Learning-based Applications Research Articles

Specifics of Data Collection and Data Processing during Formation of RailVista Dataset for Machine Learning- and Deep Learning-Based Applications.

This paper presents the methodology and outcomes of creating the Rail Vista dataset, designed for detecting defects on railway tracks using machine and deep learning techniques. The dataset comprises 200,000 high-resolution images categorized into 19 distinct classes covering various railway infrastructure defects. The data collection involved a meticulous process including complex image capture methods, distortion techniques for data enrichment, and secure storage in a data warehouse using efficient binary file formats. This structured dataset facilitates effective training of machine/deep learning models, enhancing automated defect detection systems in railway safety and maintenance applications. The study underscores the critical role of high-quality datasets in advancing machine learning applications within the railway domain, highlighting future prospects for improving safety and reliability through automated recognition technologies.

DeepOnto: A Python package for ontology engineering with deep learning

Integrating deep learning techniques, particularly language models (LMs), with knowledge representation techniques like ontologies has raised widespread attention, urging the need of a platform that supports both paradigms. Although packages such as OWL API and Jena offer robust support for basic ontology processing features, they lack the capability to transform various types of information within ontologies into formats suitable for downstream deep learning-based applications. Moreover, widely-used ontology APIs are primarily Java-based while deep learning frameworks like PyTorch and Tensorflow are mainly for Python programming. To address the needs, we present DeepOnto, a Python package designed for ontology engineering with deep learning. The package encompasses a core ontology processing module founded on the widely-recognised and reliable OWL API, encapsulating its fundamental features in a more “Pythonic” manner and extending its capabilities to incorporate other essential components including reasoning, verbalisation, normalisation, taxonomy, projection, and more. Building on this module, DeepOnto offers a suite of tools, resources, and algorithms that support various ontology engineering tasks, such as ontology alignment and completion, by harnessing deep learning methods, primarily pre-trained LMs. In this paper, we also demonstrate the practical utility of DeepOnto through two use-cases: the Digital Health Coaching in Samsung Research UK and the Bio-ML track of the Ontology Alignment Evaluation Initiative (OAEI).

Safety and efficiency of a fully automatic workflow for auto-segmentation in radiotherapy using three commercially available deep learning-based applications

Advancements in radiotherapy auto-segmentation necessitate reliable and efficient workflows. Therefore, a standardized fully automatic workflow was developed for three commercially available deep learning-based auto-segmentation applications and compared to a manual workflow for safety and efficiency. The workflow underwent safety evaluation with failure mode and effects analysis. Notably, eight failure modes were reduced, including seven with severity factors ≥7, indicating the effect on patients, and two with Risk Priority Number value >125, which assesses relative risk level. Efficiency, measured by mouse clicks, showed zero clicks with the automatic workflow. This automation illustrated improvement in both safety and efficiency of workflow

KNN-GNN: A powerful graph neural network enhanced by aggregating K-nearest neighbors in common subspace

It has been proven that graph neural networks (GNNs) are effective for a variety of graph learning-based applications. Typical GNNs iteratively aggregate messages from immediate neighbors under the homophily assumption. However, there are a larger number of heterophilous networks in our daily life, where the ability of these GNNs is limited. Recently, some GNN models have been proposed to handle networks with heterophily via some key designs like aggregating higher-order neighbors and combining immediate representations. But ”noise” information transmitted from different-order neighbors will be injected into the representations of nodes. In this paper, we propose a new GNN model, called KNN-GNN, to effectively perform the node classification task for networks with various homophily levels. The main idea of KNN-GNN is to learn a comprehensive and accurate representation for each node by integrating not only the local information from its neighborhood but also the non-local information held by its similar nodes decentralized in the network. Specifically, the local information of a node is generated from itself and its 1-hop neighbor. Then, we project all nodes into a common subspace, where similar nodes are desired to be close to each other. The non-local information of a node is gathered by aggregating its K-Nearest Neighbors searched in the common subspace. We evaluate the performance of KNN-GNN on both real and synthetic datasets including networks with diverse homophily levels. The results demonstrate that KNN-GNN outstrips the state-of-the-art baselines. Moreover, the ablation experiments show that the core designs in KNN-GNN play a critical role in node representation learning.

