A survey on semi-supervised learning

UniMatch V2: Pushing the Limit of Semi-Supervised Semantic Segmentation

Semi-supervised semantic segmentation (SSS) aims at learning rich visual knowledge from cheap unlabeled images to enhance semantic segmentation capability. Among recent works, UniMatch (Yang et al. 2023) improves its precedents tremendously by amplifying the practice of weak-to-strong consistency regularization. Subsequent works typically follow similar pipelines and propose various delicate designs. Despite the achieved progress, strangely, even in this flourishing era of numerous powerful vision models, almost all SSS works are still sticking to 1) using outdated ResNet encoders with small-scale ImageNet-1 K pre-training, and 2) evaluation on simple Pascal and Cityscapes datasets. In this work, we argue that, it is necessary to switch the baseline of SSS from ResNet-based encoders to more capable ViT-based encoders (e.g., DINOv2) that are pre-trained on massive data. A simple update on the encoder (even using 2× fewer parameters) can bring more significant improvement than careful method designs. Built on this competitive baseline, we present our upgraded and simplified UniMatch V2, inheriting the core spirit of weak-to-strong consistency from V1, but requiring less training cost and providing consistently better results. Additionally, witnessing the gradually saturated performance on Pascal and Cityscapes, we appeal that we should focus on more challenging benchmarks with complex taxonomy, such as ADE20K and COCO datasets.

A deep learning strategy to identify cell types across species from high-density extracellular recordings

High-density probes allow electrophysiological recordings from many neurons simultaneously across entire brain circuits but fail to reveal cell type. Here, we develop a strategy to identify cell types from extracellular recordings in awake animals and reveal the computational roles of neurons with distinct functional, molecular, and anatomical properties. We combine optogenetics and pharmacology using the cerebellum as a testbed to generate a curated ground-truth library of electrophysiological properties for Purkinje cells, molecular layer interneurons, Golgi cells, and mossy fibers. We train a semi-supervised deep learning classifier that predicts cell types with greater than 95% accuracy based on the waveform, discharge statistics, and layer of the recorded neuron. The classifier's predictions agree with expert classification on recordings using different probes, in different laboratories, from functionally distinct cerebellar regions, and across species. Our classifier extends the power of modern dynamical systems analyses by revealing the unique contributions of simultaneously recorded cell types during behavior.

Advanced AI and ML frameworks for transforming drug discovery and optimization: With innovative insights in polypharmacology, drug repurposing, combination therapy and nanomedicine

Artificial Intelligence (AI) and Machine Learning (ML) are transforming drug discovery by overcoming traditional challenges like high costs, time-consuming, and frequent failures. AI-driven approaches streamline key phases, including target identification, lead optimization, de novo drug design, and drug repurposing. Frameworks such as deep neural networks (DNNs), convolutional neural networks (CNNs), and deep reinforcement learning (DRL) models have shown promise in identifying drug targets, optimizing delivery systems, and accelerating drug repurposing. Generative adversarial networks (GANs) and variational autoencoders (VAEs) aid de novo drug design by creating novel drug-like compounds with desired properties. Case studies, such as DDR1 kinase inhibitors designed using generative models and CDK20 inhibitors developed via structure-based methods, highlight AI's ability to produce highly specific therapeutics. Models like SNF-CVAE and DeepDR further advance drug repurposing by uncovering new therapeutic applications for existing drugs. Advanced ML algorithms enhance precision in predicting drug efficacy, toxicity, and ADME-Tox properties, reducing development costs and improving drug-target interactions. AI also supports polypharmacology by optimizing multi-target drug interactions and enhances combination therapy through predictions of drug synergies and antagonisms. In nanomedicine, AI models like CURATE.AI and the Hartung algorithm optimize personalized treatments by predicting toxicological risks and real-time dosing adjustments with high accuracy. Despite its potential, challenges like data quality, model interpretability, and ethical concerns must be addressed. High-quality datasets, transparent models, and unbiased algorithms are essential for reliable AI applications. As AI continues to evolve, it is poised to revolutionize drug discovery and personalized medicine, advancing therapeutic development and patient care.

Towards an interpretable deep learning model of cancer

Cancer is a manifestation of dysfunctional cell states. It emerges from an interplay of intrinsic and extrinsic factors that disrupt cellular dynamics, including genetic and epigenetic alterations, as well as the tumor microenvironment. This complexity can make it challenging to infer molecular causes for treating the disease. This may be addressed by system-wide computer models of cells, as they allow rapid generation and testing of hypotheses that would be too slow or impossible to perform in the laboratory and clinic. However, so far, such models have been impeded by both experimental and computational limitations. In this perspective, we argue that they can now be achieved using deep learning algorithms to integrate omics data and prior knowledge of molecular networks. Such models would have many applications in precision oncology, e.g., for identifying drug targets and biomarkers, predicting resistance mechanisms and toxicity effects of drugs, or simulating cell-cell interactions in the microenvironment.

BenthicNet: A global compilation of seafloor images for deep learning applications

Advances in underwater imaging enable collection of extensive seafloor image datasets necessary for monitoring important benthic ecosystems. The ability to collect seafloor imagery has outpaced our capacity to analyze it, hindering mobilization of this crucial environmental information. Machine learning approaches provide opportunities to increase the efficiency with which seafloor imagery is analyzed, yet large and consistent datasets to support development of such approaches are scarce. Here we present BenthicNet: a global compilation of seafloor imagery designed to support the training and evaluation of large-scale image recognition models. An initial set of over 11.4 million images was collected and curated to represent a diversity of seafloor environments using a representative subset of 1.3 million images. These are accompanied by 3.1 million annotations translated to the CATAMI scheme, which span 190,000 of the images. A large deep learning model was trained on this compilation and preliminary results suggest it has utility for automating large and small-scale image analysis tasks. The compilation and model are made openly available for reuse.

Fault Diagnosis of Semi-Supervised Electromechanical Transmission Systems Under Imbalanced Unlabeled Sample Class Information Screening

In the health monitoring of electromechanical transmission systems, the collected state data typically consist of only a minimal amount of labeled data, with a vast majority remaining unlabeled. Consequently, deep learning-based diagnostic models encounter the challenge of scarcity in labeled data and abundance in unlabeled data. Traditional semi-supervised deep learning methods based on pseudo-label self-training, while alleviating the issue of labeled data scarcity to some extent, neglect the reliability of pseudo-label information, the accuracy of feature extraction from unlabeled data, and the imbalance in sample selection. To address these issues, this paper proposes a novel semi-supervised fault diagnosis method under imbalanced unlabeled sample class information screening. Firstly, an information screening mechanism for unlabeled data based on active learning is established. This mechanism discriminates based on the variability of intrinsic feature information in fault samples, accurately screening out unlabeled samples located near decision boundaries that are difficult to separate clearly. Then, combining the maximum membership degree of these unlabeled data in the classification space of the supervised model and interacting with the active learning expert system, label information is assigned to the screened unlabeled data. Secondly, a cost-sensitive function driven by data imbalance is constructed to address the class imbalance problem in unlabeled sample screening, adaptively adjusting the weights of different class samples during model training to guide the training of the supervised model. Ultimately, through dynamic optimization of the supervised model and the feature extraction capability of unlabeled samples, the recognition ability of the diagnostic model for unlabeled samples is significantly enhanced. Validation through two datasets, encompassing a total of 12 experimental scenarios, demonstrates that in scenarios with only a small amount of labeled data, the proposed method achieves a diagnostic accuracy increment exceeding 10% compared to existing typical methods, fully validating the effectiveness and superiority of the proposed method in practical applications.

