AI-based rock strength assessment from tunnel face images using hybrid neural networks Scientific Reports

Among these, content similarity is pivotal in learners’ online learning compared to average sentence length. It is essential to note that this work currently tests the effectiveness of CDA on only three types of English and Chinese courses in secondary schools. Future efforts will involve designing experiments to investigate whether similar characteristics and patterns exist in the classroom discourse of other disciplines. The ultimate goal is to offer methods and references for educators to enhance classroom discourse and strengthen teaching effectiveness. The basic segmentation is evaluated based on a dice similarity coefficient (DSC) that measures the overlap between the prediction and ground truth (Supplementary Fig. S2). Our model averaged 0.867 for raw outputs of the test set and 0.853 for the post-processed images (Supplementary Fig. S3).

However, the model’s score corresponding to Black patients shows a different pattern in MXR, demonstrating much smaller variation by window width and field of view. Thus, while there is some variation across datasets, varying the window width and field of view parameters can generate relatively large changes in the average predictions of the AI model by patient race. Such disparities in technical data acquisition and processing factors may exist in many imaging domains14,21,22,23 and are of particular concern from an AI perspective. These risks are further exacerbated by the common practice of adapting AI approaches from natural image tasks, which may not fully take advantage of the acquisition and processing parameters unique to medical images. You can foun additiona information about ai customer service and artificial intelligence and NLP. Thus, it is paramount to study the influence of medical image acquisition factors on AI behavior, especially in the context of bias.

2 Organization of this study

However, since the amount of training data available was limited, it was decided to reduce the complexity of the original VGG16 architecture. Hence, the fifth convolutional block of the original VGG16 architecture was removed and an ai based image recognition average pooling layer was added, followed by the two dense layers. To avoid overfitting, data augmentation was used with several augmentation techniques such as rotating, vertical flipping, zooming and different brightness levels.

• Specifically, Blocks in CNNs can contain various layers, such as convolutional, pooling, and fully connected layers.
• A system like this wouldn’t just rock humankind to its core — it could also destroy it.
• As ECGs have transitioned from analog to digital, automated computer analysis has gained traction and success in diagnoses of medical conditions (Willems et al., 1987; Schlapfer and Wellens, 2017).

Libraries were constructed using the ThruPlex DNA-seq kit (Takara) with seven cycles of amplification (library prep strategy from Brenton Lab similar to the one published in 2018)69. Library quality was assessed using the Agilent High Sensitivity DNA kit (Agilent Technologies), and pooled libraries were run on the Illumina NovaSeq at the Michael Smith Genome Sciences Center targeting 600 M reads per pooled batch. The sWGS data was run through basic processing which includes trimming with Trimmomatic70, alignment with bwa-mem271, duplicate removal with Picard72, and sorting with samtools73. If acceptable, the data was passed along to the next step of determining genomic copy numbers (QDNAseq75 + rascal76) and signature calls. The signature calling step uses techniques including mixture modeling and non-negative matrix factorization and is composed mostly of software from the CN-Signatures69 package with a few in-house modifications and additions. Interim data munging and ETL (extract, transform, load) are done primarily in bash and R (tidyverse), while visualization and plotting is performed mostly just in R using ggplot2 and pheatmap.

The maximum temperatures T1, T2, T3…Tn were extracted for each region, and the hotspot temperature max(T1, T2, T3…Tn) and the normal temperature min(T1, T2, T3…Tn) were selected. If the temperature difference exceeds 2 K, it is determined that the Bushing has occurred a potential-heating fault; otherwise it is determined to be normal. The ecommerce panorama has witnessed an excellent transformation because of the developments in artificial intelligence (AI), recently.

Common data sets and evaluation indicators

AI data classification is transforming data management by sorting and analyzing data quickly and accurately, helping businesses stay ahead. It empowers organizations to identify their data types, locations, and handle sensitive information securely. Moving forward, AI’s role in data analysis will grow, deep learning will become more common, and AI will incorporate technologies like cloud computing and big data analytics, elevating data classification further. AI data classification tools aid healthcare professionals in interpreting medical images, such as X-rays, MRI scans, and pathology slides. ML algorithms are trained on labeled datasets containing images with corresponding diagnoses. Lazy learners specialize in handling complex and nonlinear data, making them suitable for real-world applications.

