Transfer Learning With CNN for Classification of Weld Defect

Data Analytic interpretation of plant species is not widely explored. The image variety of plant species can be investigated with the help of machine learning and deep learning techniques. Machine learning and deep learning make the plant species investigation automated and almost accurate. In this paper, we have proposed a novel native and invasive plant species classification technique. We use the pre-trained deep convolution neural networks (transfer learning) to extract features from plant species and classify them into Native and Invasive. Here, state-of-the-art deep convolution neural network models like InceptionV3, MobileNetV2, ResNetV2, VGG16, and Xception have been explored to extract the feature vectors from images. Furthermore, the features of the images have been used for image classification using the Random Forest algorithm. The performance of the proposed system is verified on the proposed dataset (Native-Invasive) with aerial view images of native and invasive species. Results were obtained using cross-validation techniques. Our experiments demonstrate the potential of our proposed method for achieving excellent performance with a 95% accuracy using Xception with data augmentation, hyper-parameter optimization, and the Random Forest classification technique. The aerial images were captured and labeled by us which is very novel to our experimental setup. After using deep learning models we got promising results. The dataset used to train the plant species classifier will be made available on request.KeywordsFeature extractionTransfer learningRandom ForestDeep learningImage classificationNative plant speciesInvasive plant species

Summary A meticulous interpretation of steady-state or unsteady-state relative permeability (Kr) experimental data is required to determine a complete set of Kr curves. In this work, different machine learning (ML) models were developed to assist in a faster estimation of these curves from steady-state drainage coreflooding experimental runs. These ML algorithms include gradient boosting (GB), random forest (RF), extreme gradient boosting (XGB), and deep neural network (DNN) with a main focus on and comparison of the two latter algorithms (XGB and DNN). Based on existing mathematical models, a leading-edge framework was developed where a large database of Kr and capillary pressure (Pc) curves were generated. This database was used to perform thousands of coreflood simulation runs representing oil-water drainage steady-state experiments. The results obtained from these simulation runs, mainly pressure drop along with other conventional core analysis data, were used to estimate analytical Kr curves based on Darcy’s law. These analytically estimated Kr curves along with the previously generated Pc curves were fed as features into the ML model. The entire data set was split into 80% for training and 20% for testing. The k-fold cross-validation technique was applied to increase the model’s accuracy by splitting 80% of the training data into 10 folds. In this manner, for each of the 10 experiments, nine folds were used for training and the remaining fold was used for model validation. Once the model was trained and validated, it was subjected to blind testing on the remaining 20% of the data set. The ML model learns to capture fluid flow behavior inside the core from the training data set. In terms of applicability of these ML models, two sets of experimental data were needed as input; the first was the analytically estimated Kr curves from the steady-state drainage coreflooding experiments, while the other was the Pc curves estimated from centrifuge or mercury injection capillary pressure (MICP) measurements. The trained/tested model was then able to estimate Kr curves based on the experimental results fed as input. Furthermore, to test the performance of the ML model when only one set of experimental data is available to an end user, a recurrent neural network (RNN) algorithm was trained/tested to predict Kr curves in the absence of Pc curves as an input. The performance of the three developed models (XGB, DNN, and RNN) was assessed using the values of the coefficient of determination (R2) along with the loss calculated during training/validation of the model. The respective crossplots along with comparisons of ground truth vs. artificial intelligence (AI)-predicted curves indicated that the model is capable of making accurate predictions with an error percentage between 0.2% and 0.6% on history-matching experimental data for all three tested ML techniques. This implies that the AI-based model exhibits better efficiency and reliability in determining Kr curves when compared to conventional methods. The developed ML models by no means replace the need to conduct drainage coreflooding or centrifuge experiments but act as an alternative to existing commercial platforms that are used to interpret experimental data to predict Kr curves. The two main advantages of the developed ML models are their capability of predicting Kr curves within a matter of a few minutes as well as with limited intervention from the end user. The results also include a comparison between classical ML approaches, shallow neural networks, and DNNs in terms of accuracy in predicting the final Kr curves. The research presented here is an extension of the state-of-the-art framework proposed by Mathew et al. (2021). However, the two main aspects of the current study are the application of deep learning for the prediction of Kr curves and the application of feature engineering. The latter not only reduces the training/testing time for the ML models but also enables the end user to obtain the final predictions with the least set of experimental data. The various models discussed in this research work currently focus on the prediction of Kr curves for drainage steady-state experiments; however, the work can be extended to capture the imbibition cycle as well.

