Learning Solutions Research Articles

Deep Learning Approaches for Medical Image Analysis and Diagnosis.

In addition to enhancing diagnostic accuracy, deep learning techniques offer the potential to streamline workflows, reduce interpretation time, and ultimately improve patient outcomes. The scalability and adaptability of deep learning algorithms enable their deployment across diverse clinical settings, ranging from radiology departments to point-of-care facilities. Furthermore, ongoing research efforts focus on addressing the challenges of data heterogeneity, model interpretability, and regulatory compliance, paving the way forseamless integration of deep learning solutions into routine clinical practice. As the field continues to evolve, collaborations between clinicians, data scientists, and industry stakeholders will be paramount in harnessing the full potential of deep learning for advancing medical image analysis and diagnosis. Furthermore, the integration of deep learning algorithms with other technologies, including natural language processing and computer vision, may foster multimodal medical data analysis and clinical decision support systems to improve patient care. The future of deep learning in medical image analysis and diagnosis is promising. With each success and advancement, this technology is getting closer to being leveraged for medical purposes. Beyond medical image analysis, patient care pathways like multimodal imaging, imaging genomics, and intelligent operating rooms or intensive care units can benefitfrom deep learning models.

Abstract PO3-07-05: Multi-site validation of a deep learning solution for ER/PR profiling of breast cancer from H&E-stained pathology slides

Abstract Background: Molecular profiling of estrogen and progesterone receptors (ER/PR/Her2) is performed for all malignant breast cancers to inform the choice of targeted therapy. Though existing scoring systems are widely used and well-validated, they can involve costly preparation and variable interpretation. Additionally, discordances between histology and expected biomarker findings can prompt repeat testing to address biological, interpretative, or technical reasons for unexpected results. We evaluate PANProfiler Breast(PPB), a UKCA/CE- IVDD marked deep learning (DL)-based image analysis software, on multiple sites to determine if the majority of ER/PR assays can be replaced, relying only on routinely-used H&E-stained whole slide images. Methods: PPB was trained and validated on 5126/4619 WSIs from 5 sites to identify the ER/PR status defined by IHC assays graded in alignment with ASCO/RCPATH guidelines from five different sites in the UK were used for training and validation. The performance is evaluated separately for each site with 3-fold cross-validation, mimicking real-world distribution. Results: For ER, with a class ratio (CR) of approximately 4:1, we measure a sensitivity, specificity, and accuracy of 95.5%(±1.5%), 47.5%(±15.1%) and 88.3%(±0.6%) averaged over all sites, reaching up to 97.40%, 67.20%, and 89.30% respectively. For PR (CR approx. 3:1), the averaged sensitivity, specificity, and accuracy are 92.2%(±8.4%), 53.10%(±20.8%), and 86.6%(±2.4%), reaching up to 99.1%, 81.7%, and 88.8%, respectively. The software's performance is comparable to current SoC antibody performance in common ER/PR CDx Assays from Dako, Leica, and Roche, which have sensitivities of 98.5%(±1.3%) and specificities of 38.6%(±6.3%) for ER and sensitivities of 96.9%(±0.6%) and specificities of 23.4%(±1.4%) for PR. Performance was robust to specimen and scanner types, with accuracies of 87.6% (ER, only biopsies), 88.2% (ER, only resections), 88.5% (ER mixed types), 83.9% (PR, only resections) and 87.3% (mixed types). Accuracy across scanners varied by a standard deviation of 0.3%/1.0% for ER/PR respectively. Conclusions: We demonstrate the robustness of a DL-based ER/PR profiling method in breast cancer using only H&E-stained WSIs. This multi-site validation study is the first-of-its-kind for such an approach using real-world clinical data. Our solution could facilitate fast, accurate, and systemic screening of patients for targeted treatments if integrated into routine pathological workflows. Citation Format: Salim Arslan, Adrian Bazaga, Gareth Bryson, Oscar Carlos, Andre Geraldes, David Harrison, Alastair Ironside, Jakob Kather, Ali Khurram, David Leff, Debapriya Mehrotra, Foivos Ntelemis, John Nyonyintono, Julian Schmidt, Shikha Singhal, In Hwa Um, Steffen Wolf, Pahini Pandya. Multi-site validation of a deep learning solution for ER/PR profiling of breast cancer from H&E-stained pathology slides [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO3-07-05.

