The interpretability of deep learning models has emerged as a compelling area in artificial intelligence research.The safety criteria for medical imaging are highly stringent,and models are required for an explanation...The interpretability of deep learning models has emerged as a compelling area in artificial intelligence research.The safety criteria for medical imaging are highly stringent,and models are required for an explanation.However,existing convolutional neural network solutions for left ventricular segmentation are viewed in terms of inputs and outputs.Thus,the interpretability of CNNs has come into the spotlight.Since medical imaging data are limited,many methods to fine-tune medical imaging models that are popular in transfer models have been built using massive public Image Net datasets by the transfer learning method.Unfortunately,this generates many unreliable parameters and makes it difficult to generate plausible explanations from these models.In this study,we trained from scratch rather than relying on transfer learning,creating a novel interpretable approach for autonomously segmenting the left ventricle with a cardiac MRI.Our enhanced GPU training system implemented interpretable global average pooling for graphics using deep learning.The deep learning tasks were simplified.Simplification included data management,neural network architecture,and training.Our system monitored and analyzed the gradient changes of different layers with dynamic visualizations in real-time and selected the optimal deployment model.Our results demonstrated that the proposed method was feasible and efficient:the Dice coefficient reached 94.48%,and the accuracy reached 99.7%.It was found that no current transfer learning models could perform comparably to the ImageNet transfer learning architectures.This model is lightweight and more convenient to deploy on mobile devices than transfer learning models.展开更多
The detailed structures of the plumbing system of the early Permian Tarim flood basalt were investigated by 3-D seismic imaging.The images show that the Tarim flood basalt mainly erupted from central volcanoes distrib...The detailed structures of the plumbing system of the early Permian Tarim flood basalt were investigated by 3-D seismic imaging.The images show that the Tarim flood basalt mainly erupted from central volcanoes distributed展开更多
Objective The aim of this study was to investigate the ability of intravoxel incoherent motion(IVIM) diffusion-weighted magnetic resonance imaging(MRI) to diagnose cervical cancer and to evaluate the response of uteri...Objective The aim of this study was to investigate the ability of intravoxel incoherent motion(IVIM) diffusion-weighted magnetic resonance imaging(MRI) to diagnose cervical cancer and to evaluate the response of uterine cervical cancer to radiochemotherapy(CRT).Methods This prospective study was approved by the institutional review board, and informed consent was obtained from all patients. A total of 23 patients with primary cervical cancer who were undergoing CRT and 16 age-matched healthy subjects were prospectively recruited for IVIM(b = 0–800 s/mm2) and standard pelvic MRI. Bi-exponential analysis was performed to derive f(perfusion fraction), D*(pseudo-diffusion coefficient), and D(true molecular diffusion coefficient) in cervical cancer(n = 23) and the normal cervix(n = 16). The apparent diffusion coefficient(standard ADC) was calculated. The independent-samples t-test and paired-samples t-test were used for comparisons.Results Pre-treatment cervical cancer had the lowest standard ADC(1.15 ± 0.13 × 10-3 mm2/s) and D(0.89 ± 0.10 × 10-3 mm2/s) values, and these were significantly different from the normal cervix and posttreatment cervical cancer(P = 0.00). The f(16.67 ± 5.85%) was lowest in pre-treatment cervical cancer and was significantly different from the normal cervix and post-treatment cervical cancer(p = 0.012 and 0.00, respectively). No difference was observed in D*.Conclusion IVIM is potentially promising for differentiating between the normal cervix and cervical cancer because pre-treated cervical cancer has low perfusion and diffusion IVIM characteristics. Further, the standard ADC, D, and f of cervical cancer showed a tendency to normalize after CRT; thus, IVIM may be useful for monitoring the response to CRT in cervical cancer.展开更多
Meta-learning of dental X-rays is a machine learning technique that can be used to train models to perform new tasks quickly and with minimal input.Instead of just memorizing a task,this is accomplished through teachi...Meta-learning of dental X-rays is a machine learning technique that can be used to train models to perform new tasks quickly and with minimal input.Instead of just memorizing a task,this is accomplished through teaching a model how to learn.Algorithms for meta-learning are typically trained on a collection of training problems,each of which has a limited number of labelled instances.Multiple Xray classification tasks,including the detection of pneumonia,coronavirus disease 2019,and other disorders,have demonstrated the effectiveness of meta-learning.Meta-learning has the benefit of allowing models to be trained on dental X-ray datasets that are too few for more conventional machine learning methods.Due to the high cost and lengthy collection process associated with dental imaging datasets,this is significant for dental X-ray classification jobs.The ability to train models that are more resistant to fresh input is another benefit of meta-learning.展开更多
