemiology models more often to predict the outbreak trend of COVID-19, resulting in poor prediction accuracy. Although a few studies have employed ensemble models, there is still room for improvement in their performance. In addition, there are only a few models that use the laboratory results of patients to predict COVID-19 infection. To address these issues, research efforts should focus on improving disease prediction performance and expanding the use of medical disease prediction models. In this paper, we propose an innovative deep learning model Whale Optimization Convolutional Neural Networks (CNN), Long-Short Term Memory (LSTM) and Artificial Neural Network (ANN) called WOCLSA which incorporates three models ANN, CNN and LSTM. The WOCLSA model utilizes the Whale Optimization Algorithm to optimize the neuron number, dropout and batch size parameters in the integrated model of ANN, CNN and LSTM, thereby finding the global optimal solution parameters. WOCLSA employs 18 patient indicators as predictors, and compares its results with three other ensemble deep learning models. All models were validated with train-test split approaches. We evaluate and compare our proposed model and other models using accuracy, F1 score, recall, AUC and precision metrics. Through many studies and tests, our results show that our prediction models can identify patients with COVID-19 infection at the AUC of 91%, 91%, and 93% respectively. Other prediction results achieve a respectable accuracy of 92.82%, 92.79%, and 91.66% respectively, f1-score of 93.41%, 92.79%, and 92.33% respectively, precision of 93.41%, 92.79%, and 92.33% respectively, recall of 93.41%, 92.79%, and 92.33% respectively. All of these exceed 91%, surpassing those of comparable models. The execution time of WOCLSA is also an advantage. Therefore, the WOCLSA ensemble model can be used to assist in verifying laboratory research results and predict and to judge various diseases in public health events.
Asunto(s)
COVID-19 , Aprendizaje Profundo , Animales , COVID-19/diagnóstico , Algoritmos , Benchmarking , CetáceosRESUMEN
The chemoselective annulation of aza-ortho-quinone methide generated by in situ o-chloromethyl sulfonamide has been achieved with bifunctional acyclic olefin. This efficient approach provides access to the diastereoselective synthesis of functionalized tetrahydroquinoline derivatives containing indole scaffolds through the inverse-electron-demand aza-Diels-Alder reaction under mild reaction conditions with excellent results (up to 93% yield, > 20:1 dr). Moreover, this article realized the cyclization of α-halogeno hydrazone with electron-deficient alkene affording the tetrahydropyridazine derivatives, which had never been reported.
RESUMEN
This study investigates the relationship between ITGAL expression and immune infiltration, clinical prognosis, and specific types of T cells in melanoma tissue. The findings reveal the key role of ITGAL in melanoma and its potential mechanism of regulating tumor immune infiltrating cells, highlighting its potential as a diagnostic biomarker and therapeutic target for advanced melanoma.
RESUMEN
Enterovirus 71 (EV71) infection mainly causes hand, foot, and mouth disease (HFMD) and remains a serious public health problem to the children under the age of 5. Until now, there is no specific drug to treat HFMD in clinical and there is an urgent to explore the new target and the new drug to address clinical challenges. At present, we found histone deacetylase 11 (HDAC11) involves in supporting EV71 replication. We also used HDAC11 siRNA and an HDAC11 inhibitor FT895 to downregulate HDAC11 expression and found that targeting HDAC11 could significantly restrict EV71 replication in vitro and in vivo. Our results revealed the new role of HDAC11 participating in EV71 replication and broadened our knowledge regarding the functions of HDAC11 and the roles of HDACs in the epigenetic regulation of viral infectious diseases. Our results for the first time identified FT895 as an effective inhibitor of EV71 in vitro and in vivo, which may contribute to be a potential drug to treat HFMD.