Harnessing unlabeled data: Enhanced rare earth component content prediction based on BiLSTM-Deep autoencoder

Traditional data-driven models for predicting rare earth component content are primarily developed by relying on supervised learning methods, which suffer from limitations such as a lack of labeled data, lagging, and poor usage of a major amount of unlabeled data. This paper proposes a novel prediction approach based on the BiLSTM-Deep autoencoder enhanced traditional LSSVM algorithm, termed BiLSTM-DeepAE-LSSVM. This approach thoroughly exploits the implicit information contained in copious amounts of unlabeled data in the rare earth production process, thereby improving the traditional supervised prediction method and increasing the accuracy of component content predictions. Initially, a BiLSTM autoencoder is established for unsupervised training on the rare earth production process data, enabling the extraction of inherent time series characteristics. Subsequently, boolean vectors are introduced in the Deep autoencoder training process to perform masking operations on the input data, simulating scenarios with noise and missing data. This is facilitated by their adherence to Bernoulli distributions, which allow for the random setting of certain input vector dimensions to zero. Additionally, the Deep autoencoder is capable of extracting high-dimensional implicit features from the data. After that, the conventional supervised prediction technique, least squares support vector machine (LSSVM), is fused with the implicit characteristics derived from the well-constructed BiLSTM-Deep autoencoder, culminating in the creation of a prediction model for rare earth component content. Ultimately, the simulation verification using LaCe/PrNd extraction field data demonstrates the effectiveness of the proposed approach in harnessing substantial quantities of unlabeled data from the rare earth extraction production process, thereby bolstering the accuracy of model predictions.

CapMatch: Semi-Supervised Contrastive Transformer Capsule With Feature-Based Knowledge Distillation for Human Activity Recognition

This article proposes a semi-supervised contrastive capsule transformer method with feature-based knowledge distillation (KD) that simplifies the existing semisupervised learning (SSL) techniques for wearable human activity recognition (HAR), called CapMatch. CapMatch gracefully hybridizes supervised learning and unsupervised learning to extract rich representations from input data. In unsupervised learning, CapMatch leverages the pseudolabeling, contrastive learning (CL), and feature-based KD techniques to construct similarity learning on lower and higher level semantic information extracted from two augmentation versions of the data, "weak" and "timecut," to recognize the relationships among the obtained features of classes in the unlabeled data. CapMatch combines the outputs of the weak- and timecut-augmented models to form pseudolabeling and thus CL. Meanwhile, CapMatch uses the feature-based KD to transfer knowledge from the intermediate layers of the weak-augmented model to those of the timecut-augmented model. To effectively capture both local and global patterns of HAR data, we design a capsule transformer network consisting of four capsule-based transformer blocks and one routing layer. Experimental results show that compared with a number of state-of-the-art semi-supervised and supervised algorithms, the proposed CapMatch achieves decent performance on three commonly used HAR datasets, namely, HAPT, WISDM, and UCI_HAR. With only 10% of data labeled, CapMatch achieves values of higher than 85.00% on these datasets, outperforming 14 semi-supervised algorithms. When the proportion of labeled data reaches 30%, CapMatch obtains values of no lower than 88.00% on the datasets above, which is better than several classical supervised algorithms, e.g., decision tree and -nearest neighbor (KNN).

Enhanced Domain Adaptation for Foot Ulcer Segmentation Through Mixing Self-Trained Weak Labels

Wound management requires the measurement of the wound parameters such as its shape and area. However, computerized analysis of the wound suffers the challenge of inexact segmentation of the wound images due to limited or inaccurate labels. It is a common scenario that the source domain provides an abundance of labeled data, while the target domain provides only limited labels. To overcome this, we propose a novel approach that combines self-training learning and mixup augmentation. The neural network is trained on the source domain to generate weak labels on the target domain via the self-training process. In the second stage, generated labels are mixed up with labels from the source domain to retrain the neural network and enhance generalization across diverse datasets. The efficacy of our approach was evaluated using the DFUC 2022, FUSeg, and RMIT datasets, demonstrating substantial improvements in segmentation accuracy and robustness across different data distributions. Specifically, in single-domain experiments, segmentation on the DFUC 2022 dataset scored a dice score of 0.711, while the score on the FUSeg dataset achieved 0.859. For domain adaptation, when these datasets were used as target datasets, the dice scores were 0.714 for DFUC 2022 and 0.561 for FUSeg.

Real-Time Driver Drowsiness Detection Using Facial Analysis and Machine Learning Techniques

Drowsy driving poses a significant challenge to road safety worldwide, contributing to thousands of accidents and fatalities annually. Despite advancements in driver drowsiness detection (DDD) systems, many existing methods face limitations such as intrusiveness and delayed reaction times. This research addresses these gaps by leveraging facial analysis and state-of-the-art machine learning techniques to develop a real-time, non-intrusive DDD system. A distinctive aspect of this research is its systematic assessment of various machine and deep learning algorithms across three pivotal public datasets, the NTHUDDD, YawDD, and UTA-RLDD, known for their widespread use in drowsiness detection studies. Our evaluation covered techniques including the K-Nearest Neighbors (KNNs), support vector machines (SVMs), convolutional neural networks (CNNs), and advanced computer vision (CV) models such as YOLOv5, YOLOv8, and Faster R-CNN. Notably, the KNNs classifier reported the highest accuracy of 98.89%, a precision of 99.27%, and an F1 score of 98.86% on the UTA-RLDD. Among the CV methods, YOLOv5 and YOLOv8 demonstrated exceptional performance, achieving 100% precision and recall with mAP@0.5 values of 99.5% on the UTA-RLDD. In contrast, Faster R-CNN showed an accuracy of 81.0% and a precision of 63.4% on the same dataset. These results demonstrate the potential of our system to significantly enhance road safety by providing proactive alerts in real time.

Semisupervised Contrastive Learning for Bioactivity Prediction Using Cell Painting Image Data

Morphological profiling has recently demonstrated remarkable potential for identifying the biological activities of small molecules. Alongside the fully supervised and self-supervised machine learning methods recently proposed for bioactivity prediction from Cell Painting image data, we introduce here a semisupervised contrastive (SemiSupCon) learning approach. This approach combines the strengths of using biological annotations in supervised contrastive learning and leveraging large unannotated image data sets with self-supervised contrastive learning. SemiSupCon enhances downstream prediction performance of classifying MeSH pharmacological classifications from PubChem, as well as mode of action and biological target annotations from the Drug Repurposing Hub across two publicly available Cell Painting data sets. Notably, our approach has effectively predicted the biological activities of several unannotated compounds, and these findings were validated through literature searches. This demonstrates that our approach can potentially expedite the exploration of biological activity based on Cell Painting image data with minimal human intervention.

Artificial intelligence streamlines scientific discovery of drug-target interactions

Drug discovery is a complicated process through which new therapeutics are identified to prevent and treat specific diseases. Identification of drug-target interactions (DTIs) stands as a pivotal aspect within the realm of drug discovery and development. The traditional process of drug discovery, especially identification of DTIs, is marked by its high costs of experimental assays and low success rates. Computational methods have emerged as indispensable tools, especially those employing artificial intelligence (AI) methods, which could streamline the process, thereby reducing costs and time consumption and potentially increasing success rates. In this review, we focus on the application of AI techniques in DTI prediction. Specifically, we commence with a comprehensive overview of drug discovery and development, along with systematic prediction and validation of DTIs. We proceed to highlight the prominent databases and toolkits used in developing AI methods for DTI prediction, as well as with methodologies for evaluating their efficacy. We further extend the exploration into three primary types of state-of-the-art AI methods used in DTI prediction, including classical machine learning, deep learning and network-based methods. Finally, we summarize the key findings and outline the current challenges and future directions that AI methods face in scientific drug discovery and development.

Enhanced Lung Cancer Survival Prediction Using Semi-Supervised Pseudo-Labeling and Learning from Diverse PET/CT Datasets

OBJECTIVE

This study explores a semi-supervised learning (SSL), pseudo-labeled strategy using diverse datasets such as head and neck cancer (HNCa) to enhance lung cancer (LCa) survival outcome predictions, analyzing handcrafted and deep radiomic features (HRF/DRF) from PET/CT scans with hybrid machine learning systems (HMLSs).

