Active Learning Model Research Articles

Abstract 1041: Spatial concordance between genome and proteome in oncology: A systematic review

Abstract We recently reported that spatial transcriptome data can be used to spatially detect copy number alterations (Erickson et al., Nature, 2022). Given that copy-number alterations may also be reflected by altered protein expression in human cancers, we sought to determine what the empirical concordance of the spatial proteome and genome in human cancer samples. We performed a systematic review of Immunohistochemistry (IHC) & Fluorescence/Chromogenic In-Situ Hybridization (FISH/CISH) across human cancers. We performed a PUBMED query for records up to April 1st, 2023, yielding 9,294 records. Our inclusion criteria included human sample studies that performed both IHC and FISH/CISH against the same target, with a minimum of two samples. We excluded case reports, RNA ISH- and non-human studies, studies without IHC/ISH performed on the same target, and articles not in English or inaccessible. We screened 9,294 PUBMED article abstracts using the AS Review active learning model (van de Schoot et al., 2021), and used a pre-defined threshold of 5 % consecutive irrelevant abstracts to stop screening. A total of 1,309 abstracts were manually excluded using the AS Review, directing the active learning model to automatically exclude 3,413 abstracts out of the original 9,294. Of the 4,572 articles assessed for full-article screening, 1,167 were excluded for being out of scope, 303 for access issues, 227 for language incompatibility, and 236 for imprecise information, primarily due to non-representative samples or incorrect gene/protein target identification. Ultimately, out of 2,413 articles assessed for eligibility, 2,294 did not meet inclusion criteria, leaving 119 studies for our analysis. From these articles, we manually extracted kappa values for analysis and visualization in R. We identified articles from 19 different cancers with the majority studying Breast Cancer, Non-Small Cell Lung Cancer, Gastric Cancer, and Mesothelioma. We observed high Cohen’s Kappa values in Breast Cancer (median = 0.76), Non-Small Cell Lung Cancer (median = 0.91) and Mucinous Ovarian Carcinoma (median = 0.93). We identified studies from 14 different genes with majority being HER2, ALK and MYC. We observed high Cohen’s Kappa values in HER2 (median = 0.80), ALK (median = 0.80), ROS1 (median = 0.95) and MDM2 (median = 0.72). Our preliminary analysis indicates significant heterogeneity between tumor types for protein expression and copy-number alterations. To our knowledge, there are no previously reported systematic reviews that have empirically assessed the spatial concordance of proteome and genome, in human cancer samples. The authors often reported IHC/FISH results at patient, or section level, potentially introducing selection bias by not accounting for intra-tumoral heterogeneity at single-cell resolution. Future efforts are needed to spatially resolve gene-copy number and protein expression at single-cell resolution. Citation Format: Laura K. