A Unified Active Learning Framework for Annotating Graph Data with Application to Software Source Code Performance Prediction

Limited experimental data remains a key challenge in applying machine learning to drug discovery, particularly for cancer-related targets. In this study, we present a data-efficient active meta-deep learning framework to predict mitogen-activated protein kinase 1 (MAPK1) inhibitors, which are promising candidates for cancer-related therapies. Our approach integrates active learning (AL) with a meta-model that combines four deep architectures: a convolutional neural network, an attention, a graph convolutional network, and a graph neural network-attention, trained on molecular descriptors and graph-based representations. These models generate four probability-based features that feed into an attention-based meta-learner, improving predictive performance by 5.12% in the area under the precision-recall curve (AUPRC) and 5.48% in the Matthews correlation coefficient (MCC) using only 10% of the training data. Among the AL sampling strategies evaluated, entropy sampling showed competitive performance in selecting informative molecules for model improvement. Overall, our framework achieves an AUPRC of 0.835 ± 0.017 and MCC of 0.817 ± 0.017, on par with a traditional training method despite using only 26.7% of the training data. Compared to a conventional random forest model trained on brute-force, a 100% full training set, our approach shows a 10.6% improvement in AUPRC and modest gains in MCC, confirming the effectiveness of the proposed framework. Under severe class imbalance, balanced accuracy steadily increased across AL iterations, reaching values greater than 0.85 at the final iteration for all uncertainty-driven strategies. Molecular docking confirmed successful prioritization of the top four predicted compounds. Evaluation on an external MAPK1 data set demonstrated generalizability, with our approach achieving an AUPRC of 0.818 and an MCC of 0.403, comparable to the independent test set. These results highlight the potential of combining intelligent data selection with deep learning architectures through the meta-model to accelerate predictive performance in data-scarce drug discovery. Scientific contribution: This study contributes a novel, data-efficient active meta-deep learning framework for predicting MAPK1 inhibitors, addressing the challenge of limited experimental data in a cancer-specific target. By integrating AL with a meta-model composed of four deep architectures, the approach significantly enhances the predictive performance using only a fraction of the training data. The framework achieves superior metrics compared to traditional training methods, highlighting its potential to accelerate drug discovery in data-scarce settings.

Normally, software documentations are produced, informally. They are written in the unnatural and non-structural form of the language, such as user manuals, user requirements, design documentation, tutorials, support documentation, and so on. Recent studies show that 61% of software projects are subject to failure or challenges due to an increase in the costs and production time. Various factors may lead to this issue, and one of the major contributing factors is the lack of links between the software's source code and its related documents. The significance of software development and the possibility of making prospective changes by the development team necessitate an understanding of the links between various sections of codes and documentations. Therefore, it is crucial to design a system to link the software codes to their corresponding textual documentation. This article proposes a model for recovering the latent, but traceable links between software source codes and existing documents based on word extraction and function name separation. The contributions in this article include: (1) a model based on word extraction from document and source codes; (2) the proposal of an algorithm for splitting compound words and words that are connected to one another and completing abbreviations used in the names of functions, variables, and output commands; and (3) a new algorithm that is proposed for retrieving traceable latent links between the source code and documents. Two data sets are used in this research and the achieved results will be reported in terms of recall, precision, and F-measure. The experimental results are promising and indicate that the proposed approach significantly outperforms its counterparts.

Automatic analysis methods of electrocardiograms (ECGs) usually required large-scale annotated training data, but the annotation process is extremely time-consuming. While semi-supervised learning can leverage unlabeled data, its performance depends heavily on the quality of the initial labeled subset. Active learning has been used to identify the most informative samples for annotation, but conventional approaches face three critical limitations: (1) dependency on manual intervention for iterative query design, (2) prohibitive computational costs during sample selection, and (3) limited compatibility with semi-supervised learning frameworks. To address these limitations, we proposed an Unsupervised Active Feature-selective Semi-Supervised Learning (UAFSSL) framework for ECG analysis, including an unsupervised feature selection-based active learning module and a semi-supervised learning module. UAFSSL captures latent data distributions via unsupervised feature extraction, selects diverse and representative samples using pseudo-label clustering, and integrates seamlessly with semi-supervised learning to eliminate human intervention. We validated our algorithm on an ECG waveform segmentation task and an atrial fibrillation detection task. In the waveform segmentation task, our method improved the F1-score for P-wave delineation by 2.4% compared to random sampling, using only 5% of labeled samples. For the atrial fibrillation detection task, we evaluated our method on both the AFDB and a 24-hour dataset collected from 500 atrial fibrillation patients. Using only 200 labeled samples for model training, our method achieved AUC improvements of 2.5% and 2.2% over random sampling in five-fold cross validation. This is the first study to integrate unsupervised active learning with semi-supervised learning for automatic ECG analysis, offering a robust, automated solution to reduce annotation costs while enhancing clinical applicability.