METHODS

We collected 199 LCa patients with both PET and CT images, obtained from TCIA and our local database, alongside 408 HNCa PET/CT images from TCIA. We extracted 215 HRFs and 1024 DRFs by PySERA and a 3D autoencoder, respectively, within the ViSERA 1.0.0 software, from segmented primary tumors. The supervised strategy (SL) employed an HMLS-PCA connected with six classifiers on both HRFs and DRFs. The SSL strategy expanded the datasets by adding 408 pseudo-labeled HNCa cases (labeled by the Random Forest algorithm) to 199 LCa cases, using the same HMLS techniques. Furthermore, principal component analysis (PCA) linked with four survival prediction algorithms were utilized in the survival hazard ratio analysis.

RESULTS

The SSL strategy outperformed the SL method (p << 0.001), achieving an average accuracy of 0.85 ± 0.05 with DRFs from PET and PCA + Multi-Layer Perceptron (MLP), compared to 0.69 ± 0.06 for the SL strategy using DRFs from CT and PCA + Light Gradient Boosting (LGB). Additionally, PCA linked with Component-wise Gradient Boosting Survival Analysis on both HRFs and DRFs, as extracted from CT, had an average C-index of 0.80, with a log rank p-value << 0.001, confirmed by external testing.

CONCLUSIONS

Shifting from HRFs and SL to DRFs and SSL strategies, particularly in contexts with limited data points, enabling CT or PET alone, can significantly achieve high predictive performance.

Combining Group Contribution Method and Semisupervised Learning to Build Machine Learning Models for Predicting Hydroxyl Radical Rate Constants of Water Contaminants

Machine learning is an effective tool for predicting reaction rate constants for many organic compounds with the hydroxyl radical (HO•). Previously reported models have achieved relatively good performance, but due to scarce data (<1400 records), the applicability domain (AD) has been significantly limited. To address this limitation, we curated a much larger experimental data set (Primary data set), which contains 2358 kinetic records. We then employed both the group contribution method (GCM) and a semisupervised learning (SSL) strategy to add new data points, aiming to effectively expand the model's AD while improving model performance. The results indicated that GCM improved the model's performance for chemicals outside the AD, while SSL expanded the model's AD. The final model, after incorporating 147,168 new data points, achieved an R2 = 0.77, root-mean-square-error = 0.32, and mean-absolute-error = 0.24 on the test set. Importantly, the AD was expanded by 117% compared to the model developed solely based on the Primary data set, and the final model can be reliably applied to more than 560,000 chemicals from the DSSTox database. Further model interpretation results indicated that the model made predictions based on a correct "understanding" of the impact of key substituents and reactive sites toward HO•. This research provides an effective method for augmenting data sets, which is important in improving ML model performance and expanding AD. The final model has been made widely accessible through a free online predictor.

Prompt-and-Transfer: Dynamic Class-Aware Enhancement for Few-Shot Segmentation

For more efficient generalization to unseen domains (classes), most Few-shot Segmentation (FSS) would directly exploit pre-trained encoders and only fine-tune the decoder, especially in the current era of large models. However, such fixed feature encoders tend to be class-agnostic, inevitably activating objects that are irrelevant to the target class. In contrast, humans can effortlessly focus on specific objects in the line of sight. This paper mimics the visual perception pattern of human beings and proposes a novel and powerful prompt-driven scheme, called "Prompt and Transfer" (PAT), which constructs a dynamic class-aware prompting paradigm to tune the encoder for focusing on the interested object (target class) in the current task. Three key points are elaborated to enhance the prompting: 1) Cross-modal linguistic information is introduced to initialize prompts for each task. 2) Semantic Prompt Transfer (SPT) that precisely transfers the class-specific semantics within the images to prompts. 3) Part Mask Generator (PMG) that works in conjunction with SPT to adaptively generate different but complementary part prompts for different individuals. Surprisingly, PAT achieves competitive performance on 4 different tasks including standard FSS, Cross-domain FSS (e.g., CV, medical, and remote sensing domains), Weak-label FSS, and Zero-shot Segmentation, setting new state-of-the-arts on 11 benchmarks.

Multi-source fully test-time adaptation

Deep neural networks have significantly advanced various fields. However, these models often encounter difficulties in achieving effective generalization when the distribution of test samples varies from that of the training samples. Recently, some fully test-time adaptation methods have been proposed to adapt the trained model with the unlabeled test samples before prediction to enhance the test performance. Despite achieving remarkable results, these methods only involve one trained model, which could only provide certain side information for the test samples. In real-world scenarios, there could be multiple available trained models that are beneficial to the test samples and are complementary to each other. Consequently, to better utilize these trained models, in this paper, we propose the problem of multi-source fully test-time adaptation to adapt multiple trained models to the test samples. To address this problem, we introduce a simple yet effective method utilizing a weighted aggregation scheme and introduce two unsupervised losses. The former could adaptively assign a higher weight to a more relevant model, while the latter could jointly adapt models with online unlabeled samples. Extensive experiments on three image classification datasets show that the proposed method achieves better results than baseline methods, demonstrating the superiority in adapting to multiple models.

SemiHAR: Improving Semisupervised Human Activity Recognition via Multitask Learning

Semisupervised human activity recognition (SemiHAR) has attracted attention in recent years from various domains, such as digital health and ambient intelligence. Currently, it still faces two challenges. For one thing, discriminative features may exist among multiple sequences rather than a single sequence since activities are combinations of motions involving several body parts. For another thing, labeled data and unlabeled data suffer from distribution discrepancies due to the different behavior patterns or biological conditions of users. For that, we propose a novel SemiHAR method based on multitask learning. First, a dimension-based Markov transition field (DMTF) technique is designed to generate 2-D activity data for capturing the interactions among different dimensions. Second, we jointly consider the user recognition (UR) task and the activity recognition (AR) task to reduce the underlying discrepancy. In addition, a task relation learner (TRL) is introduced to dynamically learn task relations, which enables the primary AR task to exploit preferred knowledge from other secondary tasks. We theoretically analyze the proposed SemiHAR and provide a novel generalization result. Extensive experiments conducted on four real-world datasets demonstrate that SemiHAR outperforms other state-of-the-art methods.

A Human-in-the-Loop Method for Annotation of Events in Biomedical Signals

OBJECTIVE

Building large-scale data bases of biomedical signal recordings for training artificial-intelligence systems involves substantial human effort in data processing and annotation. In the case of event detection, experts need to exhaustively scroll through the recordings and highlight events of interest.

METHODS

We propose an iterative annotation support algorithm with a human in the loop to improve the efficiency of the annotation process. Our algorithm generates proposal events based on an event detection model trained on incomplete annotations. The human only needs to verify candidate events proposed by the tool instead of scrolling through the entire data set. Our algorithm iterates between proposal generation and verification to leverage the human-in-the-loop feedback to obtain a growing set of event annotations.

RESULTS

Our algorithm finds a substantial amount of events at a fraction of the human time spent when comparing with a benchmark method and the normal manual process, finding all events in one data set and 70% of events in another with the human-in-the-loop only viewing 20% of the data.

CONCLUSION

Our results show that combining human and computer effort can substantially speed up the annotation process for events in biomedical signal processing.

SIGNIFICANCE

Due to its simplicity and minimal reliance on task-specific information, our algorithm is broadly applicable, unlocking substantial improvements in the scalability and efficiency of biomedical signal annotation.

Melanoma Breslow Thickness Classification Using Ensemble-Based Knowledge Distillation With Semi-Supervised Convolutional Neural Networks

Melanoma is considered a global public health challenge and is responsible for more than 90% deaths related to skin cancer. Although the diagnosis of early melanoma is the main goal of dermoscopy, the discrimination between dermoscopic images of in situ and invasive melanomas can be a difficult task even for experienced dermatologists. Recent advances in artificial intelligence in the field of medical image analysis show that its application to dermoscopy with the aim of supporting and providing a second opinion to the medical expert could be of great interest. In this work, four datasets from different sources were used to train and evaluate deep learning models on in situ versus invasive melanoma classification and on Breslow thickness prediction. Supervised learning and semi-supervised learning using a multi-teacher ensemble knowledge distillation approach were considered and evaluated using a stratified 5-fold cross-validation scheme. The best models achieved AUCs of 0.80850.0242 and of 0.82320.0666 on the former and latter classification tasks, respectively. The best results were obtained using semi-supervised learning, with the best model achieving 0.8547 and 0.8768 AUC, respectively. An external test set was also evaluated, where semi-supervision achieved higher performance in all the classification tasks. The results obtained show that semi-supervised learning could improve the performance of trained models in different melanoma classification tasks compared to supervised learning. Automatic deep learning-based diagnosis systems could support medical professionals in their decision, serving as a second opinion or as a triage tool for medical centers.

