Non-training Data Research Articles

Modelling complex spatial–temporal drivers of habitat suitability for an imperilled stream fish

Fish populations rely on complex environmental conditions involving physical, chemical, and biological factors. Understanding the factors that control population persistence and productivity is essential for species management. We assessed the distribution and associated habitat features of a species at risk in Canada, Silver Shiner (Notropis photogenis), within Sixteen Mile Creek, a tributary of Lake Ontario. Using random forest models, we quantified a range of ecological factors (n = 25) to estimate habitat associations for sampled populations and life stages (juvenile, adult). A complex set of ecological factors were informative predictors of Silver Shiner distribution, including physical (stream morphology, water velocity, substrate type), and biological (aquatic and riparian vegetation) conditions. Juveniles were less responsive to habitat conditions but exhibited high seasonal variability in occurrence. Adults were most common in stream sections with greater than 0.5 m depth and stream velocity less than 0.6 m/s, and areas without silt substrate. Broadly, the models predicted Silver Shiner distribution with 68–92% accuracy in non-training data. Our findings describe the habitat conditions that Silver Shiner currently occupies in an urban drainage, which may serve as a point of reference for habitat protection and restoration. Further, predictive species distribution models can serve to identify habitat for further monitoring and restoration.

Evaluation of machine learning algorithms to predict internal concentration polarization in forward osmosis

Internal concentration polarization (ICP) is currently a major bottleneck in the forward osmosis process. Proper modelling of the internal concentration polarization is therefore vital for improving the process performance and efficiency. This study assessed the feasibility of several machine learning methods for internal concentration polarization prediction, including artificial neural networks, extreme gradient boosting (XGBoost), Categorical boosting (CatBoost), Random forest, and linear regression. Among the many algorithms evaluated, the CatBoost regression outperformed other methods in terms of coefficient of determination (R 2 ) and the mean square error. The CatBoost algorithm's prediction power was then evaluated using non-training (user-provided) data and compared to solution diffusion models. The results indicated that the machine learning algorithms could predict ICP in the process with high accuracy for the provided dataset and excellent generalizability for future testing data. Furthermore, machine learning algorithms may offer insights into the input features that majorly affect ICP modelling in the forward osmosis process. • Internal concentration polarization is predicted through machine learning models. • Artificial Neural networks and gradient tree boosting models are compared. • Neural network and Categorial boosting models showed better predictive power. • Categorical boosting outperformed other machine learning models. • Comparison against solution diffusion models is made.

Post hoc validation of a tool that accurately predicts the outcome of endoscopic therapy in Bouveret syndrome.

BackgroundBouveret syndrome is characterized by gallstone impaction in the upper gastrointestinal tract causing gastric outlet obstruction. In Bouveret syndrome, endoscopic gallstone removal can avert the need for surgery. However, in cases in which endoscopic therapy is unlikely to succeed, endoscopic attempts delay definitive treatment and compound patient risks. We previously developed a model that predicts endoscopic outcomes from data derived through a systematic review. This tool uses gallstone length, site of impaction, and the number of planned methods of lithotripsy to predict the likelihood of endoscopic success with an accuracy of 81.0%. This study aimed to evaluate our tool performance in an independent, non-training data set and assess endoscopic and surgical outcomes.MethodsSystematic searches of the PubMed, Scopus, and Cochrane databases were performed for articles published between 16 April 2018 and 1 June 2021. The data reported after our previous study were harvested and inputted into the tool to evaluate their ability to accurately predict outcomes when compared with actual outcomes.ResultsNewly collated data in fields of interest showed no significant statistical differences compared with previous training data sets. Endoscopic therapy was successful in 41.9% of cases. Gallstones of ≤4 cm had a higher chance of successful endoscopic intervention (odds ratio 6.7, 95% confidence interval 1.7–25.8, P < 0.01). Complications of surgery were reported in 29.5%; there was one fatality reported. Post hoc evaluation of our predictive tool demonstrated an AUROC score of 0.80.ConclusionsWe have demonstrated in an independent data set that the tool can be used to accurately predict outcomes of endoscopic therapy. Patients in whom endoscopic therapy is most likely to fail should be offered an early surgical opinion.

Development of a Surface Roughness Evaluation Method from Light and Shade Composition using Deep Learning

Many companies agree on the need to introduce smart factories that promote low-cost, high-efficiency operation by applying IoT technologies to manufacturing plants. To assist in efficient manufacturing, this paper proposes a system for evaluating surface roughness through deep learning AI methods from the distribution of shade on the surface of an object. This is thought to greatly relieve the meticulous and tedious process of manual quality control for precision-machined surfaces and assist in the establishment of smart manufacturing industries. To demonstrate the usefulness of the developed technique, 305 samples of paper were categorized into three classes based on Ra threshold values, and images of paper were taken using a microscope camera. Luminance values, standard deviations, mean values, and image histograms were used to train custom-designed CNN+ LSTM composite neural networks. Completely new samples of non-training data were used for validation, which showed an accuracy of 85.185%. The proposed method can be economical, efficient, and fast compared to conventional surface roughness evaluation procedures. It can be easily integrated into an assembly line and automate quality assurance processes. This method could also prove useful in reducing labor costs and streamlining quality assurance processes due to its flexible adaptability.

