Visualization for Explanation of Deep Learning-Based Defect Detection Model Using Class Activation Map

Recently,convolutional neural network(CNN)-based visual inspec-tion has been developed to detect defects on building surfaces automatically.The CNN model demonstrates remarkable accuracy in image data analysis;however,the predicted results have uncertainty in providing accurate informa-tion to users because of the“black box”problem in the deep learning model.Therefore,this study proposes a visual explanation method to overcome the uncertainty limitation of CNN-based defect identification.The visual repre-sentative gradient-weights class activation mapping(Grad-CAM)method is adopted to provide visually explainable information.A visualizing evaluation index is proposed to quantitatively analyze visual representations;this index reflects a rough estimate of the concordance rate between the visualized heat map and intended defects.In addition,an ablation study,adopting three-branch combinations with the VGG16,is implemented to identify perfor-mance variations by visualizing predicted results.Experiments reveal that the proposed model,combined with hybrid pooling,batch normalization,and multi-attention modules,achieves the best performance with an accuracy of 97.77%,corresponding to an improvement of 2.49%compared with the baseline model.Consequently,this study demonstrates that reliable results from an automatic defect classification model can be provided to an inspector through the visual representation of the predicted results using CNN models.

Gas liquid cylindrical cyclone flow regime identification using machine learning combined with experimental mechanism explanation

The flow regimes of GLCC with horizon inlet and a vertical pipe are investigated in experiments,and the velocities and pressure drops data labeled by the corresponding flow regimes are collected.Combined with the flow regimes data of other GLCC positions from other literatures in existence,the gas and liquid superficial velocities and pressure drops are used as the input of the machine learning algorithms respectively which are applied to identify the flow regimes.The choosing of input data types takes the availability of data for practical industry fields into consideration,and the twelve machine learning algorithms are chosen from the classical and popular algorithms in the area of classification,including the typical ensemble models,SVM,KNN,Bayesian Model and MLP.The results of flow regimes identification show that gas and liquid superficial velocities are the ideal type of input data for the flow regimes identification by machine learning.Most of the ensemble models can identify the flow regimes of GLCC by gas and liquid velocities with the accuracy of 0.99 and more.For the pressure drops as the input of each algorithm,it is not the suitable as gas and liquid velocities,and only XGBoost and Bagging Tree can identify the GLCC flow regimes accurately.The success and confusion of each algorithm are analyzed and explained based on the experimental phenomena of flow regimes evolution processes,the flow regimes map,and the principles of algorithms.The applicability and feasibility of each algorithm according to different types of data for GLCC flow regimes identification are proposed.

On fine-grained visual explanation in convolutional neural networks

Existing explanation methods for Convolutional Neural Networks(CNNs)lack the pixel-level visualization explanations to generate the reliable fine-grained decision features.Since there are inconsistencies between the explanation and the actual behavior of the model to be interpreted,we propose a Fine-Grained Visual Explanation for CNN,namely F-GVE,which produces a fine-grained explanation with higher consistency to the decision of the original model.The exact backward class-specific gradients with respect to the input image is obtained to highlight the object-related pixels the model used to make prediction.In addition,for better visualization and less noise,F-GVE selects an appropriate threshold to filter the gradient during the calculation and the explanation map is obtained by element-wise multiplying the gradient and the input image to show fine-grained classification decision features.Experimental results demonstrate that F-GVE has good visual performances and highlights the importance of fine-grained decision features.Moreover,the faithfulness of the explanation in this paper is high and it is effective and practical on troubleshooting and debugging detection.

专家系统工具C-ADVISOR的解释系统EXPLANATION

为了提高C—ADVISOR的实用性,我们设计了一个工具解释系统EXPLANATION。该解释系统采用预制文本法(prepared text)和执行追踪法(execution traces)对用户的提问给予解释,具有动态解释和静态解释两种功能。系统给出一种用户模型,能够针对用户的知识水平作出相应的解释。系统采用紧缩存贮技术,节省了大量的磁盘存贮空间。系统还设置一个数据库生成程序,为解释系统的建立、修改和维护提供方便的条件。

Investigation of feature contribution to shield tunneling-induced settlement using Shapley additive explanations method