BACKGROUND Gastrointestinal bleeding(GIB)is a severe and potentially life-threatening condition,especially in cases of delayed treatment.Computed tomography angiography(CTA)plays a pivotal role in the early identifica...BACKGROUND Gastrointestinal bleeding(GIB)is a severe and potentially life-threatening condition,especially in cases of delayed treatment.Computed tomography angiography(CTA)plays a pivotal role in the early identification of upper and lower GIB and in the prompt treatment of the haemorrhage.AIM To determine whether a volumetric estimation of the extravasated contrast at CTA in GIB may be a predictor of subsequent positive angiographic findings.METHODS In this retrospective single-centre study,35 patients(22 men;median age 69 years;range 16-92 years)admitted to our institution for active GIB detected at CTA and further submitted to catheter angiography between January 2018 and February 2022 were enrolled.Twenty-three(65.7%)patients underwent endoscopy before CTA.Bleeding volumetry was evaluated in both arterial and venous phases via a semi-automated dedicated software.Bleeding rate was obtained from volume change between the two phases and standardised for unit time.Patients were divided into two groups,according to the angiographic signs and their concordance with CTA.RESULTS Upper bleeding accounted for 42.9%and lower GIB for 57.1%.Mean haemoglobin value at the admission was 7.7 g/dL.A concordance between positive CTA and direct angiographic bleeding signs was found in 19(54.3%)cases.Despite no significant differences in terms of bleeding volume in the arterial phase(0.55 mL vs 0.33 mL,P=0.35),a statistically significant volume increase in the venous phase was identified in the group of patients with positive angiography(2.06 mL vs 0.9 mL,P=0.02).In the latter patient group,a significant increase in bleeding rate was also detected(2.18 mL/min vs 0.19 mL/min,P=0.02).CONCLUSION In GIB of any origin,extravasated contrast volumetric analysis at CTA could be a predictor of positive angiography and may help in avoiding further unnecessary procedures.展开更多
This paper deals with thematization as a fundamental step towards the determination of photography image discourses in order to index them. It deals with questions related to photographic reality, polysemy, denotation...This paper deals with thematization as a fundamental step towards the determination of photography image discourses in order to index them. It deals with questions related to photographic reality, polysemy, denotation and connotation, whose comprehension is fundamentally important for the interpretation of photography discourses. It also approaches aspects that have influenced photography analysis towards its thematization, such as the characterization of an image bank, the functions performed by photos, as well as technical and visual quality.展开更多
Interpreting is one of the studies that is in the phase of its theoretical bottleneck. The difficulty comes from the lack of a possible holistic understanding of the phenomenon of interpreting, not to mention the mirr...Interpreting is one of the studies that is in the phase of its theoretical bottleneck. The difficulty comes from the lack of a possible holistic understanding of the phenomenon of interpreting, not to mention the mirror image of interpreting. Different linguistic studies could do nothing since nowadays their respective studies are mostly suffering their own research bottlenecks. Furthermore, it is impossible for individual linguistic studies to elaborate on the complicated interpreting practice. The only possible mirror is Modem Western Philosophy, which can present the image of interpreting. Such a mirror image is by no means possible without the full-fledged development of language study, and in particular, the uncovering of human understanding, in Modem Western Philosophy.展开更多
Monuments and historical centers, because of their particular importance, are studied in multiple ways. The study concerns different scientific disciplines and technology. Photogrammetry and remote sensing contribute ...Monuments and historical centers, because of their particular importance, are studied in multiple ways. The study concerns different scientific disciplines and technology. Photogrammetry and remote sensing contribute essentially to this study, because of the valuable qualitative and quantitative information they offer. In this paper we search through the possibilities of very high resolution satellite imagery on historical centers study, referring to Delphi historical center. The study concerns image enhancement techniques and visual interpretation of Ikonos satellite imagery. Image enhancement techniques facilitate visual interpretation, detection and recognition, of the physiognomy and spatial arrangement of Delphi historical center and offer information about physical and architectural features in the wide area of the historical center.展开更多
Aiming at the problem of anisotropy inversion of tight sands, a new method for extracting resistivity anisotropy from array laterolog and micro-resistivity scanning imaging logging is proposed, and also the consistenc...Aiming at the problem of anisotropy inversion of tight sands, a new method for extracting resistivity anisotropy from array laterolog and micro-resistivity scanning imaging logging is proposed, and also the consistency of electric and acoustic anisotropy is discussed. Array laterolog includes resistivity anisotropy information, but numerical simulation shows that drilling fluid invasion has the greatest influence on the response, followed by the relative dip angle θ and electrical anisotropy coefficient λ. A new inversion method to determine ri, Rxo, Rt and λ is developed with the given θ and initial values of invasion radius ri, flushed zone resistivity Rxo, in-situ formation resistivity Rt. Micro-resistivity image can also be used for describing the resistivity distribution information in