Nonlinear dynamic analysis of aircraft CFRP sandwich wings under explosive blast loading: Introducing SVM-DNN algorithm to predict dynamical information

The aircraft wings are a crucial component, particularly during mechanical shock excitation. Thus, for the first time, an inventive adaptable model to simulate the nonlinear dynamics of the wings under explosive blast loading at the moment of the collision is described in this article. It is crucial to increase the structure's stability because of explosive blast loading. The mentioned structure is made of two composite face sheets and a carbon fiber-reinforced polymer (CFRP). Functionally graded (FG) graphene origami (GOri)-enabled auxetic metallic metamaterial (GOEAM) is introduced to improve the stability of this kind of applicable structure, especially in air maneuvers. The analysis is conducted utilizing Reddy's third-order shear deformation of Reddy theory (TSDRT), incorporating Hamilton's principle and Von Kármán nonlinear geometric assumptions. The governing and boundary equations are accurately computed at the domain and boundary edges of the doubly curved structure. The coupled mesh-free radial point interpolation method (MRPIM) and Newton-Raphson method are applied for numerically simulating the current doubly curved structure by employing Newmark's temporal integration method with a constant-average acceleration approach. The present findings are cross-checked against the machine learning method and open-source results from the literature. Based on physical data, machine learning integrates supervised machine learning to predict the nonlinear vibration of the system. In this case, nonlinear transient bending properties are found using hybrid machine learning methods and solutions. The findings demonstrate the significance of GOEAM for nonlinear vibrations of the composite system. The current study's conclusions could be used as a benchmark and useful recommendations for future structural design techniques with enhanced features.

Dynamic Weighting Translation Transfer Learning for Imbalanced Medical Image Classification.

Medical image diagnosis using deep learning has shown significant promise in clinical medicine. However, it often encounters two major difficulties in real-world applications: (1) domain shift, which invalidates the trained model on new datasets, and (2) class imbalance problems leading to model biases towards majority classes. To address these challenges, this paper proposes a transfer learning solution, named Dynamic Weighting Translation Transfer Learning (DTTL), for imbalanced medical image classification. The approach is grounded in information and entropy theory and comprises three modules: Cross-domain Discriminability Adaptation (CDA), Dynamic Domain Translation (DDT), and Balanced Target Learning (BTL). CDA connects discriminative feature learning between source and target domains using a synthetic discriminability loss and a domain-invariant feature learning loss. The DDT unit develops a dynamic translation process for imbalanced classes between two domains, utilizing a confidence-based selection approach to select the most useful synthesized images to create a pseudo-labeled balanced target domain. Finally, the BTL unit performs supervised learning on the reassembled target set to obtain the final diagnostic model. This paper delves into maximizing the entropy of class distributions, while simultaneously minimizing the cross-entropy between the source and target domains to reduce domain discrepancies. By incorporating entropy concepts into our framework, our method not only significantly enhances medical image classification in practical settings but also innovates the application of entropy and information theory within deep learning and medical image processing realms. Extensive experiments demonstrate that DTTL achieves the best performance compared to existing state-of-the-art methods for imbalanced medical image classification tasks.