These frameworks have been extensively documented in the existing literature for the prescribed vegetables such as tomato, chili, potato, and cucumber. The captured images contain various factors such as noise, blur, low or high illumination, unwanted background, etc. Therefore, it is crucial to process this raw data and make it worthy to classify the disease efficiently using automatic approaches. The raw data is converted into a specific format and cleaned up by removing any noise or distortion. In the next phase, images are passed to the step where the essential segmentation and feature extraction procedures are carried out.

The view position indicates the position of the patient with respect to the X-ray source. Typical view positions used in chest X-rays are anterior-posterior (AP), posterior-anterior (PA), and lateral (Fig. 1a). In addition, the X-ray equipment itself may be a standard, stationary machine or a portable device that can be moved as necessary to image the patient. To gain deeper insights into the enhancement brought about by FFT-Enhancer on model performance, we examine the heatmaps generated by both ADA and AIDA for three samples that were accurately classified by both methods, as depicted in Fig. While both approaches achieved the correct classification for these samples, a noticeable distinction arises in the heatmap output. Notably, the heatmap produced by AIDA demonstrates a closer resemblance to the annotated areas.

Design of an accurate IR model combining densenet and GQ

The deformable convolution module introduces an offset to the sampling points, as illustrated in Fig. The top part generates the index offset by processing the input feature map through a regular convolution layer, while the bottom part convolves the input feature map with the corresponding kernel to produce the output feature map14. The deformable convolution kernels are capable of adapting to the extraction of complex noise patterns in images. Image denoising involves processing degraded images that contain noise to estimate the original image. Traditional Denoising Convolutional Neural Networks (Dn-CNN) use a fixed 3 × 3 convolutional kernel for noise feature extraction in images. However, Dn-CNN mainly learns noise information from images containing noise, without accommodating shape rules, which limits the effectiveness of feature extraction with a fixed-shape convolutional kernel13.

Substations serve as fundamental units within the power system, primarily responsible for the reception, transformation, and distribution of electric energy. They house critical electrical equipment, including potential transformers, current transformers, circuit breakers, and switches1. The collective functioning and stable operation of this equipment are pivotal for ensuring the safety and reliability of power transmission. Prompt and accurate detection of abnormal temperatures is vital for assessing the operational status of electrical equipment, playing a crucial role in maintaining the safety and stability of substations5. As far back as 2008, researchers were showing how bots could be trained to break through audio CAPTCHAs intended for visually impaired users.

For example, the analysis is not systematic enough, the source of evaluation indicators is unclear, and there is no further in-depth analysis and research on various indicators. Based on this, this work addresses a current research gap by comprehensively analyzing discourse within secondary school-oriented classrooms. Focusing on the unique characteristics of the secondary school teaching environment, the present work explores the expressive features of classroom discourse and its correlation with teaching effectiveness.

Heatmaps were generated for each WSI to visualize the spatial distribution of tumors. This was accomplished by converting the prediction probability results of each patch into colors on WSI heatmaps. A higher classification score in tumor prediction is represented by a closer color to red in the heatmap image, indicating a higher likelihood of a tumor diagnosis.

PowerAI Vision can be used to deploy a deep learning model on factory floors to ensure little decision latency during production and deliver reliable results with low escape rates. Determine and label the contents of an image based on user-defined data labels (for example, “Locate and label all dogs in the image”). The methods of seeding the mechanically dissociated organoids for viability assay are aforementioned. The fluorescence intensity of captured images was quantified using ImageJ33, and the normalized data were plotted using GraphPad Prism 9 software. The increase in cell number is crucial in both 2D cell line culture and 3D organoid culture2,16,25,26,27,28,29. Because 2D cells are maintained as a single cell type, it is relatively easy to count the cell number and anticipate its culture conditions26,29,30.

The latter-layer feature maps, on the other hand, contain additional semantic information that is required for detecting and classifying things like distinct object placements and illuminations. Higher-level feature maps are valuable for classifying large objects, but they may not be enough to recognize small ones. Figure 3 Performance assessment of single-stage Object detection algorithms in different datasets. The technique eliminates the stage of generating candidate regions and combines feature extraction, regression, and classification into a single volume.