Accurate prediction of shield tunneling-induced settlement is a complex problem that requires consideration of many influential parameters.Recent studies reveal that machine learning(ML)algorithms can predict the settlement caused by tunneling.However,well-performing ML models are usually less interpretable.Irrelevant input features decrease the performance and interpretability of an ML model.Nonetheless,feature selection,a critical step in the ML pipeline,is usually ignored in most studies that focused on predicting tunneling-induced settlement.This study applies four techniques,i.e.Pearson correlation method,sequential forward selection(SFS),sequential backward selection(SBS)and Boruta algorithm,to investigate the effect of feature selection on the model’s performance when predicting the tunneling-induced maximum surface settlement(S_(max)).The data set used in this study was compiled from two metro tunnel projects excavated in Hangzhou,China using earth pressure balance(EPB)shields and consists of 14 input features and a single output(i.e.S_(max)).The ML model that is trained on features selected from the Boruta algorithm demonstrates the best performance in both the training and testing phases.The relevant features chosen from the Boruta algorithm further indicate that tunneling-induced settlement is affected by parameters related to tunnel geometry,geological conditions and shield operation.The recently proposed Shapley additive explanations(SHAP)method explores how the input features contribute to the output of a complex ML model.It is observed that the larger settlements are induced during shield tunneling in silty clay.Moreover,the SHAP analysis reveals that the low magnitudes of face pressure at the top of the shield increase the model’s output。

海登·怀特在《元史学》中混用interpretation与explanation的动机探究

有证据表明,海登?怀特的《元史学》导论(1973年11月)极有可能是根据《历史中的解释》一文(1973年初)修订而成。在《历史中的解释》中,怀特比较严格地区分了interpretation与explanation这两个概念,但在《元史学》导论中,他却混用二者,并在叙事层面偏好使用explanation。通过这一有意为之的举动,怀特消解了德国历史诠释学中interpretation与explanation的对立,拉低了interpretation的理论高度,清除了interpretation隐含的规训意味,同时赋予了explanation某种整体性、非认知性,从而超越了德国历史诠释学和战后英美学界的历史解释问题争论。

Theoretical Explanation and Improvement to the Flare Model of Lithography Based on the Kirk Test

The Kirk test has good precision for measuring stray light in optical lithography and is the usual method of measuring stray light.However,Kirk did not provide a theoretical explanation to his simulation model.We attempt to give Kirk's model a kind of theoretical explanation and a little improvement based on the model of point spread function of scattering and the theory of statistical optics.It is indicated by simulation that the improved model fits Kirk's measurement data better.

Alternative kind of hydrogen atoms as a possible explanation for the latest puzzling observation of the 21 cm radio line from the early Universe

There is a puzzling astrophysical result concerning the latest observation of the absorption profile of the redshifted radio line 21 cm from the early Universe(as described in Bowman et al.). The amplitude of the profile was more than a factor of two greater than the largest predictions. This could mean that the primordial hydrogen gas was much cooler than expected. Some explanations in the literature suggested a possible cooling of baryons either by unspecified dark matter particles or by some exotic dark matter particles with a charge a million times smaller than the electron charge. Other explanations required an additional radio background. In the present paper, we entertain a possible different explanation for the above puzzling observational result: the explanation is based on the alternative kind of hydrogen atoms(AKHA),whose existence was previously demonstrated theoretically, as well as by the analysis of atomic experiments. Namely, the AKHA are expected to decouple from the cosmic microwave background(CMB) much earlier(in the course of the Universe expansion) than usual hydrogen atoms, so that the AKHA temperature is significantly lower than that of usual hydrogen atoms. This seems to lower the excitation(spin) temperature of the hyperfine doublet(responsible for the 21 cm line) sufficiently enough for explaining the above puzzling observational result. This possible explanation appears to be more specific and natural than the previous possible explanations. Further observational studies of the redshifted 21 cm radio line from the early Universe could help to verify which explanation is the most relevant.

Regulation on the Protection of Fossils Issued by the State Council of the People's Republic of China:Scientists' Researches Protected Legally—Further Explanation

With large-scale engineering projects being carried out in China, a large number of fossil localities have been discovered and excavated by responsible agencies, but still some important fossils of great value have been removed and smuggled into foreign countries. In the last three years, more than 1345 fossil specimens have been intercepted by Customs in Shenzhen, Shanghai, Tianjin, Beijing and elsewhere, and more than 5000 fossils, most of which are listed as key fossils,

Analysis on Explanation Effect of the European Numerical Prediction on Temperature

[Objective] The research aimed to analyze explanation effect of the European numerical prediction on temperature. [Method] Based on CMSVM regression method, by using 850 hPa grid point data of the European numerical prediction from 2003 to 2009 and actual data of the maximum and minimum temperatures at 8 automatic stations in Qingyang City, prediction model of the temperature was established, and running effect of the business from 2008 to 2010 was tested and evaluated. [Result] The method had very good guidance role in real-time business running of the temperature prediction. Test and evaluation found that as forecast time prolonged, prediction accuracies of the maximum and minimum temperatures declined. When temperature anomaly was higher (actual temperature was higher than historical mean), prediction accuracy increased. Influence of the European numerical prediction was bigger. [Conclusion] Compared with other methods, operation of the prediction method was convenient, modeling was automatic, running time was short, system was stable, and prediction accuracy was high. It was suitable for implementing of the explanation work for numerical prediction product at meteorological station.