End-to-End Dataset Collection System for Sport Activities

Datasets are key to developing new machine learning-based applications but are very costly to prepare, which hinders research and development in the field. We propose an edge-to-cloud end-to-end system architecture optimized for sport activity recognition dataset collection and application deployment. Tests in authentic contexts of use in four different sports have revealed the system’s ability to effectively collect machine learning-usable data, with an energy consumption compatible with the timeframe of most of the sport types. The proposed architecture relies on a key feature of the Measurify internet of things framework for the management of measurement data (i.e., .csv dataset management) and supports a workflow designed for efficient data labeling of signal timeseries. The architecture is independent of any specific sport, and a new dataset generation application can be set up in a few days, even by novice developers. With a view to concretely supporting the R&D community, our work is released open-source.

Studying logging practice in machine learning-based applications

Context:Logging is a common practice in traditional software development. There have been multiple studies on the characteristics of logging in traditional software systems such as C/C++, Java, and Android applications. However, logging practices in Machine Learning-based (ML-based) applications are still not well understood. The size and complexity of data and models used in ML-based applications present unique challenges for logging. Objective:In this paper, we aim to bridge this knowledge gap and provide insight into the logging practices in ML-based applications, making the first attempt to characterize current logging practices within a large number of open-source ML-based applications. Method:We conducted an empirical study on 502 open-source ML applications to understand their logging practices, combining quantitative and qualitative analyses and a survey involving 31 practitioners. Results:Our quantitative analysis reveals that logging in ML applications is less common than in traditional software, with info and warn log levels being popular. Top ML-specific logging libraries include MLflow, Tensorboard, Neptune, and W&B. Qualitatively, logging is used for data and model management, especially in model training. Our survey reinforces the importance of logging in experiment tracking, complementing our qualitative findings. Conclusion:Our research carries significant implications. It reveals distinctive ML logging practices compared to traditional software. We have highlighted the prevalence of general-purpose logging libraries in ML code, indicating a potential gap in awareness regarding ML-specific logging tools. This insight benefits researchers and developers aiming to enhance ML project reproducibility and sets the stage for exploring ML-specific logging tools’ impact on machine learning system quality and trustworthiness.

Recent advancements in machine learning for bone marrow cell morphology analysis.

As machine learning progresses, techniques such as neural networks, decision trees, and support vector machines are being increasingly applied in the medical domain, especially for tasks involving large datasets, such as cell detection, recognition, classification, and visualization. Within the domain of bone marrow cell morphology analysis, deep learning offers substantial benefits due to its robustness, ability for automatic feature learning, and strong image characterization capabilities. Deep neural networks are a machine learning paradigm specifically tailored for image processing applications. Artificial intelligence serves as a potent tool in supporting the diagnostic process of clinical bone marrow cell morphology. Despite the potential of artificial intelligence to augment clinical diagnostics in this domain, manual analysis of bone marrow cell morphology remains the gold standard and an indispensable tool for identifying, diagnosing, and assessing the efficacy of hematologic disorders. However, the traditional manual approach is not without limitations and shortcomings, necessitating, the exploration of automated solutions for examining and analyzing bone marrow cytomorphology. This review provides a multidimensional account of six bone marrow cell morphology processes: automated bone marrow cell morphology detection, automated bone marrow cell morphology segmentation, automated bone marrow cell morphology identification, automated bone marrow cell morphology classification, automated bone marrow cell morphology enumeration, and automated bone marrow cell morphology diagnosis. Highlighting the attractiveness and potential of machine learning systems based on bone marrow cell morphology, the review synthesizes current research and recent advances in the application of machine learning in this field. The objective of this review is to offer recommendations to hematologists for selecting the most suitable machine learning algorithms to automate bone marrow cell morphology examinations, enabling swift and precise analysis of bone marrow cytopathic trends for early disease identification and diagnosis. Furthermore, the review endeavors to delineate potential future research avenues for machine learning-based applications in bone marrow cell morphology analysis.

COMPARISON OF VISION TRANSFORMERS AND CONVOLUTIONAL NEURAL NETWORKS FOR SKIN DISEASE CLASSIFICATION

Skin diseases are one of the most common diseases in humans. Due to its many various symptoms and types, computer vision studies have been frequently applied to its diagnosis and classification. Previous studies have frequently used machine learning methods and deep learning-based Convolutional Neural Networks (CNN) for skin disease diagnosis. Although deep learning-based applications have achieved great success in terms of detection accuracy, research continues to ensure the desired performance. However, Vision Transformer (ViT), recently proposed as a competitive alternative to CNNs, is gaining increasing popularity. This paper compares ResNet18 and ResNet50 networks, important CNN models, with ViT for classifying skin disease. The comparison is applied on a dataset containing a small number of samples. In the application performed on a dataset containing skin disease images, ViT provides 68.93% classification accuracy, while ResNet18 and ResNet50 classification accuracy is 61.65% and 61.17%, respectively. Other metrics calculated along with the accuracies also prove the superiority of ViT over ResNet models. However, ViT has a big disadvantage in terms of training time.

Rethinking Certification for Trustworthy Machine-Learning-Based Applications

Rethinking Certification for Trustworthy Machine-Learning-Based Applications

Interface-Based Search and Automatic Reassembly of CAD Models for Database Expansion and Model Reuse

This paper introduces a new framework for reassembling CAD models of mechanical parts. The generated CAD assemblies are well-constrained, with collision-free parts, and they are ready-to-use for downstream applications. First, input dead CAD models candidate for the reassembly are sorted following a part-by-part interface-based identification algorithm that is capable of characterizing each part according to the assembly slots and interfaces it offers. The slots are then hashed, and the resulting keys are used for fast search of parts in the database. Thus, the parts that can be assembled are quickly identified, and their assembly can be considered according to various scenarios. To support the reassembly steps and satisfy the constraints associated with the specified interfaces, a collision-free kinematic constraint solver is also proposed. During the reassembly steps, the geometry of the slots can also be automatically modified to adjust their dimensions, in order to ensure a perfect fit and to avoid interference at the level of the interfaces. The resulting database can also be further expanded while modifying the relative positions/orientations of the assembled parts. The approach is illustrated on several test cases covering different exploitation scenarios, ranging from simple model reuse to database expansion for machine learning-based applications. Such an automatic reassembly process is particularly innovative, and it clearly paves the way for smart assembly procedures.

Perceptions and concerns of emergency medicine practitioners about artificial intelligence in emergency triage management during the pandemic: a national survey-based study.

There have been continuous discussions over the ethics of using AI in healthcare. We sought to identify the ethical issues and viewpoints of Turkish emergency care doctors about the use of AI during epidemic triage. Ten emergency specialists were initially enlisted for this project, and their responses to open-ended questions about the ethical issues surrounding AI in the emergency room provided valuable information. A 15-question survey was created based on their input and was refined through a pilot test with 15 emergency specialty doctors. Following that, the updated survey was sent to emergency specialists via email, social media, and private email distribution. 167 emergency medicine specialists participated in the study, with an average age of 38.22 years and 6.79 years of professional experience. The majority agreed that AI could benefit patients (54.50%) and healthcare professionals (70.06%) in emergency department triage during pandemics. Regarding responsibility, 63.47% believed in shared responsibility between emergency medicine specialists and AI manufacturers/programmers for complications. Additionally, 79.04% of participants agreed that the responsibility for complications in AI applications varies depending on the nature of the complication. Concerns about privacy were expressed by 20.36% regarding deep learning-based applications, while 61.68% believed that anonymity protected privacy. Additionally, 70.66% of participants believed that AI systems would be as sensitive as humans in terms of non-discrimination. The potential advantages of deploying AI programs in emergency department triage during pandemics for patients and healthcare providers were acknowledged by emergency medicine doctors in Turkey. Nevertheless, they expressed notable ethical concerns related to the responsibility and accountability aspects of utilizing AI systems in this context.