Asunto(s)
Enterovirus Humano A , Infecciones por Enterovirus , Enterovirus , Enfermedad de Boca, Mano y Pie , Niño , Humanos , Enterovirus Humano A/genética , Epigénesis Genética , Histona Desacetilasas/genética , Histona Desacetilasas/farmacologíaRESUMEN
Osteoporosis is a serious systemic metabolic bone system disease.This study aimed to identify the target genes of isopsoralen and the signaling pathways involved in the differential expression of the genes involved in osteoclast differentiation. We hypothesized that isopsoralen may inhibit osteoclast differentiation by blocking the nuclear factor kappa-B (NF-κB) signaling pathway and verified our hypothesis through basic experiments. The 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay was used to detect the effect of isopsoralen on the proliferation and viability of primary mouse bone marrow monocytes (BMMCs). The effect of isopsoralen on receptor activator of nuclear factor kappa-B ligand (RANKL)-induced osteoclast differentiation was determined by using tartrate-resistant acid phosphatase (TRAP) staining. Quantitative real-time PCR (qRT-PCR) and Western blot were used to detect the expression of the related genes and proteins. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway of isopsoralen target genes were obtained through comprehensive analysis using the STITCH database, Cytoscape 3.8.2, and R-Studio software. Differentially expressed genes (DEGs) were found in osteoclasts induced by RANKL before and after 3 days using R-Studio, following which KEGG analysis was performed. Next, enrichment analysis was performed on the KEGG pathway shared by the target genes of isopsoralen and the differentially expressed genes during osteoclast differentiation to predict the signaling pathway underlying the inhibition of osteoclast differentiation by isopsoralen. Finally, Western blot was used to detect the effect of isopsoralen on the activation of signaling pathways to verify the results of our bioinformatics analysis. Based on the enrichment analysis of isopsoralen target genes and differentially expressed genes during osteoclastogenesis, we believe that isopsoralen can inhibit RANKL-induced osteoclastogenesis by inhibiting the NF-κB signaling pathway.
Asunto(s)
FN-kappa B , Osteogénesis , Ratones , Animales , FN-kappa B/genética , Ligandos , Transducción de SeñalRESUMEN
Plants actively respond to their neighbors by altering root placement patterns. Neighbor-modulated root responses involve root detection and interactions mediated by root-secreted functional metabolites. However, chemically mediated root placement patterns and their underlying mechanisms remain elusive. We used an allelopathic wheat model system challenged with 60 target species to identify root placement responses in window rhizobox experiments. We then tested root responses and their biochemical mechanisms in incubation experiments involving the addition of activated carbon and functional metabolites with amyloplast staining and auxin localization in roots. Wheat and each target species demonstrated intrusive, avoidant or unresponsive root placement, resulting in a total of nine combined patterns. Root placement patterns were mediated by wheat allelochemicals and (-)-loliolide signaling of neighbor species. In particular, (-)-loliolide triggered wheat allelochemical production that altered root growth and placement, degraded starch grains in the root cap and induced uneven distribution of auxin in target species roots. Root placement patterns in wheat-neighbor interactions were perception dependent and species dependent. Signaling (-)-loliolide induced the production and release of wheat allelochemicals that modulated root placement patterns. Therefore, root placement patterns are generated by both signaling chemicals and allelochemicals in allelopathic plant-plant interactions.
Asunto(s)
Plantas , Triticum , Plantas/metabolismo , Triticum/metabolismo , Ácidos Indolacéticos/metabolismo , Alelopatía , Feromonas/metabolismo , Raíces de Plantas/metabolismoRESUMEN
An efficient [4 + 1] annulation reaction between in situ generated azoalkene intermediates and α-bromocarbonyls has been established. A series of skeletally diverse aza-heterocycles with a functionalized quaternary center were obtained in up to 89% yield under mild conditions.