Machine learning solution for regional landslide susceptibility based on fault zone division strategy

Machine learning solution for regional landslide susceptibility based on fault zone division strategy

844: Head-to-head comparison for breast RT planning: knowledge-based planning vs. deep learning solution

844: Head-to-head comparison for breast RT planning: knowledge-based planning vs. deep learning solution

Context Aware Music Recommendation and Playlist Generation

There are many reasons people listen to music, and the type of music is largely determined by what the listener may be doing while they listen. For example, one may listen to one type of music while commuting, another while exercising, and yet another while relaxing. Without access to the physiological state of the user, current music recommendation methods rely on collaborative filtering - recommending music based on what other similar users listen to - and content based filtering - recommending songs based on their similarities to songs the user already prefers. With the rise in popularity of smart devices and activity trackers, physiological context can be a new channel to inform music recommendations. We propose deep learning solutions for context aware recommendation and playlist generation. Specifically, we use variational autoencoders (VAEs) to create a song embedding. We then explore multi-task multi-layer perceptrons (MLPs) and Gaussian mixture models to recommend songs based on context. We generate artificial user data to train and test our models in online learning and supervised learning settings.

기업 HRD 관점에서 본 마이크로러닝의 실천적 정의 및 활용 특징 탐색

Objectives This study was conducted to identify the definition of microlearning, misconceptions and corrections, differences from traditional e-learning, and characteristics of microlearning as recognized by the corporate HRD. Methods To identify and analyze the reality of microlearning, 16 microlearning experts from companies and universities were nominated and conducted a Delphi Survey in three rounds. Results We were able to define microlearning from three perspectives: purpose of utilization, learning method, and learning form. Second, we provided corrective opinions on the misconceptions of microlearning (“It's just a short lecture,” “Learning time should not exceed 15 minutes,” “It only includes videos,” etc. Third, we compared the characteristics of microlearning with traditional e-learning and explored our own characteristics. These features were presented in terms of learning goals, learning content, learning activities, and learning time. Conclusions Through this study, we have prepared basic data to promote the proper understanding and practice of microlearning and to activate microlearning as a digital learning solution in the corporate HRD.

Live Stock Shelter Management System using IoT and ML

In India, domestic animals are vital contributors to the country's economy, serving as a significant source of income for farmers and fulfilling various work and service roles. Keeping these animals safe and healthy in their shelters is really important because they mean a lot to farmers. Livestock farming faces big challenges, especially when it's raining, because bad weather can make the animals sick and less productive, causing problems like hoof issues.This project proposes an affordable shelter solution for Indian livestock, integrating IoT and machine learning to monitor environmental conditions. Sensors track temperature, humidity, gas levels, and soil moisture, with data analysed using the Arduino IDE and K means algorithm. Through predictive analysis, the system detects potential abnormalities like high temperatures, excessive humidity, elevated soil moisture, or the presence of harmful gases within the shelter. The predictive analysis also identifies specific stress conditions, such as heat stress from high temperatures, cold stress because, the temperature drops below the lower critical temperature, hoof diseases caused by increased soil moisture, and respiratory issues from harmful gas presence. Timely notifications are sent to user upon identifying such irregularities, enabling prompt interventions to mitigate health risks and improve herd welfare and productivity

A comprehensive overview of core modules in visual SLAM framework

Visual Simultaneous Localization and Mapping (VSLAM) technology has become a key technology in autonomous driving and robot navigation. Relying on camera sensors, VSLAM can provide a richer and more precise perception means, and its advancement has accelerated in recent years. However, current review studies are often limited to in-depth analysis of a specific module and lack a comprehensive review of the entire VSLAM framework. The VSLAM system consists of five core components: (1) The camera sensor module is responsible for capturing visual information about the surrounding environment. (2) The front-end module uses image data to roughly estimate the camera’s position and orientation. (3) The back-end module optimizes and processes the pose information estimated by the front-end. (4) The loop detection module is used to correct accumulated errors in the system. (5) The mapping module is responsible for generating environmental maps. This review provides a systematic and comprehensive analysis of the SLAM framework by taking the core components of VSLAM as the entry point. Deep learning brings new development opportunities for VSLAM, but it still needs to solve the problems of data dependence, cost and real-time in practical application. We deeply explore the challenges of combining VSLAM with deep learning and feasible solutions. This review provides a valuable reference for the development of VSLAM. This will help push VSLAM technology to become smarter and more efficient. Thus, it can better meet the needs of future intelligent autonomous systems in multiple fields.