Analysis of machine learning-based applications for intelligent construction of shield tunnels

With the quick advancement of the level of information science in shield tunnel construction, the monitoring methods of shield equipment during tunnel boring work are increasingly improved and the recorded construction data includes not only information on the internal workings of the shield equipment, but also on its interaction with the external strata. Machine learning data analysis is powerful and has a wider range of applications and scope than traditional data analysis methods in the civil construction industry. Through the use of machine learning methods, the data and information collected can be mined and analysed in depth to find the intrinsic connections and linkages that can help improve the safety and efficiency ragarding shield tunnel construction. This work presents a literature analysis of current situations of machine learning for shield tunnel construction at home and abroad, briefly describes the basic principles of machine learning methods, summarises and analyses the research situation in shield tunnel construction, reviews the progress of machine learning-based shield equipment condition analysis, intelligent prediction and control methods for shield tunneling parameters and shield tunneling surface deformation prediction, and summarises the current research The study also summarises the shortcomings of current research. Finally, an outlook on the development of shield tunneling towards intelligence is presented.

An Interactive Image Segmentation Method Based on Multi-Level Semantic Fusion.

Understanding and analyzing 2D/3D sensor data is crucial for a wide range of machine learning-based applications, including object detection, scene segmentation, and salient object detection. In this context, interactive object segmentation is a vital task in image editing and medical diagnosis, involving the accurate separation of the target object from its background based on user annotation information. However, existing interactive object segmentation methods struggle to effectively leverage such information to guide object-segmentation models. To address these challenges, this paper proposes an interactive image-segmentation technique for static images based on multi-level semantic fusion. Our method utilizes user-guidance information both inside and outside the target object to segment it from the static image, making it applicable to both 2D and 3D sensor data. The proposed method introduces a cross-stage feature aggregation module, enabling the effective propagation of multi-scale features from previous stages to the current stage. This mechanism prevents the loss of semantic information caused by multiple upsampling and downsampling of the network, allowing the current stage to make better use of semantic information from the previous stage. Additionally, we incorporate a feature channel attention mechanism to address the issue of rough network segmentation edges. This mechanism captures richer feature details from the feature channel level, leading to finer segmentation edges. In the experimental evaluation conducted on the PASCAL Visual Object Classes (VOC) 2012 dataset, our proposed interactive image segmentation method based on multi-level semantic fusion demonstrates an intersection over union (IOU) accuracy approximately 2.1% higher than the currently popular interactive image segmentation method in static images. The comparative analysis highlights the improved performance and effectiveness of our method. Furthermore, our method exhibits potential applications in various fields, including medical imaging and robotics. Its compatibility with other machine learning methods for visual semantic analysis allows for integration into existing workflows. These aspects emphasize the significance of our contributions in advancing interactive image-segmentation techniques and their practical utility in real-world applications.

Agile Services Provisioning for Learning-Based Applications in Fog Computing Networks

Fog computing has emerged as a promising solution for service provisioning. To support the increasing number of smart end devices and their requirement for intelligent data processing, numerous learning-based applications will be deployed at the edge network to provide services with an acceptable level of latency. However, such applications assume that the training and the actual data are independent and identically distributed, which is impractical in the dynamic and heterogeneous environment of the real world. This makes service matchmaking a challenge because even if the service functionality matches user requirements, the mismatched actual data will inevitably result in the degradation of service quality. This paper targets the mismatching issue of learning-based applications and the high requirement of intelligent data processing at the network edge and proposes a novel service provisioning model. This model introduces a novel service description model to resolve data mismatch, a clustering algorithm that pre-processes requests to deal with high concurrency requirements, and a heuristic joint service searching method to reduce traffic costs. This work was evaluated through a case study and simulations. Evaluation results show that the proposed model can achieve a good quality of services and reduce response time as well as traffic costs.