RESUMEN
The aim of this study was to investigate the responses of non-starch polysaccharide (NSP) enzymes and protease combination on growth performance, meat quality, and nutrients digestibility of yellow-feathered broilers fed with corn-soybean meal basal diets with normal and subnormal crude protein (CP) levels. The experimental design was completely randomized with a 2 × 2 factorial arrangement of treatments, including six replicates of 20 birds per pen. Two basal diets were formulated with normal CP level as positive control (PC) and subnormal CP level without extra essential amino acid (AA) supplementation as negative control (NC). The basal diets were supplemented without or with NSP enzymes and protease. Broilers fed with the NC diet had lower (P < 0.05) final body weight (BW), average daily weight gain (ADG) on days 1-21, 22-56 and 1-56 and higher (P < 0.05) feed-to-gain ratio (F/G) on day 22-56 than those fed with PC diet. The broilers fed with the NC diet had higher (P < 0.05) L* and b* values in thigh muscle, crypt depth in the duodenum, and dry matter (DM) digestibility as well as lower (P < 0.05) villus height, musculature thicknesses, and villus height: crypt depth in the duodenum than those fed with the PC diet. Dietary NSP enzymes and protease combination increased (P < 0.05) the ADG and F/G of the broilers on days 1-56, and pH values in breast and thigh muscles as well as the digestibility of DM, gross energy (GE), CP and most AAs of the broilers on day 56. Compared with the PC diet, the combination of NSP enzymes and protease exhibited greater (P < 0.05) improvements in the digestibility of DM, CP, and some AAs (Asp, Ile, and Leu) in the broilers fed with the NC diet. In conclusion, reducing CP diet without essential AA supplementation impaired the growth performance and meat color of the thigh muscles of the broilers. The combination of NSP enzymes and protease effectively improved the growth performance, meat quality, and nutritional values of the broilers. In terms of the digestibility of DM, CP, and some AAs, the magnitude of response to the addition of NSP enzymes and protease was greater in the low nutritional-quality diet with a subnormal CP level.
RESUMEN
Spermatogonial stem cells (SSCs) are the basis of spermatogenesis. Systematically exploring the critical factors associated with the formation of SSCs will provide new insight to improve the formation efficiency, and their practical application. Here we explore the regulatory mechanism of the ECM-receptor interaction signaling pathway and related genes during differentiation of SSCs in chicken. Firstly, the positive cell rate of SSCs protein marker was detected by immunofluorescence and flow cytometry and qRT-PCR was used to identify, the expression of related marker genes after 10 days of RA-induction. Secondly, the ESCs on 0d/ 4d /10d after RA- induction/self-differentiation were collected, and the total RNA was then extracted from cells. Finally, high-throughput analysis methods (RNA-seq) were used to sequence the transcriptome of these cells. After PCA analysis of the RNA-seq data, Venny analysis, GO and KEGG enrichment were further used to find the key signaling pathways and genes in the RA-induction process. The results showed that on day 10 of RA-induction, grape cluster growth cells expressed integrinß1, the specific marker protein of SSCs cells, and the integrinß1 positive rate was 35.1%. Also, SSCs marker genes CVH, Integrinß1, Integrinα6 were significantly up-regulated during RA-induction. Moreover, the significantly enriched pathway, ECM-receptor interaction signaling, in current study may play a crucial role in RA-induction. Then, JASPAR was used to predict the differential gene transcription factors in the signaling pathway, finding that RA receptor was a transcription factor of COL5A1, COL5A2 and COL3A1. The qRT-PCR results showed that the expression levels of RA receptors (RXRA, RARA and RXRG) and the predicted genes (COL5A1, COL5A2 and COL3A1) were both significantly increased during RA-induction. Also, dual-luciferase reporter assay showed that RA could affect the luciferin activities of COL5A1, COL5A2 and COL3A1. These results suggest that RA plays a crucial role in the formation of chicken spermatogonial stem cells via the transcription levels of COL5A1, COL5A2 and COL3A1 to regulate the ECM-receptor interaction signaling pathway. Additionally, knockdown of COL5A1/COL5A2/COL3A1 could effectively reduce the formation efficiency of SSCs. This indicated that the interference of RA receptor binding genes in the ECM-receptor interaction signaling pathway could decrease the efficiency of RA induced SSCs formation. Therefore, this study concludes that RA promotes formation of chicken spermatogonial stem cells by regulating the ECM-receptor interaction signaling pathway.