A theoretical explanation of the decays of Majorana oscillations

Majorana zero modes in the hybrid semiconductor-superconductornanowire is one of the promising candidates for topologicalquantum computing. Recently, in nanowires with a superconductingisland, the signature of Majorana zero modescan be revealed as a subgap state whose energy oscillatesaround zero in magnetic field. This oscillation was interpretedas overlapping Majoranas. However, the oscillation amplitudeeither dies away after an overshoot or decays, sharply oppositeto the theoretically predicted enhanced oscillations for Majoranabound states, as the magnetic field increases. Several theoreticalstudies have tried to address this discrepancy, but arepartially successful. This discrepancy has raised the concernson the conclusive identification of Majorana bound states, andhas even endangered the scheme of Majorana qubits basedon the nanowires.

THEORETICAL EXPLANATION OF SPIN REORIENTATION IN DyTiFe_(11) COMPOUND

DyTiFe_(11) compound is a ferromagnetic substance.It has tetragonal body-centered ThMn_(12)-type crystallographic structure.At room temperature,the easy magnetization direction is the c-axis.A spin reorientation begins to appear at about 175K.The contribution of Fe sublattice to magnetocrystalline anisotropy was determined by experiments and that of Dy sublattice was obtained by using single ion model calculation.Results show that the spin reorientation arises from the competition of anisotropy between Fe and Dy sublattices.

Henderson Island’s Plastic: An Explanation

Plastic has been accumulating on the beaches of Henderson Island in vast quantities according to recent news reports. It is proposed that the plastic is brought to the island by a very broad permanent surface current flowing southeastward past the island. Other characteristics of the flow are that its temperature is relatively high, its depth is shallow (about 100 m), its speed is sluggish (10 - 20 cm/sec), and by broad is meant more than 5000 km along 28 S. Henderson Island is located at the east/west midpoint of this wide warm current (130 W). By knowing more definitely where the plastic is coming from, than the vague suggestions provided by the news sources, it may be possible in the future to slow down or stop the piling up of trash on what were pristine beaches of this World Heritage Site.

An Explanation for the Undetection of Radio Pulsar in Supernova 1987A

Supernova 1987 A is a core collapse supernova in the Large Magellanic Cloud, inside which the product is most likely a neutron star. Despite the most sensitive available detection instruments from radio to γ-ray wavebands being exploited in the pass thirty years, there have not yet been any pulse signals detected. By considering the density of the medium plasma in the remnant of 1987 A, we find that the plasma cut-off frequency is approximately7 GHz, a value higher than the conventional observational waveband of radio pulsars. As derived, with the expansion of the supernova remnant, the radio signal will be detected in 2073 A.D. at 3 GHz.

EXPLANATION OF MIDDLE-PAGE

This picture of "International Space Year" provided by Geo-Space Co. of Austria, which is obtained by geometric accurate correction and digital mounting on NOAA/AVHHR meteorological satellite image, displays the vegetable rates of Europen Continent in detail. The snow-cover of the Alps, Scandinavia and Iceland and dry and hot features of Mediterranean area are very clear. The vegetable rates give multi-temporal information for the study of environmental and seasonal change, biomass estimation and crop production, which has been the

Explanation vs Performance in Data Mining: A Case Study with Predicting Runaway Projects

Often, the explanatory power of a learned model must be traded off against model performance. In the case of predict-ing runaway software projects, we show that the twin goals of high performance and good explanatory power are achievable after applying a variety of data mining techniques (discrimination, feature subset selection, rule covering algorithms). This result is a new high water mark in predicting runaway projects. Measured in terms of precision, this new model is as good as can be expected for our data. Other methods might out-perform our result (e.g. by generating a smaller, more explainable model) but no other method could out-perform the precision of our learned model.

EXPLANATION OF BACK-PAGE

Based on Landsat image received by the Remote Sensing Ground Station of Chinese Academy of Sciences on Feb. 23, 1988, the image is obtained by geometric accurate-correction and flase-colour composition. The ancient way of fresh-water stream, bridge, block, railway, highway can be easily distinguished.

CAS 132-2006 Standard for Flat Panel Display TV Set Installation Service Formulating and Enforcing Main Point Explanation and Analysis

CAS 132-2006 formulation background and work courses Flat panel display technology is maturing alongside the development of science and technology. The prices of flat panel display TV sets are becoming lower and lower. Flat panel display TV sets have begun entering the mainstream and are now the first choice of some consumers when making new TV purchases.Display TV Set Installation Service Point Explanation and AnalysisRepair, Retreat and Change Duty Stipulations. Peculiar use conditions, usage environments, and the practical circumstances of user consumption service requirements, etc., were also taken into account on the basis of extensive investigation, testing, and verification.……