different directions, and the electrical characteristics from micro-resistivity log in different azimuths, lateral and vertical, can be compared to extract electric anisotropy information. Directional arrangement of mineral particles in tight sands and fracture development are the intrinsic causes of anisotropy, which in turn brings about anisotropy in resistivity and acoustic velocity, so the resistivity anisotropy and acoustic velocity anisotropy are consistent in trends. Analysis of log data of several wells show that the electrical anisotropy and acoustic anisotropy extracted from array laterolog, micro-resistivity imaging and cross-dipole acoustic logs respectively are consistent in trend and magnitude, proving the inversion method is accurate and the anisotropies of different formation physical parameters caused by the intrinsic structure of tight sand reservoir are consistent. This research provides a new idea for evaluating anisotropy of tight sands.展开更多
Chinese idioms,like English proverbs,are a language form,with a certain philosophical motive and ideological basis,given meaning through people's lifelong experiences.Throughout the ages,a wide range of beliefs,cu...Chinese idioms,like English proverbs,are a language form,with a certain philosophical motive and ideological basis,given meaning through people's lifelong experiences.Throughout the ages,a wide range of beliefs,customs and natural scenery have been condensed into text forms in both Chinese idioms and English proverbs that display certain aspects of the essence of the two cultures.This paper focuses on the study of animal images in idioms,and elaborates on the different interpretations of specific images caused by the differences in history,regions,customs,and values of the two cultures.The purpose of this research is to provide a breakthrough more relevant to students who devote themselves to the study and research of Chinese and English languages and culture from the perspective of cultural contrast and analysis,so as to make a more thorough understanding of differences between Chinese and Western customs,culture and ideology.Therefore,language translations and cultural communications can be improved through a more accurate understanding of the Chinese and Western cultures.展开更多
From the perspective of new criticism, this article attempts at a profound study of Robert Frost's Desert Places so as to better understand the value, reveal the meaning and enjoy the charm of this works.
The practice has proved that it is an economic and effective method to investigate placer gold deposit by using multi-level information sources of remote sensing and multi-variate analysis methods, especially for the ...The practice has proved that it is an economic and effective method to investigate placer gold deposit by using multi-level information sources of remote sensing and multi-variate analysis methods, especially for the area with a sparse population and difficult condition like the Da Hinggan Mountains, China.The information sources used in our work includes Landsat TM, aerial infrared photography and their mosaic image maps and enlarged photos with different scales. According to statistic data, in the study area the gold-bearing rocks are mainly granite, alaskite, granodiorite and some old metamorphic rocks. On gold-bearing geological structures, the fault zones in the four directions (NE, NNE, NW and EW) are obvious, in which NNE and EW are the most key fault zones. On fluvial geomorphology the flow courses stored placer are in the tributaries of the 4th and 5th levels, especially in straight or slight curve reaches. On the basis of analysis the interpretative signs were set up, and the interpretative展开更多
Beginning with the analysis of the behavior of natural ants, this paper illuminates the principle and method that, by adopting image texture energy as pheromone and finding their way on the track of the pheromone, art...Beginning with the analysis of the behavior of natural ants, this paper illuminates the principle and method that, by adopting image texture energy as pheromone and finding their way on the track of the pheromone, artificial ants have the ability to identify and remember through similar measurement of pheromone. Based on the quantity of experiments, this paper analyzes some factors that influence the ability of artificial ants and draws some conclusions about the law of ant perception.展开更多
Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder.In this ...Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder.In this study,we develop an interpretable deep learning model capable of accurately classifying amorphous configurations and characterizing their structural properties.The results demonstrate that the multi-dimensional hybrid convolutional neural network can classify the two-dimensional(2D)liquids and amorphous solids of molecular dynamics simulation.The classification process does not make a priori assumptions on the amorphous particle environment,and the accuracy is 92.75%,which is better than other convolutional neural networks.Moreover,our model utilizes the gradient-weighted activation-like mapping method,which generates activation-like heat maps that can precisely identify important structures in the amorphous configuration maps.We obtain an order parameter from the heatmap and conduct finite scale analysis of this parameter.Our findings demonstrate that the order parameter effectively captures the amorphous phase transition process across various systems.These results hold significant scientific implications for the study of amorphous structural characteristics via deep learning.展开更多
基金The National Natural Science Foundation of China (62176048)provided funding for this research.