Asunto(s)
Células Madre Germinales Adultas/efectos de los fármacos , Células Madre Germinales Adultas/metabolismo , Receptores de Superficie Celular/metabolismo , Transducción de Señal , Espermatogonias/efectos de los fármacos , Espermatogonias/metabolismo , Tretinoina/farmacología , Animales , Diferenciación Celular , Pollos , Colágeno Tipo III/genética , Colágeno Tipo III/metabolismo , Colágeno Tipo V/genética , Colágeno Tipo V/metabolismo , Regulación de la Expresión Génica , Técnicas de Silenciamiento del Gen/métodos , MasculinoRESUMEN
LiNi0.8Co0.1Mn0.1O2 (NCM811) became a research hot point because of its low cost, environmental friendliness, and excellent electrochemical performance. However, Li+/Ni2+ intermixing is an essential factor affecting its applicability. Doping could be an important method to improve the electrochemical performance of NCM811-based cathode materials. In this work, La and Al co-doped NCM811 was prepared by a solid-state method. Results from X-ray diffraction (XRD), scanning electron microscopy (SEM), and energy-dispersive spectroscopy (EDS) and electrochemical performance were discussed in depth. These showed that when La and Al doping concentrations were 1 and 0.5%, the samples showed the best performance. The as-improved performances were mainly attributed to the reduced Li+/Ni2+ intermixing, suppressed phase transition, and decreased potential polarization and impedance.
RESUMEN
As a detection tool to identify metal or alloy, metallographic quantitative analysis has received increasing attention for its ability to evaluate quality control and reveal mechanical properties. The detection procedure is mainly operated manually to locate and characterize the constitution in metallographic images. The automatic detection is still a challenge even with the emergence of several excellent models. Benefiting from the development of deep learning, with regard to two different metallurgical structural steel image datasets, we propose two attention-aware deep neural networks, Modified Attention U-Net (MAUNet) and Self-adaptive Attention-aware Soft Anchor-Point Detector (SASAPD), to identify structures and evaluate their performance. Specifically, in the case of analyzing single-phase metallographic image, MAUNet investigates the difference between low-frequency and high-frequency and prevents duplication of low-resolution information in skip connection used in an U-Net like structure, and incorporates spatial-channel attention module with the decoder to enhance interpretability of features. In the case of analyzing multi-phase metallographic image, SASAPD explores and ranks the importance of anchor points, forming soft-weighted samples in subsequent loss design, and self-adaptively evaluates the contributions of attention-aware pyramid features to assist in detecting elements in different sizes. Extensive experiments on the above two datasets demonstrate the superiority and effectiveness of our two deep neural networks compared to state-of-the-art models on different metrics.
RESUMEN
To explore the influences of different cultivated areas on the chemical profiles of Eucommia ulmoides leaves (EUL) and rapidly authenticate its geographical origins, 187 samples from 13 provinces in China were systematically investigated using three data fusion strategies (low, mid, and high level) combined with two discrimination model algorithms (partial least squares discrimination analysis; random forest, RF). RF models constructed by high-level data fusion with different modes of different spectral data (Fourier transform near-infrared spectrum and attenuated total reflection Fourier transform mid-infrared spectrum) were most suitable for identifying EULs from different geographical origins. The accuracy rates of calibration and validation set were 92.86% and 93.44%, respectively. In addition, climate parameters were systematically investigated the cluster difference in our study. Some interesting and novel information could be found from the clustering tree diagram of hierarchical cluster analysis. The Xinjiang Autonomous Region (Region 5) located in the high latitude area was the only region in the middle temperate zone of all sample collection areas in which the samples belonged to an individual class no matter their distance in the tree diagram. The samples were from a relatively high elevation in the Shennongjia Forest District in Hubei Province (>1200 m), which is the main difference from the samples from Xiangyang City (78 m). Thus, the sample clusters from region 9 are different from the sample clusters from other regions. The results would provide a reference for further research to those samples from the special cluster.