Brain-computer interface of focus and motor imagery using wavelet and recurrent neural networks

Brain-computer interface is a technology that allows operating a device without involving muscles and sound, but directly from the brain through the processed electrical signals. The technology works by capturing electrical or magnetic signals from the brain, which are then processed to obtain information contained therein. Usually, BCI uses information from electroencephalogram (EEG) signals based on various variables reviewed. This study proposed BCI to move external devices such as a drone simulator based on EEG signal information. From the EEG signal was extracted to get motor imagery (MI) and focus variable using wavelet. Then, they were classified by recurrent neural networks (RNN). In overcoming the problem of vanishing memory from RNN, was used long short-term memory (LSTM). The results showed that BCI used wavelet, and RNN can drive external devices of non-training data with an accuracy of 79.6%. The experiment gave AdaDelta model is better than the Adam model in terms of accuracy and value losses. Whereas in computational learning time, Adam's model is faster than AdaDelta's model.

Prospectivity Mapping of Mineral Deposits in Northern Norway Using Radial Basis Function Neural Networks

In this paper, the radial basis function neural network (RBFNN) is used to generate a prospectivity map for undiscovered copper-rich (Cu) deposits in the Finnmark region, northern Norway. To generate the input data for RBFNN, geological and geophysical data, including up to 86 known mineral occurrences hosted in mafic host-rocks, were combined at different resolutions. Mineral occurrences were integrated into “deposit” and “non-deposit” training sets. Running RBFNN on different input vectors, with a k-fold cross-validation method, showed that increasing the number of iterations and radial basis functions resulted in: (1) a reduction of training mean squared error (MSE) down to 0.1, depending on the grid resolution, and (2) reaching correct classification rates of 0.9 and 0.6 for training and validation, respectively. The latter depends on: (1) the selection of “non-deposit” training data throughout the study area, (2) the scale at which data was acquired, and (3) the dissimilarity of input vectors. The “deposit” input data were correctly identified by the trained model (up to 83%) after proceeding to classification of non-training data. Up to 885 km2 of the Finnmark region studied is favorable for Cu mineralization based on the resulting mineral prospectivity map. The prospectivity map can be used as a reconnaissance guide for future detailed ground surveys.

Application of semi-supervised fuzzy c-means method in clustering multivariate geochemical data, a case study from the Dalli Cu-Au porphyry deposit in central Iran

Supervised and unsupervised learning methods are widely used to classify and cluster multivariate geochemical data. Supervised learning methods incorporate training functions to classify the geochemical data, whereas unsupervised learning methods extract hidden structures of the data and assign them to various clusters. A semi-supervised learning method is a hybrid learning method that simultaneously extracts the hidden structure of non-training data and uses training data to improve the clustering analysis. In this research, initially eleven soil geochemical variables associated with the Dalli Cu-Au porphyry deposit, located in the central part of Iran, were selected by using hieratical clustering analysis and expert knowledge. Then, the semi-supervised fuzzy c-means clustering method (ssFCM) was used to separate multivariate soil geochemical anomalies from background for further drilling. The results were compared with the fuzzy c-mean clustering (FCM) analysis applied to the same samples. The fundamental concept of the ssFCM method is similar to the widely used FCM method with the exception that the training data, in this case trenching data, were used as an objective function in the clustering analysis. The soil classification results were validated by using cluster validity indices, cross-validation and the uncertainty measurement. The validation results demonstrated that the ssFCM method was superior in classifying the multivariate soil geochemical data compared to the FCM method. For further validation, the membership values of the favorable classes identified by both FCM and ssFCM methods were converted to grid maps and compared with the spatial distribution of copper anomalies along the trenches and surface projection of the boreholes. This comparison suggests that the favorable multivariate soil geochemical anomalies identified by the ssFCM analysis correlate well with copper mineralization in rock channel and drill core samples.