Digital forgery is increasing day by day due to the increase in the usage of digital media files and the ease of editing media files. But digital media editing is a serious issue, as digital media are the primary source of evidence for criminal cases in the courtrooms. Consequently, there is an urge to detect and localize forgeries in digital media to aid digital forensics. The main aim of the proposed work is to detect forgeries in digital images. Image splicing forgery involves copying an image region from one image and pasting it into another image. A deep learning-based technique is proposed to detect image splicing forgery. A pre-trained deep Convolution Neural Network is transferred for the proposed application. The network is trained using spliced and original images to adapt it to the image splicing detection problem. The layers in the network are modified and fine-tuned to make it perform well for the new unseen dataset. This re-designed convolution network discriminates spliced and original images accurately. Also, the proposed work locates the spliced image regions if both the source and spliced images are provided as input image pairs. The proposed work is tested on CASIA and Columbia splicing image datasets and achieved good results.

Highlights Estimating the moisture content (MC) of cereal grain and oilseed in real time is a challenge. We explored the accuracy of MC estimation with machine learning models when cereal grain and oilseed were dried with a microwave device. The frequency, attenuation, phase shift, and dielectric properties were utilized for model training. Out of five machine learning models, a custom DNN model performed the best, and the RF model was the next best-performing model. ABSTRACT. Estimating moisture content (MC) in cereal grains and oilseeds plays a major role in their quality control, ensuring safe storage and determining their market value. Conventional MC measurement methods, such as oven drying, are time-consuming and not suitable for continuous monitoring. This study focuses on evaluating various machine learning algorithms to estimate MC in cereal grains and oilseeds by analyzing their dielectric properties (dielectric constant, e', and dielectric loss factor, e"). Machine learning (ML) models, including a modified Deep Neural Network (DNN), Autoencoder, LeNet5, Support Vector Machine (SVM), and Random Forest (RF), were evaluated for MC determination. The dataset consisted of multiple cereal grain and oilseed types of oats, soybeans, sorghum, corn, barley, and wheat (hereafter referred to as “grains” for simplicity). The models were trained on three test sets: (I) only dielectric properties from all grains, (II) all measured physical properties from all grains, and (III) all measured physical properties from wheat. Results showed that the DNN outperformed other models on test sets I and III with an R 2 value of 0.9084 and 0.9995, respectively, whereas RF showed the best performance in test set II with an R 2 value of 0.9784. These results highlight the effectiveness of ML algorithms in estimating the MC level of grains accurately, providing an alternative to traditional methods and enabling real-time and non-destructive monitoring in agricultural processes. Keywords: Cereal grains, Dielectric properties, Machine learning, Microwave sensing, Moisture content, Oilseeds.

Deep learning models are highly parameterized, resulting in difficulty in inference and transfer learning for image recognition tasks. In this work, we propose a layered pathway evolution method to compress a deep convolutional neural network (DCNN) for classification of masses in digital breast tomosynthesis (DBT). The objective is to prune the number of tunable parameters while preserving the classification accuracy. In the first stage transfer learning, 19 632 augmented regions-of-interest (ROIs) from 2454 mass lesions on mammograms were used to train a pre-trained DCNN on ImageNet. In the second stage transfer learning, the DCNN was used as a feature extractor followed by feature selection and random forest classification. The pathway evolution was performed using genetic algorithm in an iterative approach with tournament selection driven by count-preserving crossover and mutation. The second stage was trained with 9120 DBT ROIs from 228 mass lesions using leave-one-case-out cross-validation. The DCNN was reduced by 87% in the number of neurons, 34% in the number of parameters, and 95% in the number of multiply-and-add operations required in the convolutional layers. The test AUC on 89 mass lesions from 94 independent DBT cases before and after pruning were 0.88 and 0.90, respectively, and the difference was not statistically significant (p > 0.05). The proposed DCNN compression approach can reduce the number of required operations by 95% while maintaining the classification performance. The approach can be extended to other deep neural networks and imaging tasks where transfer learning is appropriate.