Active learning has recently attracted increasing attention in the task of person re-identification, due to its unique scalability that not only maximally reduces the annotation cost but also retains the satisfying performance. Although some preliminary active learning methods have been explored in scalable person re-identification task, they have the following two problems: 1) the inefficiency in the selection process of image pairs due to the huge search space, and 2) the ineffectiveness caused by ignoring the impact of unlabeled data in model training. Considering that, we propose a Multi-grained Active Semi-Supervised learning framework, named MASS, to address the scalable person re-identification problem existing in the practical scenarios. Specifically, we firstly design a cluster-scatter procedure to alleviate the inefficiency problem, which consists of two components: cluster step and scatter step. The cluster step shrinks the search space into individual small clusters by a coarse-grained clustering method, and the subsequent scatter step further mines the hard distinguished image pairs from unlabelled set to purify the learned clusters by a novel centrality-based adaptive purification strategy. Afterward, we introduce a customized purification loss for the purified clustering, which utilizes the complementary information in both labeled and unlabeled data to optimize the model for solving the ineffectiveness problem. The cluster-scatter procedure and the model optimization are performed in an iterative fashion to achieve the promising performance while greatly reducing the annotation cost. Extensive experimental results have demonstrated that MASS can even achieve a competitive performance with fully supervised methods in the case of extremely less annotation requirements.

Software ownership allows the owner to restrict the distribution of software and to prevent others from reading the software's source code and building upon it. However, free software is released to users under software licenses that give them the right to read the source code, modify it, reuse it, and distribute the software to others. Proponents of free software such as Richard M. Stallman and Eben Moglen argue that the social disutility of software ownership is a sufficient justification for prohibiting it. This social disutility includes the social instability of disregarding laws and agreements covering software use and distribution, inequality of software access, and the inability to help others by sharing software with them. Here I consider these and other social disutility claims against withholding specific software rights from users, in particular, the rights to read the source code, duplicate, distribute, modify, imitate, and reuse portions of the software within new programs. I find that generally while withholding these rights from software users does cause some degree of social disutility, only the rights to duplicate, modify and imitate cannot legitimately be denied to users on this basis. The social disutility of withholding the rights to distribute the software, read its source code and reuse portions of it in new programs is insufficient to prohibit software owners from denying them to users. A compromise between the software owner and user can minimise the social disutility of withholding these particular rights from users. However, the social disutility caused by software patents is sufficient for rejecting such patents as they restrict the methods of reducing social disutility possible with other forms of software ownership.

Dual-metal-site catalysts (DMSCs) are increasingly important catalysts in the field of electrochemical carbon dioxide reduction reaction (CO2RR) in recent years. However, rapid screening of suitable metal combinations of DMSCs remains a huge challenge. Herein, we constructed an active learning (AL) framework to study CO2RR to HCOOH. This AL framework turned out a success in the accurate prediction of 282 DMSCs for CO2RR through interactive learning between users and machine learning (ML) models. Among the 42 DMSCs calculated in three iteration loops of AL, 29 DMSCs were obtained, where the screening success rate was as high as 70%. Furthermore, we found five experimentally unexplored DMSCs that exhibited better CO2RR activity and selectivity than pure Bi. Low prediction errors on other DMSCs show that the AL model possessed outstanding universality. The results prove the excellent potential of the AL method and provide guidance on the design of high-performance electrocatalysts for CO2RR.

Active learning is an important technology to solve the lack of data in crack detection model training. However, the sampling strategies of most existing active learning methods for crack detection are based on the uncertainty or representation of the samples, which cannot effectively balance the exploitation and exploration of active learning. To solve this problem, this study proposes an active learning method for crack detection based on subset searching and weighted sampling. First, a new active learning framework is established to successively search subsets with large uncertainty from the candidate dataset, and select training samples with large diversity from the subsets to update the crack detection model. Second, to realize the active learning process, a subset searching method based on sample relative error is proposed to adaptively select subsets with large uncertainty, and a weighted sampling method based on flow-based deep generative network is introduced to select training samples with large diversity form the subsets. Third, a termination criterion for active learning directly based on the prediction accuracy of the trained model is proposed to adaptively determine the maximum number of iterations. Finally, the proposed method is tested using two open-source crack datasets. The experimental comparison results on the Bridge Crack Library dataset show that the proposed method has higher calculation efficiency and prediction accuracy in crack detection than the uncertainty-based and representation-based active learning methods. The test results on the DeepCrack dataset show that the crack detection model trained by the proposed method has good transferability on different datasets with multi-scale concrete cracks and scenes.