Fusion-driven semi-supervised learning-based lung nodules classification with dual-discriminator and dual-generator generative adversarial network

BACKGROUND

The detection and classification of lung nodules are crucial in medical imaging, as they significantly impact patient outcomes related to lung cancer diagnosis and treatment. However, existing models often suffer from mode collapse and poor generalizability, as they fail to capture the complete diversity of the data distribution. This study addresses these challenges by proposing a novel generative adversarial network (GAN) architecture tailored for semi-supervised lung nodule classification.

METHODS

The proposed DDDG-GAN model consists of dual generators and discriminators. Each generator specializes in benign or malignant nodules, generating diverse, high-fidelity synthetic images for each class. This dual-generator setup prevents mode collapse. The dual-discriminator framework enhances the model's generalization capability, ensuring better performance on unseen data. Feature fusion techniques are incorporated to refine the model's discriminatory power between benign and malignant nodules. The model is evaluated in two scenarios: (1) training and testing on the LIDC-IDRI dataset and (2) training on LIDC-IDRI, testing on the unseen LUNA16 dataset and the unseen LUNGx dataset.

RESULTS

In Scenario 1, the DDDG-GAN achieved an accuracy of 92.56%, a precision of 90.12%, a recall of 95.87%, and an F1 score of 92.77%. In Scenario 2, the model demonstrated robust performance with an accuracy of 72.6%, a precision of 72.3%, a recall of 73.82%, and an F1 score of 73.39% when testing using Luna16 and an accuracy of 71.23%, a precision of 67.56%, a recall of 73.52%, and an F1 score of 70.42% when testing using LungX. The results indicate that the proposed model outperforms state-of-the-art semi-supervised learning approaches.

CONCLUSIONS

The DDDG-GAN model mitigates mode collapse and improves generalizability in lung nodule classification. It demonstrates superior performance on both the LIDC-IDRI and the unseen LUNA16 and LungX datasets, offering significant potential for improving diagnostic accuracy in clinical practice.

Prior knowledge-guided vision-transformer-based unsupervised domain adaptation for intubation prediction in lung disease at one week

Data-driven approaches have achieved great success in various medical image analysis tasks. However, fully-supervised data-driven approaches require unprecedentedly large amounts of labeled data and often suffer from poor generalization to unseen new data due to domain shifts. Various unsupervised domain adaptation (UDA) methods have been actively explored to solve these problems. Anatomical and spatial priors in medical imaging are common and have been incorporated into data-driven approaches to ease the need for labeled data as well as to achieve better generalization and interpretation. Inspired by the effectiveness of recent transformer-based methods in medical image analysis, the adaptability of transformer-based models has been investigated. How to incorporate prior knowledge for transformer-based UDA models remains under-explored. In this paper, we introduce a prior knowledge-guided and transformer-based unsupervised domain adaptation (PUDA) pipeline. It regularizes the vision transformer attention heads using anatomical and spatial prior information that is shared by both the source and target domain, which provides additional insight into the similarity between the underlying data distribution across domains. Besides the global alignment of class tokens, it assigns local weights to guide the token distribution alignment via adversarial training. We evaluate our proposed method on a clinical outcome prediction task, where Computed Tomography (CT) and Chest X-ray (CXR) data are collected and used to predict the intubation status of patients in a week. Abnormal lesions are regarded as anatomical and spatial prior information for this task and are annotated in the source domain scans. Extensive experiments show the effectiveness of the proposed PUDA method.

Addressing data limitations in leakage detection of water distribution systems: Data creation, data requirement reduction, and knowledge transfer

Leakage in water distribution systems is a significant problem worldwide, leading to wastage of water resources, compromised water quality and excess energy consumption. Leakage detection is essential to reduce the duration of leaks and data-driven methods are increasingly being used for this purpose. However, these models are data hungry and available observed data, especially leakage data, is limited in most cases. In addition, these data need to be manually processed to label whether leaks occur, which is time-consuming and costly. These are significant obstacles for the development and application of these methods. This article provides a comprehensive review of relevant journal papers, categorizing all data-driven methods into unsupervised anomaly detection, semi-supervised anomaly detection and supervised classification methods based on how the data are utilized for developing these methods. In addition, strategies to address data limitations are summarized from both data and model perspectives, including data creation, reduction of a model's data requirements and knowledge transfer. After detailing these strategies, research gaps are identified. Based on these, future research directions are suggested, highlighting the need for further research in data augmentation, development of semi-supervised classification methods, exploration of multi-classification methods with model updating mechanisms, and development of novel knowledge transfer methods.

Semi-supervised Federated Learning for Digital Twin 6G-enabled IIoT: A Bayesian estimated approach

INTRODUCTION

In recent years, the proliferation of Industrial Internet of Things (IIoT) devices has resulted in a substantial increase in data generation across various domains, including the nascent 6G networks. Digital Twins (DTs), serving as virtual replicas of physical entities, have gained popularity within the realm of IoT due to their capacity to simulate and optimize physical systems in a cost-effective manner. Nonetheless, the security of DTs and the safeguarding of the sensitive data they generate have emerged as paramount concerns. Fortunately, the Federated Fearning (FL) system has emerged as a promising solution to address the challenge of data privacy within DTs. Nonetheless, the requisite acquisition of a significant volume of labeled data for training purposes poses a formidable challenge, particularly in a DT environment that blends real and virtual data.

OBJECTIVES

To tackle this challenge, this study presents an innovative Semi-supervised FL (SSFL) framework designed to overcome the scarcity of labeled data through the strategic utilization of pseudo-labels.

METHODS

Specifically, our proposed SSFL algorithm, named SSFL-MBE, introduces a novel approach by combining Mix data augmentation and Bayesian Estimation consistency regularization loss, thereby integrating robust augmentation techniques to enhance model generalization. Furthermore, we introduce a Bayesian-estimated pseudo-label loss that leverages prior probabilistic knowledge to enhance model performance. Our investigation focuses particularly on a demanding scenario where labeled and unlabeled data are segregated across disparate locations, specifically, the server and various clients.

RESULTS

Comprehensive evaluations conducted on CIFAR-10 and MNIST datasets conclusively demonstrate that our proposed algorithm consistently surpasses mainstream SSFL baseline models, exhibiting an enhancement in model performance ranging from 0.5% to 1.5%.

CONCLUSION

Overall, this work contributes to the development of more efficient and secure approaches for model training in DT-empowered FL settings, which is crucial for the deployment of IIoTs in 6G-enabled environments.

A simulation data-driven semi-supervised framework based on MK-KNN graph and ESSGAT for bearing fault diagnosis

Current supervised intelligent fault diagnosis relies on abundant labeled data. However, collecting and labeling data are typically both expensive and time-consuming. Fault diagnosis with unlabeled data remains a significant challenge. To address this issue, a simulation data-driven semi-supervised framework based on multi-kernel K-nearest neighbor (MK-KNN) and edge self-supervised graph attention network (ESSGAT) is proposed. The novel MK-KNN establishes the neighborhood relationships between simulation data and real data. The developed multi-kernel function mitigates the risks of overfitting and underfitting, thereby enhancing the robustness of the simulation-real graphs. The designed ESSGAT employs two forms of self-supervised attention to predict the presence of edges, increasing the weights of crucial neighboring nodes in the MK-KNN graph. The performance of the proposed method is evaluated using a public bearing dataset and a self-constructed dataset of high-speed train axle box bearings. The results show that the proposed method achieves better diagnostic performance compared with other state-of-the-art graph construction methods and graph convolutional networks.