Using Machine Learning for Chronic Disease Diagnosis and Prediction

Machine learning had a strong effect on day-to-day life processes, and in the future, it's going to continue influencing this field. A key where machine learning illustrates promise in healthcare, especially in diagnosing and predicting chronic diseases, conditions that last for at least a year. This paper examines various methods through which machine learning is used in healthcare, consider how these models might be improved to perform specialized tasks, like detecting cancer from individual cells, as they are used to analyse medical images. The paper also explores the inner workings of machine learning and enlightens on some of its current limitations. The article addresses concerns such as model bias and the significant expenses associated with implementing machine learning solutions

Few-shot satellite image classification for bringing deep learning on board OPS-SAT

Bringing artificial intelligence on board Earth observation satellites unlocks unprecedented possibilities to extract actionable items from various image modalities at the global scale in real time. This is of paramount importance nowadays, as downlinking large amounts of imagery is not only prohibitively expensive but also time-consuming. However, building deep learning solutions that could be deployed on board an edge device is challenging due to the limited manually-annotated satellite datasets and hardware constraints of an edge device. This paper addresses these challenges through harnessing a blend of data-centric and model-centric approaches to build a well-generalizing yet efficient and resource-frugal deep learning model for multi-class satellite image classification in the few-shot learning settings. This integrated strategy is formulated to enhance classification performance while accommodating the unique demands of an image analysis chain on board OPS-SAT, a nanosatellite operated by the European Space Agency. The experiments performed over a real-world dataset of OPS-SAT images delves into the interactions between data- and model-centric techniques, underscores the significance of synthesizing artificial training data and emphasizes the value of ensemble learning. However, they also caution against negative transfer in domain adaptation. This study sheds light on effective model training strategies and highlights the multifaceted challenges inherent in deep learning for practical Earth observation, contributing insights to the field of satellite image classification within the constraints of nanosatellite operations. To ensure reproducibility of our study, the implementation is available at https://github.com/ShendoxParadox/Few-shot-satellite-image-classification-OPS-SAT.

A Metadata-Enhanced Deep Learning Method for Sea Surface Height and Mesoscale Eddy Prediction

Predicting the mesoscale eddies in the ocean is crucial for advancing our understanding of the ocean and climate systems. Establishing spatio-temporal correlation among input data is a significant challenge in mesoscale eddy prediction tasks, especially for deep learning techniques. In this paper, we first present a deep learning solution based on a video prediction model to capture the spatio-temporal correlation and predict future sea surface height data accurately. To enhance the performance of the model, we introduced a novel metadata embedding module that utilizes neural networks to fuse remote sensing metadata with input data, resulting in increased accuracy. To the best of our knowledge, our model outperforms the state-of-the-art method for predicting sea level anomalies. Consequently, a mesoscale eddy detection algorithm will be applied to the predicted sea surface height data to generate mesoscale eddies in future. The proposed solution achieves competitive results, indicating that the prediction error for the eddy center position is 5.6 km for a 3-day prediction and 13.6 km for a 7-day prediction.

Differentiating stable and unstable protein using convolution neural network and molecular dynamics simulations