We present NeMO-Net, the first open-source fully convolutional neural network (FCNN) and interactive learning and training software aimed at assessing the present and past dynamics of shallow marine systems through habitat mapping into geomorphological (9 classes) and biological classes (22 classes). Shallow marine systems, particularly coral reefs, are under significant pressures due to climate change, ocean acidification, and other anthropogenic pressures, leading to rapid, often devastating changes, in these fragile and diverse ecosystems. Historically, remote sensing of shallow marine habitats has been limited to meter-scale imagery due to the optical effects of ocean wave distortion, refraction, and optical attenuation. NeMO-Net combines 3D cm-scale distortion-free imagery captured using NASA's airborne FluidCam and fluid lensing remote sensing technology with low resolution airborne and spaceborne datasets of varying spatial resolutions, spectral spaces, calibrations, and temporal cadence in a supercomputer-based deep learning framework. NeMO-Net augments and improves the benthic habitat classification accuracy of low-resolution datasets across large geographic and temporal scales using high-resolution training data from FluidCam. NeMO-Net's FCNN uses ResNet and RefineNet to perform semantic segmentation and cloud masking of remote sensing imagery of shallow marine systems from drones, manned aircraft, and satellites, including FluidCam, WorldView, Planet, Sentinel, and Landsat. Deep Laplacian Pyramid Super-Resolution Networks (LapSRN) alongside Domain Adversarial Neural Networks (DANNs) are used to augment low resolution imagery with high resolution drone-based datasets as well as recognize domain-invariant features across multiple instruments to achieve high classification accuracies, ameliorating inter-sensor spatial, spectral and temporal heterogeneities. An online active learning and citizen science application is used to allows users to provide interactive training data for NeMO-Net in 2D and 3D, fully integrated within an active learning framework. Preliminary results from a test case in Fiji demonstrate 9-class classification accuracy exceeding 84%.

Vulnerability verification is an important process in ensuring the security of software systems. To support users in this process, we present the design study of DaV3is, which utilizes visual event sequence analysis techniques to enable the comparison and tracing of automatically detected data flows through the software's source code, thereby allowing users to take advantage of sequence similarities to reduce the verification workload. To that end, we characterize the domain problem based on input from domain users, describe our design rationale based on best-practices from the visual analytics literature, and evaluate individual design decisions, usability, and utility in studies with three stakeholder groups. The evaluations yielded overall positive responses, showing the suitability of our design and providing valuable insight for future research.

Software that operates actively needs to be adequately managed and maintained to avoid future problems. One of the problems in the software is the existence of code smells. A code smell indicates that a problem will, in turn, occur in the software's source code. OpenSID is open-source software used to create village administration websites. The detection carried out in this study shows that the OpenSID source code contains 32 code smells of High Cyclomatic Complexity type, which, if left untreated, can cause problems in the future. One of the problems is the reduced level of maintainability of the software. This research performs refactoring using the Extract Function method on a High Cyclomatic Complexity code smell to analyze what effect refactoring has on the code smell and its impact on the value of the OpenSID Maintainability Index. The research used the following tools: PHPMD, PHPDepend, and PhpStorm®. The results showed that after refactoring, the High Cyclomatic Complexity code smells were gone in some functions, and the average value of the OpenSID Maintainability Index increased from 37.0337 to 42.0912. Even though, in the end, the maintainability of this software is still in the category of difficult to maintain, this Maintainability Index increases to 13.66% has shown that the refactoring Extract Function method in code smells High Cyclomatic Complexity can increase the maintainability of this software.

Software plays an increasingly important role in today's world. With the advent of open-source software, an increasing number of developers begin to focus on and apply open-source software as a basic tool for their program development. However, at the same time, introducing open-source software into their own software introduces various types of defects and disadvantages. These unknown risks may cause incalculable economic loss aud credit crises if they were to be exploited in the future. Therefore, it is an important and urgent problem to detect the components of open-source software that may be reused when outsourcing software. To help detecting opensource software components in large-scale software projects, this paper proposes automatic identification technology for subnetworks with similar structural characteristics. This technology is based on node role classification, node similarity matching, and similar subnetwork search. This subject applies complex network technology to the comparison of software networks. In contrast to traditional code detection technology, this study does not constrain the text information of the software's source code. Considering that the basic skeleton structure of an application, in the processes of code reuse and features, remains the same, its network structure is used instead of its software structure to avoid problems such as poor detection results of similar codes as a result of text modification. This technology starts from the software features and the structure of software network.