Privacy-preserving multi-source semi-supervised domain adaptation for seizure prediction

Domain adaptation (DA) has been frequently used to solve the inter-patient variability problem in EEG-based seizure prediction. However, existing DA methods require access to the existing patients' data when adapting the model, which leads to privacy concerns. Besides, most of them treat the whole existing patients' data as one single source and attempt to minimize the discrepancy with the target patient. This manner ignores the inter-patient variability among source patients, making the adaptation more difficult. Considering theses issues simultaneously, we present a novel multi-source-free semi-supervised domain adaptive seizure prediction model (MSF-SSDA-SPM). MSF-SSDA-SPM considers each source patient as one single source and generates a pretrained model from each source. Without requiring access to the source data, MSF-SSDA-SPM performs adaptation just using these pretrained source models and limited labeled target data. Specifically, we freeze the classifiers of all the source models and optimize the source feature extractors in a joint manner. Then we design a knowledge distillation strategy to integrate the knowledge of these well-adapted source models into one single target model. On the CHB-MIT dataset, MSF-SSDA-SPM attains a sensitivity of 88.6%, a FPR of 0.182/h and an AUC of 0.856; on the Kaggle dataset, it achieves 78.6%, 0.178/h and 0.784, respectively. Experimental results demonstrate that MSF-SSDA-SPM achieves both high privacy-protection and promising prediction performance.

On Leveraging Machine Learning in Sport Science in the Hypothetico-deductive Framework

Supervised machine learning (ML) offers an exciting suite of algorithms that could benefit research in sport science. In principle, supervised ML approaches were designed for pure prediction, as opposed to explanation, leading to a rise in powerful, but opaque, algorithms. Recently, two subdomains of ML-explainable ML, which allows us to "peek into the black box," and interpretable ML, which encourages using algorithms that are inherently interpretable-have grown in popularity. The increased transparency of these powerful ML algorithms may provide considerable support for the hypothetico-deductive framework, in which hypotheses are generated from prior beliefs and theory, and are assessed against data collected specifically to test that hypothesis. However, this paper shows why ML algorithms are fundamentally different from statistical methods, even when using explainable or interpretable approaches. Translating potential insights from supervised ML algorithms, while in many cases seemingly straightforward, can have unanticipated challenges. While supervised ML cannot be used to replace statistical methods, we propose ways in which the sport sciences community can take advantage of supervised ML in the hypothetico-deductive framework. In this manuscript we argue that supervised machine learning can and should augment our exploratory investigations in sport science, but that leveraging potential insights from supervised ML algorithms should be undertaken with caution. We justify our position through a careful examination of supervised machine learning, and provide a useful analogy to help elucidate our findings. Three case studies are provided to demonstrate how supervised machine learning can be integrated into exploratory analysis. Supervised machine learning should be integrated into the scientific workflow with requisite caution. The approaches described in this paper provide ways to safely leverage the strengths of machine learning-like the flexibility ML algorithms can provide for fitting complex patterns-while avoiding potential pitfalls-at best, like wasted effort and money, and at worst, like misguided clinical recommendations-that may arise when trying to integrate findings from ML algorithms into domain knowledge. KEY POINTS: Some supervised machine learning algorithms and statistical models are used to solve the same problem, y = f(x) + ε, but differ fundamentally in motivation and approach. The hypothetico-deductive framework-in which hypotheses are generated from prior beliefs and theory, and are assessed against data collected specifically to test that hypothesis-is one of the core frameworks comprising the scientific method. In the hypothetico-deductive framework, supervised machine learning can be used in an exploratory capacity. However, it cannot replace the use of statistical methods, even as explainable and interpretable machine learning methods become increasingly popular. Improper use of supervised machine learning in the hypothetico-deductive framework is tantamount to p-value hacking in statistical methods.

Application of artificial intelligence tools in wastewater and waste gas treatment systems: Recent advances and prospects

The non-linear complex relationships among the process variables in wastewater and waste gas treatment systems possess a significant challenge for real-time systems modelling. Data driven artificial intelligence (AI) tools are increasingly being adopted to predict the process performance, cost-effective process monitoring, and the control of different waste treatment systems, including those involving resource recovery. This review presents an in-depth analysis of the applications of emerging AI tools in physico-chemical and biological processes for the treatment of air pollutants, water and wastewater, and resource recovery processes. Additionally, the successful implementation of AI-controlled wastewater and waste gas treatment systems, along with real-time monitoring at the industrial scale are discussed.

Demystifying unsupervised learning: how it helps and hurts

Humans and machines rarely have access to explicit external feedback or supervision, yet manage to learn. Most modern machine learning systems succeed because they benefit from unsupervised data. Humans are also expected to benefit and yet, mysteriously, empirical results are mixed. Does unsupervised learning help humans or not? Here, we argue that the mixed results are not conflicting answers to this question, but reflect that humans self-reinforce their predictions in the absence of supervision, which can help or hurt depending on whether predictions and task align. We use this framework to synthesize empirical results across various domains to clarify when unsupervised learning will help or hurt. This provides new insights into the fundamentals of learning with implications for instruction and lifelong learning.

Advances in artificial intelligence for meibomian gland evaluation: A comprehensive review

Meibomian gland dysfunction (MGD) is increasingly recognized as a critical contributor to evaporative dry eye, significantly impacting visual quality. With a global prevalence estimated at 35.8 %, it presents substantial challenges for clinicians. Conventional manual evaluation techniques for MGD face limitations characterized by inefficiencies, high subjectivity, limited big data processing capabilities, and a dearth of quantitative analytical tools. With rapidly advancing artificial intelligence (AI) techniques revolutionizing ophthalmology, studies are now leveraging sophisticated AI methodologies--including computer vision, unsupervised learning, and supervised learning--to facilitate comprehensive analyses of meibomian gland (MG) evaluations. These evaluations employ various techniques, including slit lamp examination, infrared imaging, confocal microscopy, and optical coherence tomography. This paradigm shift promises enhanced accuracy and consistency in disease evaluation and severity classification. While AI has achieved preliminary strides in meibomian gland evaluation, ongoing advancements in system development and clinical validation are imperative. We review the evolution of MG evaluation, juxtapose AI-driven methods with traditional approaches, elucidate the specific roles of diverse AI technologies, and explore their practical applications using various evaluation techniques. Moreover, we delve into critical considerations for the clinical deployment of AI technologies and envisages future prospects, providing novel insights into MG evaluation and fostering technological and clinical progress in this arena.

Machine learning algorithms and biomarkers identification for pancreatic cancer diagnosis using multi-omics data integration

PURPOSE

Pancreatic cancer is a lethal type of cancer with most of the cases being diagnosed in an advanced stage and poor prognosis. Developing new diagnostic and prognostic markers for pancreatic cancer can significantly improve early detection and patient outcomes. These biomarkers can potentially revolutionize medical practice by enabling personalized, more effective, targeted treatments, ultimately improving patient outcomes.

METHODS

The search strategy was developed following PRISMA guidelines. A comprehensive search was performed across four electronic databases: PubMed, Scopus, EMBASE, and Web of Science, covering all English publications up to September 2022. The Newcastle-Ottawa Scale (NOS) was utilized to assess bias, categorizing studies as "good," "fair," or "poor" quality based on their NOS scores. Descriptive statistics for all included studies were compiled and reviewed, along with the NOS scores for each study to indicate their quality assessment.

RESULTS

Our results showed that SVM and RF are the most widely used algorithms in machine learning and data analysis, particularly for biomarker identification. SVM, a supervised learning algorithm, is employed for both classification and regression by mapping data points in high-dimensional space to identify the optimal separating hyperplane between classes.

CONCLUSIONS

The application of machine-learning algorithms in the search for novel biomarkers in pancreatic cancer represents a significant advancement in the field. By harnessing the power of artificial intelligence, researchers are poised to make strides towards earlier detection and more effective treatment, ultimately improving patient outcomes in this challenging disease.