A total of 3663 image samples were used during training and testing, all carefully selected from the extensive PlantVillage dataset. The system’s output demonstrates an impressively high accuracy rate (87%).Similarly, researchers (Basavaiah and Anthony, 2020) observed the practice of various ML approaches to identify tomato plant disease. Texture, color, and form were used since they are well-known global feature descriptors. The authors used KNN, LR, DT, RF, SVM, and other algorithms for model training. The RF model outperformed many other ML algorithms in our analysis with an impressive 94% accuracy rate (Table 5).

• Our model averaged 0.867 for raw outputs of the test set and 0.853 for the post-processed images (Supplementary Fig. S3).
• Squeeze-and-excitation networks (SENet) add attention in the channel dimension.
• In HAR, particularly in sports, Cem Direkoglu et al.11 introduced an approach for team activity recognition based on known player positions.
• Determine why you need AI data classification—is it to enhance customer experience, predict future trends, or detect anomalies?
• Among the metrics used for the development and evaluation of OrgaExtractor (Supplementary Table S3), the projected area, perimeter, major axis length, and eccentricity were visualized through diagrams.

The resulting image serves as a representative image from the source dataset during the training phase. Moreover, our study revealed that the top patches of slides exhibited subtype-specific histologic features, such as tumor epithelium, while the bottom five patches predominantly contained nonspecific stromal or necrotic areas. We employed a class-discriminative localization method to identify and highlight the relevant histological features on these patches.

This integration of multi-scale encoder features and skip connections at matching resolutions allows the transfer of fine-grained local information to the decoder. This multi-resolution representation capability enables the model to produce highly accurate segmentation masks. This augmentation process is crucial for enhancing the model’s robustness and helps expand the dataset, thereby improving model performance. Each augmentation method contributes to creating a diverse set of training images, which helps in reducing overfitting and improving generalization. During tunnel construction, assessing the strength of the rock at the tunnel face is crucial due to the complex and variable geological conditions, which pose significant challenges for accurate evaluation.

In addition, to reduce false positives we used a minimum threshold probability of 90% for tumor patches. Finally, for consistency, we applied the trained model on the discovery set, including the cases that were manually annotated by a pathologist. In summary, various fields have extensively studied models for IR classification and processing, resulting in improved recognition accuracy.

The method uses an image enhancement technique, enhancing the effectiveness of the Convolutional Neural Network (CNN) model. The optimized CNN model includes four preprocessing stages, including filter width variations, hyper-parameter optimization, max-pooling, and dropout layers, yielding promising results. The optimized CNN model, trained for 25 epochs, achieved an accuracy rate of 99.99% (Table 7). To avoid the issue, a third modification was done and it was decided to train the entire model, instead of using transfer learning.

In 2006, Geoffrey Hinton and his students published a paper related to deep learning (Hinton and Salakhutdinov, 2006), which opened the door to object detection and recognition using deep learning. Under the fully convolutional network, similar to SSD, RON uses VGG-16 as the backbone network, the difference is that RON ChatGPT changes the 14th and 15th fully connected layers of the VGG-16 network into a kernel size of 2 × 2. In tests, RON achieves state-of-the-art object detection performance, with input 384×384 size images, the mAP reaches 81.3% on the PASCAL VOC2007 dataset, and the mAP improves to 80.7% on the PASCAL VOC 2012 dataset.

For this strategy and the test set resampling approach, we evaluate the originally trained AI models without modification. The training set resampling approach requires training new models, which we then evaluate on the resampled test sets. Various techniques aim to mitigate generalization errors in histopathology images by manipulating color spaces, categorized into stain color augmentation and normalization. Augmentation simulates ChatGPT App diverse stain variations for stain-invariant models, while normalization aligns training and test color distributions to reduce stain variation. Within the domain of color augmentation, methodologies range from basic techniques to advanced H&E-based approaches21,22,23. Typically, these methods involve direct modifications to images within the H&E color space, aiming to replicate specific variations in H&E staining.