As the supervisory control and data acquisition (SCADA) become more and more distributed and complex, the job of designing data acquisition server (DAS) software is very challenging. This paper advances a new design method for data acquisition server software based on multi-agent. By using multi-agent technology, the flexible intelligent communication protocol interface and optimal load balancing policy are designed to tackle the problems of variety communication protocols and heavy network loads. It can easily add new protocol or modify protocol without any modification of the protocol interface software's source code. So, the performance such as scalability and maintainability of the DAS has been greatly improved. This method has already been applied in the real project, and the desirable result of the project field proves its validity.

Identifying the appropriate kernel function/matrix for a given dataset is essential to all kernel-based learning techniques. A number of kernel learning algorithms have been proposed to learn kernel functions or matrices from side information (e.g., either labeled examples or pairwise constraints). However, most previous studies are limited to passive kernel learning in which side information is provided beforehand. In this paper we present a framework of Active Kernel Learning (AKL) that actively identifies the most informative pairwise constraints for kernel learning. The key challenge of active kernel learning is how to measure the informativeness of an example pair given its class label is unknown. To this end, we propose a min-max approach for active kernel learning that selects the example pair that results in a large classification margin regardless of its assigned class label. We furthermore approximate the related optimization problem into a convex programming problem. We evaluate the effectiveness of the proposed algorithm by comparing it to two other implementations of active kernel learning. Empirical study with nine datasets on semi-supervised data clustering shows that the proposed algorithm is more effective than its competitors.

Active learning is a useful technique for tasks for which unlabeled data is abundant but manual labeling is expensive. One example of such a task is semantic role labeling (SRL), which relies heavily on labels from trained linguistic experts. One challenge in applying active learning algorithms for SRL is that the complete knowledge of the SRL model is often unavailable, against the common assumption that active learning methods are aware of the details of the underlying models. In this paper, we present an active learning framework for black-box SRL models (i.e., models whose details are unknown). In lieu of a query strategy based on model details, we propose a neural query strategy model that embeds both language and semantic information to automatically learn the query strategy from predictions of an SRL model alone. Our experimental results demonstrate the effectiveness of both this new active learning framework and the neural query strategy model.

Recently, Time Series Remote Sensing Images (TSRSIs) have been proven to be a significant resource for land use/land cover (LULC) mapping. Deep learning methods perform well in managing and processing temporal dependencies and have shown remarkable advancements within this domain. Although deep learning methods have exhibited outstanding performance in classifying TSRSIs, they rely on enough labeled time series samples for effective training. Labeling data with a wide geographical range and a long time span is highly time-consuming and labor-intensive. Active learning (AL) is a promising method of selecting the most informative data for labeling to save human labeling efforts. It has been widely applied in the remote sensing community, except for the classification of TSRSIs. The main challenge of AL in TSRSI classification is dealing with the internal temporal dependencies within TSRSIs and evaluating the informativeness of unlabeled time series data. In this paper, we propose a data-driven active deep learning framework for TSRSI classification to address the problem of limited labeled time series samples. First, a temporal classifier for TSRSI classification tasks is designed. Next, we propose an effective active learning method to select informative time series samples for labeling, which considers representativeness and uncertainty. For representativeness, we use the K-shape method to cluster time series data. For uncertainty, we construct an auxiliary deep network to evaluate the uncertainty of unlabeled data. The features with rich temporal information in the classifier's middle-hidden layers will be fed into the auxiliary deep network. Then, we define a new loss function with the aim of improving the deep model's performance. Finally, the proposed method in this paper was verified on two TSRSI datasets. The results demonstrate a significant advantage of our method over other approaches to TSRSI. On the MUDS dataset, when the initial number of samples was 100 after our method selected and labeled 2000 samples, an accuracy improvement of 4.92% was achieved. On the DynamicEarthNet dataset, when the initial number of samples was 1000 after our method selected and labeled 2000 samples, an accuracy improvement of 7.81% was attained. On the PASTIS dataset, when the initial number of samples was 1000 after our method selected and labeled 2000 samples, an accuracy improvement of 4.89% was achieved. Our code is available in Data Availability Statement.