Protein stability is a critical aspect of molecular biology and biochemistry, hinges on an intricate balance of thermodynamic and structural factors. Determining protein stability is crucial for understanding and manipulating biological machineries, as it directly correlated with the protein function. Thus, this study delves into the intricacies of protein stability, highlighting its dependence on various factors, including thermodynamics, thermal conditions, and structural properties. Moreover, a notable focus is placed on the free energy change of unfolding (ΔGunfolding), change in heat capacity (ΔCp) with protein structural transition, melting temperature (Tm) and number of disulfide bonds, which are critical parameters in understanding protein stability. In this study, a machine learning (ML) predictive model was developed to estimate these four parameters using the primary sequence of the protein. The shortfall of available tools for protein stability prediction based on multiple parameters propelled the completion of this study. Convolutional Neural Network (CNN) with multiple layers was adopted to develop a more reliable ML model. Individual predictive models were prepared for each property, and all the prepared models showed results with high accuracy. The R2 (coefficient of determination) of these models were 0.79, 0.78, 0.92 and 0.92, respectively, for ΔG, ΔCp, Tm and disulfide bonds. A case study on stability analysis of two homologous proteins was presented to validate the results predicted through the developed model. The case study included in silico analysis of protein stability using molecular docking and molecular dynamic simulations. This validation study assured the accuracy of each model in predicting the stability associated properties. The alignment of physics-based principles with ML models has provided an opportunity to develop a fast machine learning solution to replace the computationally demanding physics-based calculations used to determine protein stability. Furthermore, this work provided valuable insights into the impact of mutation on protein stability, which has implications for the field of protein engineering. The source codes are available at https://github.com/Growdeatechnology.

Learning Management of Islamic Religious Education for Children with Intellectual Disabilities: Multicase Study

This research aims to examine the learning of Islamic Education for children with intellectual disabilities in Sekolah Luar Biasa (SLB) Negeri Binjai and Sekolah Luar Biasa Negeri 1 Binjai, including planning, implementation, learning evaluation, as well as barriers and solutions in learning. The research method uses a qualitative approach with a multi-site study to explore the learning process at SLB Negeri Binjai and SLB Negeri 1 Binjai The research findings indicate that Islamic Education learning in both SLBs includes the introduction of basic Islamic concepts, teaching methods tailored to the needs of students, and parental involvement in the learning process. However, there are still obstacles such as limited resources and a lack of training for Islamic Education teachers. In conclusion, this research provides a significant contribution to efforts to improve the quality of Islamic education for children with special needs SLB Negeri Binjai and SLB Negeri 1 Binjai.

Reliable arrival time picking of acoustic emission using ensemble machine learning models

This study presents an innovative method for accurately picking the first-wave arrival time in acoustic emission (AE) localization, particularly effective in environments with low or variable signal-to-noise ratios (SNR). Utilizing an ensemble learning model, it synergizes multiple automatic arrival time estimation algorithms to enhance both consistency and robustness. The model, rooted in decision tree methodologies, integrates a variety of techniques, including Akaike information criterion (AIC), improved AIC, Hinkley, energy ratios, and wavelet transformation-based binary map. Its efficacy is demonstrated through testing on 549 manually annotated AE datasets, where the model’s predictions were benchmarked against manual picks, showcasing superior accuracy and robustness in AE event source localization. Evaluations of the model’s performance across various ensemble machine learning models highlighted its ability to significantly diminish localization errors, achieving an average absolute error of less than 1.5 mm. The study also delved into the impact of base pickers and the size of the training dataset on the model’s predictions. Findings indicated consistent performance across different decision tree models, with the accuracy of base pickers and training set size playing a significant role in outcomes. This research culminates in a decision tree-based ensemble machine learning solution that effectively estimates AE first-wave arrival times with high accuracy and robustness, even amidst fluctuating SNR conditions. Its adaptability and interpretability greatly reduce source localization errors in practical AE monitoring, effectively overcoming challenges associated with imprecise arrival time picking.

Multi-view GCN for loan default risk prediction

As a significant application of machine learning in financial scenarios, loan default risk prediction aims to evaluate the client’s default probability. However, most existing deep learning solutions treat each application as an independent individual, neglecting the explicit connections among different application records. Besides, these attempts suffer from the problem of missing data and imbalanced distribution (i.e., the default records are small samples against all the applications). We believe similar records could provide some auxiliary signals, which are of critical importance to alleviate the data missing issue and facilitate data argumentation. To this end, we propose multi-view loan application graphs, dubbed MLAGs. By evaluating the similarity between the records, a loan application graph can be constructed. Furthermore, we arrange different similarity thresholds to organize various graph structures for multi-graph constructions; thus, a variety of representations can be generated via information propagation and aggregation for small sample argumentation. Consequently, the imbalanced data distribution and missing values issues can be alleviated effectively. We conduct experiments on three public datasets from real-world home credit and P2P lending platforms, which show that MGCN outperforms both conventional and deep learning models. Ablation studies also illustrated the validity of each module design.