Optimisation and interpretation of machine and deep learning models for improved water quality management in Lake Loktak

In recent years, a significant amount of research has focused on analyzing the effectiveness of machine learning (ML) models for malware detection. These approaches have ranged from methods such as decision trees and clustering to more complex approaches like support vector machine (SVM) and deep neural networks. In particular, neural networks have proven to be very effective in detecting complex and advanced malware. This, however, comes with a caveat. Neural networks are notoriously complex. Therefore, the decisions that they make are often just accepted without questioning why the model made that specific decision. The black box characteristic of neural networks has challenged researchers to explore methods to explain black-box models such as SVM and neural networks and their decision-making process. Transparency and explainability give the experts and malware analysts assurance and trustworthiness about the ML models’ decisions. In addition, it helps in generating comprehensive reports that can be used to enhance cyber threat intelligence sharing. As such, this much needed analysis drives our work in this paper to explore the explainability and interpretability of ML models in the field of online malware detection. In this paper, we used the Shapley Additive exPlanations (SHAP) explainability technique to achieve efficient performance in interpreting the outcome of different ML models such as SVM Linear, SVM-RBF (Radial Basis Function), Random Forest (RF), Feed-Forward Neural Net (FFNN), and Convolutional Neural Network (CNN) models trained on an online malware dataset. To explain the output of these models, explainability techniques such as KernalSHAP, TreeSHAP, and DeepSHAP are applied to the obtained results.

In the human brain, the incoming light to the retina is transformed into meaningful representations that allow us to interact with the world. In a similar vein, the RGB pixel values are transformed by a deep neural network (DNN) into meaningful representations relevant to solving a computer vision task it was trained for. Therefore, in my research, I aim to reveal insights into the visual representations in the human visual cortex and DNNs solving vision tasks. In the previous decade, DNNs have emerged as the state-of-the-art models for predicting neural responses in the human and monkey visual cortex. Research has shown that training on a task related to a brain region’s function leads to better predictivity than a randomly initialized network. Based on this observation, we proposed that we can use DNNs trained on different computer vision tasks to identify functional mapping of the human visual cortex. To validate our proposed idea, we first investigate a brain region occipital place area (OPA) using DNNs trained on scene parsing task and scene classification task. From the previous investigations about OPA’s functions, we knew that it encodes navigational affordances that require spatial information about the scene. Therefore, we hypothesized that OPA’s representation should be closer to a scene parsing model than a scene classification model as the scene parsing task explicitly requires spatial information about the scene. Our results showed that scene parsing models had representation closer to OPA than scene classification models thus validating our approach. We then selected multiple DNNs performing a wide range of computer vision tasks ranging from low-level tasks such as edge detection, 3D tasks such as surface normals, and semantic tasks such as semantic segmentation. We compared the representations of these DNNs with all the regions in the visual cortex, thus revealing the functional representations of different regions of the visual cortex. Our results highly converged with previous investigations of these brain regions validating the feasibility of the proposed approach in finding functional representations of the human brain. Our results also provided new insights into underinvestigated brain regions that can serve as starting hypotheses and promote further investigation into those brain regions. We applied the same approach to find representational insights about the DNNs. A DNN usually consists of multiple layers with each layer performing a computation leading to the final layer that performs prediction for a given task. Training on different tasks could lead to very different representations. Therefore, we first investigate at which stage does the representation in DNNs trained on different tasks starts to differ. We further investigate if the DNNs trained on similar tasks lead to similar representations and on dissimilar tasks lead to more dissimilar representations. We selected the same set of DNNs used in the previous work that were trained on the Taskonomy dataset on a diverse range of 2D, 3D and semantic tasks. Then, given a DNN trained on a particular task, we compared the representation of multiple layers to corresponding layers in other DNNs. From this analysis, we aimed to reveal where in the network architecture task-specific representation is prominent. We found that task specificity increases as we go deeper into the DNN architecture and similar tasks start to cluster in groups. We found that the grouping we found using representational similarity was highly correlated with grouping based on transfer learning thus creating an interesting application of the approach to model selection in transfer learning. During previous works, several new measures were introduced to compare DNN representations. So, we identified the commonalities in different measures and unified different measures into a single framework referred to as duality diagram similarity. This work opens up new possibilities for similarity measures to understand DNN representations. While demonstrating a much higher correlation with transfer learning than previous state-of-the-art measures we extend it to understanding layer-wise representations of models trained on the Imagenet and Places dataset using different tasks and demonstrate its applicability to layer selection for transfer learning. In all the previous works, we used the task-specific DNN representations to understand the representations in the human visual cortex and other DNNs. We were able to interpret our findings in terms of computer vision tasks such as edge detection, semantic segmentation, depth estimation, etc. however we were not able to map the representations to human interpretable concepts. Therefore in our most recent work, we developed a new method that associates individual artificial neurons with human interpretable concepts. Overall, the works in this thesis revealed new insights into the representation of the visual cortex and DNNs...