Exploring Inherent Consistency for Semi-Supervised Anatomical Structure Segmentation in Medical Imaging

Due to the exorbitant expense of obtaining labeled data in the field of medical image analysis, semi-supervised learning has emerged as a favorable method for the segmentation of anatomical structures. Although semi-supervised learning techniques have shown great potential in this field, existing methods only utilize image-level spatial consistency to impose unsupervised regularization on data in label space. Considering that anatomical structures often possess inherent anatomical properties that have not been focused on in previous works, this study introduces the inherent consistency into semi-supervised anatomical structure segmentation. First, the prediction and the ground-truth are projected into an embedding space to obtain latent representations that encapsulate the inherent anatomical properties of the structures. Then, two inherent consistency constraints are designed to leverage these inherent properties by aligning these latent representations. The proposed method is plug-and-play and can be seamlessly integrated with existing methods, thereby collaborating to improve segmentation performance and enhance the anatomical plausibility of the results. To evaluate the effectiveness of the proposed method, experiments are conducted on three public datasets (ACDC, LA, and Pancreas). Extensive experimental results demonstrate that the proposed method exhibits good generalizability and outperforms several state-of-the-art methods.

From zero to hero: Harnessing transformers for biomedical named entity recognition in zero- and few-shot contexts

Supervised named entity recognition (NER) in the biomedical domain depends on large sets of annotated texts with the given named entities. The creation of such datasets can be time-consuming and expensive, while extraction of new entities requires additional annotation tasks and retraining the model. This paper proposes a method for zero- and few-shot NER in the biomedical domain to address these challenges. The method is based on transforming the task of multi-class token classification into binary token classification and pre-training on a large number of datasets and biomedical entities, which allows the model to learn semantic relations between the given and potentially novel named entity labels. We have achieved average F1 scores of 35.44% for zero-shot NER, 50.10% for one-shot NER, 69.94% for 10-shot NER, and 79.51% for 100-shot NER on 9 diverse evaluated biomedical entities with fine-tuned PubMedBERT-based model. The results demonstrate the effectiveness of the proposed method for recognizing new biomedical entities with no or limited number of examples, outperforming previous transformer-based methods, and being comparable to GPT3-based models using models with over 1000 times fewer parameters. We make models and developed code publicly available.

Taming Overconfident Prediction on Unlabeled Data From Hindsight

Minimizing prediction uncertainty on unlabeled data is a key factor to achieve good performance in semi-supervised learning (SSL). The prediction uncertainty is typically expressed as the entropy computed by the transformed probabilities in output space. Most existing works distill low-entropy prediction by either accepting the determining class (with the largest probability) as the true label or suppressing subtle predictions (with the smaller probabilities). Unarguably, these distillation strategies are usually heuristic and less informative for model training. From this discernment, this article proposes a dual mechanism, named adaptive sharpening (ADS), which first applies a soft-threshold to adaptively mask out determinate and negligible predictions, and then seamlessly sharpens the informed predictions, distilling certain predictions with the informed ones only. More importantly, we theoretically analyze the traits of ADS by comparing it with various distillation strategies. Numerous experiments verify that ADS significantly improves state-of-the-art SSL methods by making it a plug-in. Our proposed ADS forges a cornerstone for future distillation-based SSL research.

Machine learning and artificial intelligence within pediatric autoimmune diseases: applications, challenges, future perspective

INTRODUCTION

Autoimmune disorders affect 4.5% to 9.4% of children, significantly reducing their quality of life. The diagnosis and prognosis of autoimmune diseases are uncertain because of the variety of onset and development. Machine learning can identify clinically relevant patterns from vast amounts of data. Hence, its introduction has been beneficial in the diagnosis and management of patients.

AREAS COVERED

This narrative review was conducted through searching various electronic databases, including PubMed, Scopus, and Web of Science. This study thoroughly explores the current knowledge and identifies the remaining gaps in the applications of machine learning specifically in the context of pediatric autoimmune and related diseases.

EXPERT OPINION

Machine learning algorithms have the potential to completely change how pediatric autoimmune disorders are identified, treated, and managed. Machine learning can assist physicians in making more precise and fast judgments, identifying new biomarkers and therapeutic targets, and personalizing treatment strategies for each patient by utilizing massive datasets and powerful analytics.

SMS Scam Detection Application Based on Optical Character Recognition for Image Data Using Unsupervised and Deep Semi-Supervised Learning

The growing problem of unsolicited text messages (smishing) and data irregularities necessitates stronger spam detection solutions. This paper explores the development of a sophisticated model designed to identify smishing messages by understanding the complex relationships among words, images, and context-specific factors, areas that remain underexplored in existing research. To address this, we merge a UCI spam dataset of regular text messages with real-world spam data, leveraging OCR technology for comprehensive analysis. The study employs a combination of traditional machine learning models, including K-means, Non-Negative Matrix Factorization, and Gaussian Mixture Models, along with feature extraction techniques such as TF-IDF and PCA. Additionally, deep learning models like RNN-Flatten, LSTM, and Bi-LSTM are utilized. The selection of these models is driven by their complementary strengths in capturing both the linear and non-linear relationships inherent in smishing messages. Machine learning models are chosen for their efficiency in handling structured text data, while deep learning models are selected for their superior ability to capture sequential dependencies and contextual nuances. The performance of these models is rigorously evaluated using metrics like accuracy, precision, recall, and F1 score, enabling a comparative analysis between the machine learning and deep learning approaches. Notably, the K-means feature extraction with vectorizer achieved 91.01% accuracy, and the KNN-Flatten model reached 94.13% accuracy, emerging as the top performer. The rationale behind highlighting these models is their potential to significantly improve smishing detection rates. For instance, the high accuracy of the KNN-Flatten model suggests its applicability in real-time spam detection systems, but its computational complexity might limit scalability in large-scale deployments. Similarly, while K-means with vectorizer excels in accuracy, it may struggle with the dynamic and evolving nature of smishing attacks, necessitating continual retraining.

Turbocharging protein binding site prediction with geometric attention, inter-resolution transfer learning, and homology-based augmentation

BACKGROUND

Locating small molecule binding sites in target proteins, in the resolution of either pocket or residue, is critical in many drug-discovery scenarios. Since it is not always easy to find such binding sites using conventional methods, different deep learning methods to predict binding sites out of protein structures have been developed in recent years. The existing deep learning based methods have several limitations, including (1) the inefficiency of the CNN-only architecture, (2) loss of information due to excessive post-processing, and (3) the under-utilization of available data sources.

METHODS

We present a new model architecture and training method that resolves the aforementioned problems. First, by layering geometric self-attention units on top of residue-level 3D CNN outputs, our model overcomes the problems of CNN-only architectures. Second, by configuring the fundamental units of computation as residues and pockets instead of voxels, our method reduced the information loss from post-processing. Lastly, by employing inter-resolution transfer learning and homology-based augmentation, our method maximizes the utilization of available data sources to a significant extent.

RESULTS

The proposed method significantly outperformed all state-of-the-art baselines regarding both resolutions-pocket and residue. An ablation study demonstrated the indispensability of our proposed architecture, as well as transfer learning and homology-based augmentation, for achieving optimal performance. We further scrutinized our model's performance through a case study involving human serum albumin, which demonstrated our model's superior capability in identifying multiple binding sites of the protein, outperforming the existing methods.

CONCLUSIONS

We believe that our contribution to the literature is twofold. Firstly, we introduce a novel computational method for binding site prediction with practical applications, substantiated by its strong performance across diverse benchmarks and case studies. Secondly, the innovative aspects in our method- specifically, the design of the model architecture, inter-resolution transfer learning, and homology-based augmentation-would serve as useful components for future work.

The changing scenario of drug discovery using AI to deep learning: Recent advancement, success stories, collaborations, and challenges

Due to the transformation of artificial intelligence (AI) tools and technologies, AI-driven drug discovery has come to the forefront. It reduces the time and expenditure. Due to these advantages, pharmaceutical industries are concentrating on AI-driven drug discovery. Several drug molecules have been discovered using AI-based techniques and tools, and several newly AI-discovered drug molecules have already entered clinical trials. In this review, we first present the data and their resources in the pharmaceutical sector for AI-driven drug discovery and illustrated some significant algorithms or techniques used for AI and ML which are used in this field. We gave an overview of the deep neural network (NN) models and compared them with artificial NNs. Then, we illustrate the recent advancement of the landscape of drug discovery using AI to deep learning, such as the identification of drug targets, prediction of their structure, estimation of drug-target interaction, estimation of drug-target binding affinity, design of de novo drug, prediction of drug toxicity, estimation of absorption, distribution, metabolism, excretion, toxicity; and estimation of drug-drug interaction. Moreover, we highlighted the success stories of AI-driven drug discovery and discussed several collaboration and the challenges in this area. The discussions in the article will enrich the pharmaceutical industry.