文摘The interpretability of deep learning models has emerged as a compelling area in artificial intelligence research.The safety criteria for medical imaging are highly stringent,and models are required for an explanation.However,existing convolutional neural network solutions for left ventricular segmentation are viewed in terms of inputs and outputs.Thus,the interpretability of CNNs has come into the spotlight.Since medical imaging data are limited,many methods to fine-tune medical imaging models that are popular in transfer models have been built using massive public Image Net datasets by the transfer learning method.Unfortunately,this generates many unreliable parameters and makes it difficult to generate plausible explanations from these models.In this study,we trained from scratch rather than relying on transfer learning,creating a novel interpretable approach for autonomously segmenting the left ventricle with a cardiac MRI.Our enhanced GPU training system implemented interpretable global average pooling for graphics using deep learning.The deep learning tasks were simplified.Simplification included data management,neural network architecture,and training.Our system monitored and analyzed the gradient changes of different layers with dynamic visualizations in real-time and selected the optimal deployment model.Our results demonstrated that the proposed method was feasible and efficient:the Dice coefficient reached 94.48%,and the accuracy reached 99.7%.It was found that no current transfer learning models could perform comparably to the ImageNet transfer learning architectures.This model is lightweight and more convenient to deploy on mobile devices than transfer learning models.
文摘The detailed structures of the plumbing system of the early Permian Tarim flood basalt were investigated by 3-D seismic imaging.The images show that the Tarim flood basalt mainly erupted from central volcanoes distributed
基金Supported by grants from the the National Natural Science Foundation of China(No.81371524,81271529)the Hubei Provincial Natural Science Foundation of China(No.2014CFB298)
文摘Objective The aim of this study was to investigate the ability of intravoxel incoherent motion(IVIM) diffusion-weighted magnetic resonance imaging(MRI) to diagnose cervical cancer and to evaluate the response of uterine cervical cancer to radiochemotherapy(CRT).Methods This prospective study was approved by the institutional review board, and informed consent was obtained from all patients. A total of 23 patients with primary cervical cancer who were undergoing CRT and 16 age-matched healthy subjects were prospectively recruited for IVIM(b = 0–800 s/mm2) and standard pelvic MRI. Bi-exponential analysis was performed to derive f(perfusion fraction), D*(pseudo-diffusion coefficient), and D(true molecular diffusion coefficient) in cervical cancer(n = 23) and the normal cervix(n = 16). The apparent diffusion coefficient(standard ADC) was calculated. The independent-samples t-test and paired-samples t-test were used for comparisons.Results Pre-treatment cervical cancer had the lowest standard ADC(1.15 ± 0.13 × 10-3 mm2/s) and D(0.89 ± 0.10 × 10-3 mm2/s) values, and these were significantly different from the normal cervix and posttreatment cervical cancer(P = 0.00). The f(16.67 ± 5.85%) was lowest in pre-treatment cervical cancer and was significantly different from the normal cervix and post-treatment cervical cancer(p = 0.012 and 0.00, respectively). No difference was observed in D*.Conclusion IVIM is potentially promising for differentiating between the normal cervix and cervical cancer because pre-treated cervical cancer has low perfusion and diffusion IVIM characteristics. Further, the standard ADC, D, and f of cervical cancer showed a tendency to normalize after CRT; thus, IVIM may be useful for monitoring the response to CRT in cervical cancer.