Resource allocation and device pairing for energy-efficient NOMA-enabled federated edge learning

In the era of Fifth Generation of cellular networks (5G), Internet of Things (IoT) devices are becoming much more popular and accumulate a lot of data. With the advantages of privacy protection and resource saving, Federated Edge Learning (FEEL) is a promising machine learning paradigm which helps improve the service quality of deep learning-based applications and makes people’s lives intelligent by processing these data. In FEEL, users only train the models locally without uploading their own data, thus alleviating the data security issue. Nevertheless, this paradigm also faces a major challenge. The training tasks are carried out on the mobile devices, which poses great pressure to their limited battery lives. In 5G, Non-Orthogonal Multiple Access (NOMA) has become an enabling technology to support massive connections because of its higher spectral efficiency and throughput over OMA technologies. Considering the potential of FEEL and the broad application prospects of NOMA, we design a NOMA-enabled FEEL framework and minimize its Energy Consumption (EC) by optimizing the device pairing, communication resource (i.e., transmission powers and time slots) and computation resource (i.e., CPU frequencies and local training threshold). The original problem can be decomposed into two subproblems: resource allocation and device pairing subproblems. For the resource allocation subproblem, with the help of the linear search method, we can find a near-optimal solution. Then, the optimality of the proposed solution (i.e., how far is the near-optimal from the optimum) is proved by sensitivity analysis. For the device pairing subproblem, we propose two low-complexity pairing schemes, which have their own characteristics. Simulation results demonstrate the effectiveness of the proposed resource allocation and device pairing strategies and highlight the advantage of the NOMA-enabled FEEL framework over the existing TDMA-enabled framework in terms of EC.

A Literature Review on Students' Views on the Utilization of Zoom for English Language Instruction During the Pandemic

This research aims to investigate the implementation and effectiveness of using the Zoom application for English Language Teaching (ELT) during the pandemic. The study focuses on students who utilized the Zoom application as a method for ELT. The research methodology combines both quantitative and qualitative approaches, utilizing an online questionnaire to collect data. The researcher analyzed the data, integrated their ideas, and formulated concluding remarks based on the collected information. The review includes three primary studies published between 2020 and 2021. The first study, conducted by Irwandi (2020), examines the implementation of the electric method for ELT through distance learning during the COVID-19 pandemic. The second study, also conducted by Irwandi (2020), investigates students' perceptions of using Zoom and WhatsApp for ELT during the pandemic. The third study, conducted by Suadi (2021), explores the effectiveness of learning-based applications, including Zoom, for ELT in the pandemic era. The review findings indicate that students were able to actively practice English, responded positively to the use of Zoom, and demonstrated the ability to learn independently.

Pharmaceutical patent landscaping: A novel approach to understand patents from the drug discovery perspective

Patents play a crucial role in the drug discovery process by providing legal protection for discoveries and incentivising investments in research and development. By identifying patterns within patent data resources, researchers can gain insight into the market trends and priorities of the pharmaceutical and biotechnology industries, as well as provide additional perspectives on more fundamental aspects such as the emergence of potential new drug targets. In this paper, we used the patent enrichment tool, PEMT, to extract, integrate, and analyse patent literature for rare diseases (RD) and Alzheimer's disease (AD). This is followed by a systematic review of the underlying patent landscape to decipher trends and applications in patents for these diseases. To do so, we discuss prominent organisations involved in drug discovery research in AD and RD. This allows us to gain an understanding of the importance of AD and RD from specific organisational (pharmaceutical or university) perspectives. Next, we analyse the historical focus of patents in relation to individual therapeutic targets and correlate them with market scenarios allowing the identification of prominent targets for a disease. Lastly, we identified drug repurposing activities within the two diseases with the help of patents. This resulted in identifying existing repurposed drugs and novel potential therapeutic approaches applicable to the indication areas. The study demonstrates the expanded applicability of patent documents from legal to drug discovery, design, and research, thus, providing a valuable resource for future drug discovery efforts. Moreover, this study is an attempt towards understanding the importance of data underlying patent documents and raising the need for preparing the data for machine learning-based applications.