The evolution and improvements of deep learning are being used to tackle any research obstacles that could be converted into classification problems in all spheres of life. Each Deep convolutional neural network (DCNN) design's output is determined by the depth and value of the hyperparameters, which explains why so many of them have been proposed. These DCNN architectures must be created entirely from scratch, and they can only be used for the applications for which they were intended. Transfer learning may be used to modify these pre-trained networks so they are more appropriate for particular purposes. This article aims to evaluate the empirical performance of the applicability of pre-trained DCNN models to identify human face presentation threats (FPAD). Human FPAD is one of the most significant and crucial areas of research right now because of the introduction of ambient computing, which necessitates contact-free identification of persons with the help of their biometric traits. Six pre-trained DCNN models are taken into account for an experimental evaluation in human FPAD alias VGG19, VGG16, DensNet121, MobileNet, Xception, and InceptionV3. The investigation makes use of the NUAA and Replay-Attack benchmark FPAD datasets. Thepade's sorted block truncation coding (SBTC) 10-ary features are merged with deep learning features produced from the finest performing finetuned DCNNs to enhance the FPAD capabilities of analyzed machine learning (ML) classifiers. The integration of features of Thepade's SBTC 10-ary and DCNN has considerably increased the FPAD accuracy of ML classifiers with slightly more computations of feature extraction.

Welding defects detection and classification is very important to guarantee the welding quality. Over the last 30 years, there has been a large amount of research attempting to develop an automatic (or semiautomatic) system for the detection and classification of weld defects in continuous welds using radiography. In this paper, we describe an automatic system for classification of welding defects from radiographic images and compare with KNN and SVM classifiers. We classify and recognize the linear defects such as lack of penetrations, incomplete fusion and external undercut. Experimental results have shown the classification method is useful for the lengthy defects and obtained through our method is better than the two classifiers methods.

Automated detection of leukemia by pretrained deep neural networks and transfer learning: A comparison

Purpose: The aim of the article is to develop an algorithm for classifying cracks in the analyzed images using modern methods of deep machine learning and transfer learning based on pretrained convolutional neural network - Inception-ResNet-v2. Design/Methodology/Approach: Transfer learning based on the pretrained convolutional neural network was used to categorize the images. The fully conected layer of the Inception-ResNet-v2 network has been modified. The last layer was trained using a two-class (binary) linear SVM (Support Vector Machine). In total, 20,000 training cases (images) were used to train the fully connected layer within transfer learning process. The research analyzed the possibility of using the deep neural networks for quick and fully automatic identification of cracks / defects on the surface of analyzed parts. Findings: The results indicate that pretrained convolutional neural network using SVM to train a fully connected layer is a very effective solution for visual crack / fault detection. In the analyzed model, a positive classification was obtained at the level of 99.89%. Practical Implications: The model presented in the article can be used in quality control carried out by process monitoring. An effective model for identifying defective parts can be used in both logistics and production processes. Originality/Value: A novelty is the use of a freely available, deep neural network trained to classify 1000 categories of various images for binary categorization of faults (cracks). The algorithm was adjusted by replacing the primary, 1000-output fully connected layer in the Inception-ResNet-v2 network with a binary layer (2 categories). The fully connected layer has been trained using the classification version of the popular SVM learner, but thanks to the combination of this layer with the sophisticated fearure extraction ability of the pre-trained Inception-ResNet-v2 deep network, the resulting predictive model enables the classification of defects with a very high level of accuracy.

Feature fusion methods have been demonstrated to be effective for many computer vision based applications. These methods generally use multiple hand-crafted features. However, in recent days, features extracted through transfer leaning procedures have been proved to be robust than the hand-crafted features in myriad applications, such as object classification and recognition. The transfer learning is a highly appreciated strategy in deep convolutional neural networks (DCNNs) due to its multifaceted benefits. It heartens us to explore the effect of fusing multiple transfer learning features of different DCNN architectures. Thus, in this work, we extract features of image statistics by exploiting three different pre-trained DCNNs through transfer learning. Then, we transform the features into a generalized subspace through a recently introduced Autoencoder network and fuse them to form intra-class invariant feature vector that is used to train a multi-class Support Vector Machine (SVM). The experimental results on various datasets, including object and action image statistics show that the fusion of multiple transfer learning features improves classification accuracy as compared to fusion of multiple hand-crafted features and usage of single component transfer learning features.