文摘Meta-learning of dental X-rays is a machine learning technique that can be used to train models to perform new tasks quickly and with minimal input.Instead of just memorizing a task,this is accomplished through teaching a model how to learn.Algorithms for meta-learning are typically trained on a collection of training problems,each of which has a limited number of labelled instances.Multiple Xray classification tasks,including the detection of pneumonia,coronavirus disease 2019,and other disorders,have demonstrated the effectiveness of meta-learning.Meta-learning has the benefit of allowing models to be trained on dental X-ray datasets that are too few for more conventional machine learning methods.Due to the high cost and lengthy collection process associated with dental imaging datasets,this is significant for dental X-ray classification jobs.The ability to train models that are more resistant to fresh input is another benefit of meta-learning.
文摘BACKGROUND Gastrointestinal bleeding(GIB)is a severe and potentially life-threatening condition,especially in cases of delayed treatment.Computed tomography angiography(CTA)plays a pivotal role in the early identification of upper and lower GIB and in the prompt treatment of the haemorrhage.AIM To determine whether a volumetric estimation of the extravasated contrast at CTA in GIB may be a predictor of subsequent positive angiographic findings.METHODS In this retrospective single-centre study,35 patients(22 men;median age 69 years;range 16-92 years)admitted to our institution for active GIB detected at CTA and further submitted to catheter angiography between January 2018 and February 2022 were enrolled.Twenty-three(65.7%)patients underwent endoscopy before CTA.Bleeding volumetry was evaluated in both arterial and venous phases via a semi-automated dedicated software.Bleeding rate was obtained from volume change between the two phases and standardised for unit time.Patients were divided into two groups,according to the angiographic signs and their concordance with CTA.RESULTS Upper bleeding accounted for 42.9%and lower GIB for 57.1%.Mean haemoglobin value at the admission was 7.7 g/dL.A concordance between positive CTA and direct angiographic bleeding signs was found in 19(54.3%)cases.Despite no significant differences in terms of bleeding volume in the arterial phase(0.55 mL vs 0.33 mL,P=0.35),a statistically significant volume increase in the venous phase was identified in the group of patients with positive angiography(2.06 mL vs 0.9 mL,P=0.02).In the latter patient group,a significant increase in bleeding rate was also detected(2.18 mL/min vs 0.19 mL/min,P=0.02).CONCLUSION In GIB of any origin,extravasated contrast volumetric analysis at CTA could be a predictor of positive angiography and may help in avoiding further unnecessary procedures.
文摘This paper deals with thematization as a fundamental step towards the determination of photography image discourses in order to index them. It deals with questions related to photographic reality, polysemy, denotation and connotation, whose comprehension is fundamentally important for the interpretation of photography discourses. It also approaches aspects that have influenced photography analysis towards its thematization, such as the characterization of an image bank, the functions performed by photos, as well as technical and visual quality.
文摘Interpreting is one of the studies that is in the phase of its theoretical bottleneck. The difficulty comes from the lack of a possible holistic understanding of the phenomenon of interpreting, not to mention the mirror image of interpreting. Different linguistic studies could do nothing since nowadays their respective studies are mostly suffering their own research bottlenecks. Furthermore, it is impossible for individual linguistic studies to elaborate on the complicated interpreting practice. The only possible mirror is Modem Western Philosophy, which can present the image of interpreting. Such a mirror image is by no means possible without the full-fledged development of language study, and in particular, the uncovering of human understanding, in Modem Western Philosophy.
文摘Monuments and historical centers, because of their particular importance, are studied in multiple ways. The study concerns different scientific disciplines and technology. Photogrammetry and remote sensing contribute essentially to this study, because of the valuable qualitative and quantitative information they offer. In this paper we search through the possibilities of very high resolution satellite imagery on historical centers study, referring to Delphi historical center. The study concerns image enhancement techniques and visual interpretation of Ikonos satellite imagery. Image enhancement techniques facilitate visual interpretation, detection and recognition, of the physiognomy and spatial arrangement of Delphi historical center and offer information about physical and architectural features in the wide area of the historical center.