A Futuristic Development in 3D Printing Technique Using Nanomaterials with a Step Toward 4D Printing

3D bioprinting has shown great promise in tissue engineering and regenerative medicine for creating patient-specific tissue scaffolds and medicinal devices. The quickness, accurate imaging, and design targeting of this emerging technology have excited biomedical engineers and translational medicine researchers. Recently, scaffolds made from 3D bioprinted tissue have become more clinically effective due to nanomaterials and nanotechnology. Because of quantum confinement effects and high surface area/volume ratios, nanomaterials and nanotechnological techniques have unique physical, chemical, and biological features. The use of nanomaterials and 3D bioprinting has led to scaffolds with improved physicochemical and biological properties. Nanotechnology and nanomaterials affect 3D bioprinted tissue engineered scaffolds for regenerative medicine and tissue engineering. Biomaterials and cells that respond to stimuli change the structural shape in 4D bioprinting. With such dynamic designs, tissue architecture can change morphologically. New 4D bioprinting techniques will aid in bioactuation, biorobotics, and biosensing. The potential of 4D bioprinting in biomedical technologies is also discussed in this article.

Machine learning assisted sorting of active microswimmers

Active matter systems, being in a non-equilibrium state, exhibit complex behaviors, such as self-organization, giving rise to emergent phenomena. There are many examples of active particles with biological origins, including bacteria and spermatozoa, or with artificial origins, such as self-propelled swimmers and Janus particles. The ability to manipulate active particles is vital for their effective application, e.g., separating motile spermatozoa from nonmotile and dead ones, to increase fertilization chance. In this study, we proposed a mechanism-an apparatus-to sort and demix active particles based on their motility values (Péclet number). Initially, using Brownian simulations, we demonstrated the feasibility of sorting self-propelled particles. Following this, we employed machine learning methods, supplemented with data from comprehensive simulations that we conducted for this study, to model the complex behavior of active particles. This enabled us to sort them based on their Péclet number. Finally, we evaluated the performance of the developed models and showed their effectiveness in demixing and sorting the active particles. Our findings can find applications in various fields, including physics, biology, and biomedical science, where the sorting and manipulation of active particles play a pivotal role.

A Semi-Supervised Transformer-Based Deep Learning Framework for Automated Tooth Segmentation and Identification on Panoramic Radiographs

Automated tooth segmentation and identification on dental radiographs are crucial steps in establishing digital dental workflows. While deep learning networks have been developed for these tasks, their performance has been inferior in partially edentulous individuals. This study proposes a novel semi-supervised Transformer-based framework (SemiTNet), specifically designed to improve tooth segmentation and identification performance on panoramic radiographs, particularly in partially edentulous cases, and establish an open-source dataset to serve as a unified benchmark. A total of 16,317 panoramic radiographs (1589 labeled and 14,728 unlabeled images) were collected from various datasets to create a large-scale dataset (TSI15k). The labeled images were divided into training and test sets at a 7:1 ratio, while the unlabeled images were used for semi-supervised learning. The SemiTNet was developed using a semi-supervised learning method with a label-guided teacher-student knowledge distillation strategy, incorporating a Transformer-based architecture. The performance of SemiTNet was evaluated on the test set using the intersection over union (IoU), Dice coefficient, precision, recall, and F1 score, and compared with five state-of-the-art networks. Paired t-tests were performed to compare the evaluation metrics between SemiTNet and the other networks. SemiTNet outperformed other networks, achieving the highest accuracy for tooth segmentation and identification, while requiring minimal model size. SemiTNet's performance was near-perfect for fully dentate individuals (all metrics over 99.69%) and excellent for partially edentulous individuals (all metrics over 93%). In edentulous cases, SemiTNet obtained statistically significantly higher tooth identification performance than all other networks. The proposed SemiTNet outperformed previous high-complexity, state-of-the-art networks, particularly in partially edentulous cases. The established open-source TSI15k dataset could serve as a unified benchmark for future studies.

Retinal Vessel Segmentation by a Transformer-U-Net Hybrid Model With Dual-Path Decoder

This paper introduces an effective and efficient framework for retinal vessel segmentation. First, we design a Transformer-CNN hybrid model in which a Transformer module is inserted inside the U-Net to capture long-range interactions. Second, we design a dual-path decoder in the U-Net framework, which contains two decoding paths for multi-task outputs. Specifically, we train the extra decoder to predict vessel skeletons as an auxiliary task which helps the model learn balanced features. The proposed framework, named as TSNet, not only achieves good performances in a fully supervised learning manner but also enables a rough skeleton annotation process. The annotators only need to roughly delineate vessel skeletons instead of giving precise pixel-wise vessel annotations. To learn with rough skeleton annotations plus a few precise vessel annotations, we propose a skeleton semi-supervised learning scheme. We adopt a mean teacher model to produce pseudo vessel annotations and conduct annotation correction for roughly labeled skeletons annotations. This learning scheme can achieve promising performance with fewer annotation efforts. We have evaluated TSNet through extensive experiments on five benchmarking datasets. Experimental results show that TSNet yields state-of-the-art performances on retinal vessel segmentation and provides an efficient training scheme in practice.

Few-Shot Object Detection: A Comprehensive Survey

Humans are able to learn to recognize new objects even from a few examples. In contrast, training deep-learning-based object detectors requires huge amounts of annotated data. To avoid the need to acquire and annotate these huge amounts of data, few-shot object detection (FSOD) aims to learn from few object instances of new categories in the target domain. In this survey, we provide an overview of the state of the art in FSOD. We categorize approaches according to their training scheme and architectural layout. For each type of approach, we describe the general realization as well as concepts to improve the performance on novel categories. Whenever appropriate, we give short takeaways regarding these concepts in order to highlight the best ideas. Eventually, we introduce commonly used datasets and their evaluation protocols and analyze the reported benchmark results. As a result, we emphasize common challenges in evaluation and identify the most promising current trends in this emerging field of FSOD.

Using clinical text to refine unspecific condition codes in Dutch general practitioner EHR data

OBJECTIVE

Observational studies using electronic health record (EHR) databases often face challenges due to unspecific clinical codes that can obscure detailed medical information, hindering precise data analysis. In this study, we aimed to assess the feasibility of refining these unspecific condition codes into more specific codes in a Dutch general practitioner (GP) EHR database by leveraging the available clinical free text.

METHODS

We utilized three approaches for text classification-search queries, semi-supervised learning, and supervised learning-to improve the specificity of ten unspecific International Classification of Primary Care (ICPC-1) codes. Two text representations and three machine learning algorithms were evaluated for the (semi-)supervised models. Additionally, we measured the improvement achieved by the refinement process on all code occurrences in the database.

RESULTS

The classification models performed well for most codes. In general, no single classification approach consistently outperformed the others. However, there were variations in the relative performance of the classification approaches within each code and in the use of different text representations and machine learning algorithms. Class imbalance and limited training data affected the performance of the (semi-)supervised models, yet the simple search queries remained particularly effective. Ultimately, the developed models improved the specificity of over half of all the unspecific code occurrences in the database.

CONCLUSIONS

Our findings show the feasibility of using information from clinical text to improve the specificity of unspecific condition codes in observational healthcare databases, even with a limited range of machine-learning techniques and modest annotated training sets. Future work could investigate transfer learning, integration of structured data, alternative semi-supervised methods, and validation of models across healthcare settings. The improved level of detail enriches the interpretation of medical information and can benefit observational research and patient care.