This study aims to develop and validate a machine learning-based mortality risk prediction model for V-A ECMO patients to improve the precision of clinical decision-making. This multicenter retrospective cohort study included 280 patients receiving V-A ECMO from the Second Affiliated Hospital of Guangxi Medical University, Yulin First People’s Hospital, and the MIMIC-IV database. The data from the Second Affiliated Hospital of Guangxi Medical University and the MIMIC-IV database were merged and randomly divided in a 7:3 ratio into a training set and an internal validation set, respectively. The dataset from Yulin First People’s Hospital was reserved as an external validation cohort. The primary study outcome was defined as in-hospital mortality.Feature selection was conducted using Lasso regression, followed by the development of six machine learning models: Logistic Regression, Random Forest (RF), Deep Neural Network (DNN), Support Vector Machine (SVM), LightGBM, and CatBoost. Model performance was assessed using the Area Under the Curve (AUC), accuracy, sensitivity, specificity, and F1 score. Model validation was performed through calibration and decision curve analysis. Feature importance was evaluated using SHAP, and subgroup analysis was conducted to assess the model’s applicability across different clinical scenarios. In internal validation, the Logistic Regression model performed the best, with an AUC of 0.86 (95% CI: 0.77–0.93), accuracy of 0.76, sensitivity of 0.73, specificity of 0.79, and an F1 score of 0.73. It outperformed other models (RF: AUC = 0.79, DNN: AUC = 0.78, SVM: AUC = 0.76, LightGBM: AUC = 0.71, CatBoost: AUC = 0.77). External validation yielded consistent results, with the Logistic Regression model’s AUC at 0.75 (95% CI: 0.56–0.92), accuracy of 0.69, sensitivity of 0.64, specificity of 0.73, and an F1 score of 0.66. Calibration curve analysis revealed that the Logistic Regression model had the lowest Brier score (0.1496), indicating the most reliable predicted probabilities. Decision curve analysis demonstrated that the model provided the highest net benefit across most decision thresholds. SHAP analysis identified lactate, age, and albumin as key predictors of mortality, with lactate and age positively correlated, and albumin negatively correlated. Subgroup analysis revealed better performance in the cardiac arrest group (AUC = 0.81), non-sepsis group (AUC = 0.75), and non-diabetes group (AUC = 0.78). The Logistic Regression-based mortality risk prediction model for V-A ECMO patients demonstrated comparable or even favorable performance to more complex machine learning models, with the advantage of higher interpretability.By explicitly incorporating lactate, age, and albumin as the principal predictors, this model facilitates precise risk stratification and provides practical support for clinical decision-making in ECMO management.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-25423-4.

It is estimated that 7% of women in the western world will develop palpable breast cysts in their lifetime. Even though cysts have been correlated with risk of developing breast cancer, many of them are benign and do not require follow-up. We develop a method to discriminate benign solitary cysts from malignant masses in digital mammography. We think a system like this can have merit in the clinic as a decision aid or complementary to specialized modalities. We employ a deep convolutional neural network (CNN) to classify cyst and mass patches. Deep CNNs have been shown to be powerful classifiers, but need a large amount of training data for which medical problems are often difficult to come by. The key contribution of this paper is that we show good performance can be obtained on a small dataset by pretraining the network on a large dataset of a related task. We subsequently investigate the following: (a) when a mammographic exam is performed, two different views of the same breast are recorded. We investigate the merit of combining the output of the classifier from these two views. (b) We evaluate the importance of the resolution of the patches fed to the network. (c) A method dubbed tissue augmentation is subsequently employed, where we extract normal tissue from normal patches and superimpose this onto the actual samples aiming for a classifier invariant to occluding tissue. (d) We combine the representation extracted using the deep CNN with our previously developed features. We show that using the proposed deep learning method, an area under the ROC curve (AUC) value of 0.80 can be obtained on a set of benign solitary cysts and malignant mass findings recalled in screening. We find that it works significantly better than our previously developed approach by comparing the AUC of the ROC using bootstrapping. By combining views, the results can be further improved, though this difference was not found to be significant. We find no significant difference between using a resolution of 100 versus 200 micron. The proposed tissue augmentations give a small improvement in performance, but this improvement was also not found to be significant. The final system obtained an AUC of 0.80 with 95% confidence interval [0.78, 0.83], calculated using bootstrapping. The system works best for lesions larger than 27mm where it obtains an AUC value of 0.87. We have presented a computer-aided diagnosis (CADx) method to discriminate cysts from solid lesion in mammography using features from a deep CNN trained on a large set of mass candidates, obtaining an AUC of 0.80 on a set of diagnostic exams recalled from screening. We believe the system shows great potential and comes close to the performance of recently developed spectral mammography. We think the system can be further improved when more data and computational power becomes available.