基金Supported by the Scientific Research and Technological Development Project of CNPC(2019A-3608)
文摘Aiming at the problem of anisotropy inversion of tight sands, a new method for extracting resistivity anisotropy from array laterolog and micro-resistivity scanning imaging logging is proposed, and also the consistency of electric and acoustic anisotropy is discussed. Array laterolog includes resistivity anisotropy information, but numerical simulation shows that drilling fluid invasion has the greatest influence on the response, followed by the relative dip angle θ and electrical anisotropy coefficient λ. A new inversion method to determine ri, Rxo, Rt and λ is developed with the given θ and initial values of invasion radius ri, flushed zone resistivity Rxo, in-situ formation resistivity Rt. Micro-resistivity image can also be used for describing the resistivity distribution information in different directions, and the electrical characteristics from micro-resistivity log in different azimuths, lateral and vertical, can be compared to extract electric anisotropy information. Directional arrangement of mineral particles in tight sands and fracture development are the intrinsic causes of anisotropy, which in turn brings about anisotropy in resistivity and acoustic velocity, so the resistivity anisotropy and acoustic velocity anisotropy are consistent in trends. Analysis of log data of several wells show that the electrical anisotropy and acoustic anisotropy extracted from array laterolog, micro-resistivity imaging and cross-dipole acoustic logs respectively are consistent in trend and magnitude, proving the inversion method is accurate and the anisotropies of different formation physical parameters caused by the intrinsic structure of tight sand reservoir are consistent. This research provides a new idea for evaluating anisotropy of tight sands.
文摘Chinese idioms,like English proverbs,are a language form,with a certain philosophical motive and ideological basis,given meaning through people's lifelong experiences.Throughout the ages,a wide range of beliefs,customs and natural scenery have been condensed into text forms in both Chinese idioms and English proverbs that display certain aspects of the essence of the two cultures.This paper focuses on the study of animal images in idioms,and elaborates on the different interpretations of specific images caused by the differences in history,regions,customs,and values of the two cultures.The purpose of this research is to provide a breakthrough more relevant to students who devote themselves to the study and research of Chinese and English languages and culture from the perspective of cultural contrast and analysis,so as to make a more thorough understanding of differences between Chinese and Western customs,culture and ideology.Therefore,language translations and cultural communications can be improved through a more accurate understanding of the Chinese and Western cultures.
文摘From the perspective of new criticism, this article attempts at a profound study of Robert Frost's Desert Places so as to better understand the value, reveal the meaning and enjoy the charm of this works.
文摘The practice has proved that it is an economic and effective method to investigate placer gold deposit by using multi-level information sources of remote sensing and multi-variate analysis methods, especially for the area with a sparse population and difficult condition like the Da Hinggan Mountains, China.The information sources used in our work includes Landsat TM, aerial infrared photography and their mosaic image maps and enlarged photos with different scales. According to statistic data, in the study area the gold-bearing rocks are mainly granite, alaskite, granodiorite and some old metamorphic rocks. On gold-bearing geological structures, the fault zones in the four directions (NE, NNE, NW and EW) are obvious, in which NNE and EW are the most key fault zones. On fluvial geomorphology the flow courses stored placer are in the tributaries of the 4th and 5th levels, especially in straight or slight curve reaches. On the basis of analysis the interpretative signs were set up, and the interpretative
基金Founded by the National Science Foundation of China (No.42071094) .
文摘Beginning with the analysis of the behavior of natural ants, this paper illuminates the principle and method that, by adopting image texture energy as pheromone and finding their way on the track of the pheromone, artificial ants have the ability to identify and remember through similar measurement of pheromone. Based on the quantity of experiments, this paper analyzes some factors that influence the ability of artificial ants and draws some conclusions about the law of ant perception.
基金National Natural Science Foundation of China(Grant No.11702289)the Key Core Technology and Generic Technology Research and Development Project of Shanxi Province,China(Grant No.2020XXX013)the National Key Research and Development Project of China。
文摘Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder.In this study,we develop an interpretable deep learning model capable of accurately classifying amorphous configurations and characterizing their structural properties.The results demonstrate that the multi-dimensional hybrid convolutional neural network can classify the two-dimensional(2D)liquids and amorphous solids of molecular dynamics simulation.The classification process does not make a priori assumptions on the amorphous particle environment,and the accuracy is 92.75%,which is better than other convolutional neural networks.Moreover,our model utilizes the gradient-weighted activation-like mapping method,which generates activation-like heat maps that can precisely identify important structures in the amorphous configuration maps.We obtain an order parameter from the heatmap and conduct finite scale analysis of this parameter.Our findings demonstrate that the order parameter effectively captures the amorphous phase transition process across various systems.These results hold significant scientific implications for the study of amorphous structural characteristics via deep learning.