Enhanced privacy-preserving distributed deep learning with application to fog-based IoT

Generally, privacy-preserving distributed deep learning (PPDDL) solutions provided so far present a trade-off between privacy and efficiency/effectiveness, especially, in terms of high communication and run-time costs (i.e. efficiency) and declined accuracy (i.e. effectiveness). Additionally, not much attention has been paid to proposing privacy-preserving solutions that are multi-key compliant, able to extend to new participants and well fit for an unbounded number of participants as well as being post-quantum robust. Also, application of DDL to fog-based IoT needs more attention. To this effect, we present enhanced PPDDL by proposing two solutions (basic and advanced). The basic combines LWE-based additive homomorphic encryption and partial sharing, making it communication cost friendly. The advanced, however, uses a lattice-based fully dynamic multi-key fully homomorphic encryption (FHE) scheme and partial sharing with the additional advantages of being multi-key compliant, able to extend to new participants and well fit for an unbounded number of participants. Furthermore, we put our work in the context of Fog-Based IoT. Extensive experimental evaluations for the basic solution which also best suits the Fog-Based IoT, show that the PPDDL accuracy is maintained, recording accuracy of about 97% for the MNIST dataset. For a much deeper architecture, accuracy could reach about 99%. To assess the generalizability of the accuracy trend observed in the PPDDL, the proposed solution was also evaluated using the CIFAR-10 dataset with interesting results recorded in this study. This paper is innovative as it combines partial sharing and homomorphic encryption in DDL . Solutions are post-quantum robust and communication cost efficient.

Investigating attention mechanisms for plant disease identification in challenging environments

There is an increasing demand for efficient and precise plant disease detection methods that can quickly identify disease outbreaks. For this, researchers have developed various machine learning and image processing techniques. However, real-field images present challenges due to complex backgrounds, similarities between different disease symptoms, and the need to detect multiple diseases simultaneously. These obstacles hinder the development of a reliable classification model. The attention mechanisms emerge as a critical factor in enhancing the robustness of classification models by selectively focusing on relevant regions or features within infected regions in an image. This paper provides details about various types of attention mechanisms and explores the utilization of these techniques for the machine learning solutions created by researchers for image segmentation, feature extraction, object detection, and classification for efficient plant disease identification. Experiments are conducted on three models: MobileNetV2, EfficientNetV2, and ShuffleNetV2, to assess the effectiveness of attention modules. For this, Squeeze and Excitation layers, the Convolutional Block Attention Module, and transformer modules have been integrated into these models, and their performance has been evaluated using different metrics. The outcomes show that adding attention modules enhances the original models' functionality.

Interpretable machine learning-based text classification method for construction quality defect reports

Efficient identification and remediation of construction defects are critical for ensuring the quality and success of engineering projects. However, the complexity of construction environments poses challenges to this endeavor. Current research predominantly relies on statistical and causal analyses of defect detection reports, yet these methods are time-consuming and error-prone due to the unstructured nature of such reports. To address this, machine learning techniques have been applied to classify defect texts rapidly and accurately. However, existing studies primarily focus on model performance enhancement, neglecting interpretability and the effect of imbalanced data. This study introduces RF-SMOTE, an oversampling technique based on Random Forest (RF), to address the limitations of traditional methods like SMOTE. Comparative analyses demonstrated the efficacy of RF-SMOTE in mitigating imbalanced data effects. Further, the application of SHAP-based interpretability methods in construction management decision-making was explored, filling gaps in existing research. Contributions include providing interpretable machine learning solutions, discussing the effect of imbalanced data, and proposing SHAP-based application scenarios.