Performance evaluation of semi-supervised learning frameworks for multi-class weed detection

Precision weed management (PWM), driven by machine vision and deep learning (DL) advancements, not only enhances agricultural product quality and optimizes crop yield but also provides a sustainable alternative to herbicide use. However, existing DL-based algorithms on weed detection are mainly developed based on supervised learning approaches, typically demanding large-scale datasets with manual-labeled annotations, which can be time-consuming and labor-intensive. As such, label-efficient learning methods, especially semi-supervised learning, have gained increased attention in the broader domain of computer vision and have demonstrated promising performance. These methods aim to utilize a small number of labeled data samples along with a great number of unlabeled samples to develop high-performing models comparable to the supervised learning counterpart trained on a large amount of labeled data samples. In this study, we assess the effectiveness of a semi-supervised learning framework for multi-class weed detection, employing two well-known object detection frameworks, namely FCOS (Fully Convolutional One-Stage Object Detection) and Faster-RCNN (Faster Region-based Convolutional Networks). Specifically, we evaluate a generalized student-teacher framework with an improved pseudo-label generation module to produce reliable pseudo-labels for the unlabeled data. To enhance generalization, an ensemble student network is employed to facilitate the training process. Experimental results show that the proposed approach is able to achieve approximately 76% and 96% detection accuracy as the supervised methods with only 10% of labeled data in CottonWeedDet3 and CottonWeedDet12, respectively. We offer access to the source code (https://github.com/JiajiaLi04/SemiWeeds), contributing a valuable resource for ongoing semi-supervised learning research in weed detection and beyond.

Investigating Contrastive Pair Learning's Frontiers in Supervised, Semisupervised, and Self-Supervised Learning

In recent years, contrastive learning has been a highly favored method for self-supervised representation learning, which significantly improves the unsupervised training of deep image models. Self-supervised learning is a subset of unsupervised learning in which the learning process is supervised by creating pseudolabels from the data themselves. Using supervised final adjustments after unsupervised pretraining is one way to take the most valuable information from a vast collection of unlabeled data and teach from a small number of labeled instances. This study aims firstly to compare contrastive learning with other traditional learning models; secondly to demonstrate by experimental studies the superiority of contrastive learning during classification; thirdly to fine-tune performance using pretrained models and appropriate hyperparameter selection; and finally to address the challenge of using contrastive learning techniques to produce data representations with semantic meaning that are independent of irrelevant factors like position, lighting, and background. Relying on contrastive techniques, the model efficiently captures meaningful representations by discerning similarities and differences between modified copies of the same image. The proposed strategy, involving unsupervised pretraining followed by supervised fine-tuning, improves the robustness, accuracy, and knowledge extraction of deep image models. The results show that even with a modest 5% of data labeled, the semisupervised model achieves an accuracy of 57.72%. However, the use of supervised learning with a contrastive approach and careful hyperparameter tuning increases accuracy to 85.43%. Further adjustment of the hyperparameters resulted in an excellent accuracy of 88.70%.

Progress and Opportunities for Machine Learning in Materials and Processes of Additive Manufacturing

In recent years, there has been widespread adoption of machine learning (ML) technologies to unravel intricate relationships among diverse parameters in various additive manufacturing (AM) techniques. These ML models excel at recognizing complex patterns from extensive, well-curated datasets, thereby unveiling latent knowledge crucial for informed decision-making during the AM process. The collaborative synergy between ML and AM holds the potential to revolutionize the design and production of AM-printed parts. This review delves into the challenges and opportunities emerging at the intersection of these two dynamic fields. It provides a comprehensive analysis of the publication landscape for ML-related research in the field of AM, explores common ML applications in AM research (such as quality control, process optimization, design optimization, microstructure analysis, and material formulation), and concludes by presenting an outlook that underscores the utilization of advanced ML models, the development of emerging sensors, and ML applications in emerging AM-related fields. Notably, ML has garnered increased attention in AM due to its superior performance across various AM-related applications. It is envisioned that the integration of ML into AM processes will significantly enhance 3D printing capabilities across diverse AM-related research areas.

CarcassFormer: an end-to-end transformer-based framework for simultaneous localization, segmentation and classification of poultry carcass defect

In the food industry, assessing the quality of poultry carcasses during processing is a crucial step. This study proposes an effective approach for automating the assessment of carcass quality without requiring skilled labor or inspector involvement. The proposed system is based on machine learning (ML) and computer vision (CV) techniques, enabling automated defect detection and carcass quality assessment. To this end, an end-to-end framework called CarcassFormer is introduced. It is built upon a Transformer-based architecture designed to effectively extract visual representations while simultaneously detecting, segmenting, and classifying poultry carcass defects. Our proposed framework is capable of analyzing imperfections resulting from production and transport welfare issues, as well as processing plant stunner, scalder, picker, and other equipment malfunctions. To benchmark the framework, a dataset of 7,321 images was initially acquired, which contained both single and multiple carcasses per image. In this study, the performance of the CarcassFormer system is compared with other state-of-the-art (SOTA) approaches for both classification, detection, and segmentation tasks. Through extensive quantitative experiments, our framework consistently outperforms existing methods, demonstrating re- markable improvements across various evaluation metrics such as AP, AP@50, and AP@75. Furthermore, the qualitative results highlight the strengths of CarcassFormer in capturing fine details, including feathers, and accurately localizing and segmenting carcasses with high precision. To facilitate further research and collaboration, the source code and trained models will be made publicly available upon acceptance.

AI-Powered Microfluidics: Shaping the Future of Phenotypic Drug Discovery

Phenotypic drug discovery (PDD), which involves harnessing biological systems directly to uncover effective drugs, has undergone a resurgence in recent years. The rapid advancement of artificial intelligence (AI) over the past few years presents numerous opportunities for augmenting phenotypic drug screening on microfluidic platforms, leveraging its predictive capabilities, data analysis, efficient data processing, etc. Microfluidics coupled with AI is poised to revolutionize the landscape of phenotypic drug discovery. By integrating advanced microfluidic platforms with AI algorithms, researchers can rapidly screen large libraries of compounds, identify novel drug candidates, and elucidate complex biological pathways with unprecedented speed and efficiency. This review provides an overview of recent advances and challenges in AI-based microfluidics and their applications in drug discovery. We discuss the synergistic combination of microfluidic systems for high-throughput screening and AI-driven analysis for phenotype characterization, drug-target interactions, and predictive modeling. In addition, we highlight the potential of AI-powered microfluidics to achieve an automated drug screening system. Overall, AI-powered microfluidics represents a promising approach to shaping the future of phenotypic drug discovery by enabling rapid, cost-effective, and accurate identification of therapeutically relevant compounds.

Multi-omics based artificial intelligence for cancer research

With significant advancements of next generation sequencing technologies, large amounts of multi-omics data, including genomics, epigenomics, transcriptomics, proteomics, and metabolomics, have been accumulated, offering an unprecedented opportunity to explore the heterogeneity and complexity of cancer across various molecular levels and scales. One of the promising aspects of multi-omics lies in its capacity to offer a holistic view of the biological networks and pathways underpinning cancer, facilitating a deeper understanding of its development, progression, and response to treatment. However, the exponential growth of data generated by multi-omics studies present significant analytical challenges. Processing, analyzing, integrating, and interpreting these multi-omics datasets to extract meaningful insights is an ambitious task that stands at the forefront of current cancer research. The application of artificial intelligence (AI) has emerged as a powerful solution to these challenges, demonstrating exceptional capabilities in deciphering complex patterns and extracting valuable information from large-scale, intricate omics datasets. This review delves into the synergy of AI and multi-omics, highlighting its revolutionary impact on oncology. We dissect how this confluence is reshaping the landscape of cancer research and clinical practice, particularly in the realms of early detection, diagnosis, prognosis, treatment and pathology. Additionally, we elaborate the latest AI methods for multi-omics integration to provide a comprehensive insight of the complex biological mechanisms and inherent heterogeneity of cancer. Finally, we discuss the current challenges of data harmonization, algorithm interpretability, and ethical considerations. Addressing these challenges necessitates a multidisciplinary collaboration, paving the promising way for more precise, personalized, and effective treatments for cancer patients.