Data integration modeling applied to drill hole planning through semi-supervised learning: A case study from the Dalli Cu-Au porphyry deposit in the central Iran

In this study, the application of a transductive support vector machine (TSVM), an innovative semi-supervised learning algorithm, has been proposed for mapping the potential drill targets at a detailed exploration stage. The semi-supervised learning method is a hybrid of supervised and unsupervised learning approach that simultaneously uses both training and non-training data to design a classifier. By using the TSVM algorithm, exploration layers at the Dalli porphyry Cu-Au deposit in the central Iran were integrated to locate the boundary of the Cu-Au mineralization for further drilling. By applying this algorithm on the non-training (unlabeled) and limited training (labeled) Dalli exploration data, the study area was classified in two domains of Cu-Au ore and waste. Then, the results were validated by the earlier block models created, using the available borehole and trench data. In addition to TSVM, the support vector machine (SVM) algorithm was also implemented on the study area for comparison. Thirty percent of the labeled exploration data was used to evaluate the performance of these two algorithms. The results revealed 87 percent correct recognition accuracy for the TSVM algorithm and 82 percent for the SVM algorithm. The deepest inclined borehole, recently drilled in the western part of the Dalli deposit, indicated that the boundary of Cu-Au mineralization, as identified by the TSVM algorithm, was only 15 m off from the actual boundary intersected by this borehole. According to the results of the TSVM algorithm, six new boreholes were suggested for further drilling at the Dalli deposit. This study showed that the TSVM algorithm could be a useful tool for enhancing the mineralization zones and consequently, ensuring a more accurate drill hole planning.

Identification of Viruses Using Microfluidic Protein Profiling and Bayesian Classification

We present a rapid method for the identification of viruses using microfluidic chip gel electrophoresis (CGE) of high-copy number proteins to generate unique protein profiles. Viral proteins are solubilized by heating at 95 degrees C in borate buffer containing detergent (5 min), then labeled with fluorescamine dye (10 s), and analyzed using the microChemLab CGE system (5 min). Analyses of closely related T2 and T4 bacteriophage demonstrate sufficient assay sensitivity and peak resolution to distinguish the two phage. CGE analyses of four additional viruses--MS2 bacteriophage, Epstein-Barr, respiratory syncytial, and vaccinia viruses--demonstrate reproducible and visually distinct protein profiles. To evaluate the suitability of the method for unique identification of viruses, we employed a Bayesian classification approach. Using a subset of 126 replicate electropherograms of the six viruses and phage for training purposes, successful classification with non-training data was 66/69 or 95% with no false positives. The classification method is based on a single attribute (elution time), although other attributes such as peak width, peak amplitude, or peak shape could be incorporated and may improve performance further. The encouraging results suggest a rapid and simple way to identify viruses without requiring specialty reagents such as PCR probes and antibodies.

Studies on Human Skin Extraction from Hyperspectral Data using Particle Swarm Optimization

If the automatic detection of the victims is feasible after large-scale disasters, the search efficiency would be significantly improved compared to the traditional visual search method from aircraft. In order to extract the human skin correctly, we focus on the training of an Artificial Neural Network (ANN) by the hyperspectral data. From the standpoint of avoiding the over-fitting problem generated by multi-channel inputs of the hyperspectral data, it is necessary to be trained by the optimal bands. In this paper, we propose the coupled search method for the optimal number and combination of bands in order to extract the target from the hyperspectral data. The coupled search method is composed of two search methods. The first is a new search method of the maximum value of the evaluation function which searches the optimal number of bands from the non-training data. The other is the search method using the Particle Swarm Optimization which searched the optimal combination of bands from the training data. The ANN is trained by the selected combination of bands, and the results are evaluated. Moreover, the trained network obtained using the coupled search method is compared with the ANN trained by all the bands and the Normalized Difference Human Index.

Inter-crystal scatter identification for a depth-sensitive detector using support vector machine for small animal positron emission tomography

In a conventional positron emission tomography (PET) detector, detected events are projected onto a 2D position histogram by an Anger calculation for crystal identification. However, the measured histogram is affected by inter-crystal scatterings (ICS) which occur in the entire detector. Peaks which are projected for each crystal in the histogram are blurred, and this causes ICS mispositioning. A depth-of-interaction (DOI) detector has been developed for the small animal PET scanner jPET-RD. This DOI detector uses 32×32 crystals with four layers and a 256-channel multi-anode flat panel photomultiplier tube (FP-PMT) which was developed by Hamamatsu Photonics K.K. Each crystal element is 1.45×1.45×4.5 mm 3. The FP-PMT has a large detective area (49×49 mm 2) and a small anode pitch (3.04 mm). Therefore, the FP-PMT can extensively trace the behavior of incident γ rays in the crystals including ICS event. We, therefore, propose a novel method for ICS estimation using a statistical pattern recognition algorithm based on a support vector machine (SVM). In this study, we applied the SVM for discriminating photoelectric events from ICS events generated from multiple-anode outputs. The SVM was trained by uniform irradiation events generated from a detector simulator using a Monte Carlo calculation. The success rate for ICS event identification is about 78% for non-training data. The SVM can achieve a true subtraction of ICS events from measured events, and it is also useful for random correction in PET.