Comparing Image Captioning Techniques using Deep Learning Models

Websites today generate tremendous amounts of data and this data needs to be processed effectively by the creators. One such important factor is processing images on the website and generating effective data from it through techniques like Web Scraping. This process can be done through techniques like Image Captioning. Image captioning is a powerful process that involves generating descriptive image captions. The ability to generate detailed and accurate descriptions of images is extremely valuable in many different fields, particularly in machine learning-based applications and systems. Images often contain a wealth of information that can be difficult for machines to understand without the help of advanced algorithms. Image captioning is one way to extract and use this information for a variety of purposes. Image captioning can also be used in self-driving cars to help vehicles navigate the environment more effectively. It can also be used to develop software for blind people, providing them with descriptions of the images that they cannot see. To accomplish image captioning, we rely on models based on deep learning, specifically advancements in natural language processing. One approach to creating these models involves using pre-trained CNNs and recurrent neural networks (RNNs) like ResNet and LSTM. These pre-trained models can be combined with natural language processing techniques to generate captions for images. As deep learning continues to evolve, there will likely be even more advanced techniques and models that can be used to enhance the accuracy and efficiency of image captioning. Overall, image captioning is a valuable tool that can help us better understand and utilize the vast amount of information contained in images.

Performance analysis of edge, fog and cloud computing paradigms for real-time video quality assessment and phishing detection

PurposeThe purpose of this paper is to perform a comparative analysis of the delay associated in running two real-time machine learning-based applications, namely, a video quality assessment (VQA) and a phishing detection application by using the edge, fog and cloud computing paradigms.Design/methodology/approachThe VQA algorithm was developed using Android Studio and run on a mobile phone for the edge paradigm. For the fog paradigm, it was hosted on a Java server and for the cloud paradigm on the IBM and Firebase clouds. The phishing detection algorithm was embedded into a browser extension for the edge paradigm. For the fog paradigm, it was hosted on a Node.js server and for the cloud paradigm on Firebase.FindingsFor the VQA algorithm, the edge paradigm had the highest response time while the cloud paradigm had the lowest, as the algorithm was computationally intensive. For the phishing detection algorithm, the edge paradigm had the lowest response time, and the cloud paradigm had the highest, as the algorithm had a low computational complexity. Since the determining factor for the response time was the latency, the edge paradigm provided the smallest delay as all processing were local.Research limitations/implicationsThe main limitation of this work is that the experiments were performed on a small scale due to time and budget constraints.Originality/valueA detailed analysis with real applications has been provided to show how the complexity of an application can determine the best computing paradigm on which it can be deployed.

Privacy-Preserving Federated Graph Neural Network Learning on Non-IID Graph Data

Since the concept of federated learning (FL) was proposed by Google in 2017, many applications have been combined with FL technology due to its outstanding performance in data integration, computing performance, privacy protection, etc. However, most traditional federated learning-based applications focus on image processing and natural language processing with few achievements in graph neural networks due to the graph’s nonindependent identically distributed (IID) nature. Representation learning on graph-structured data generates graph embedding, which helps machines understand graphs effectively. Meanwhile, privacy protection plays a more meaningful role in analyzing graph-structured data such as social networks. Hence, this paper proposes PPFL-GNN, a novel privacy-preserving federated graph neural network framework for node representation learning, which is a pioneer work for graph neural network-based federated learning. In PPFL-GNN, clients utilize a local graph dataset to generate graph embeddings and integrate information from other collaborative clients to utilize federated learning to produce more accurate representation results. More importantly, by integrating embedding alignment techniques in PPFL-GNN, we overcome the obstacles of federated learning on non-IID graph data and can further reduce privacy exposure by sharing preferred information.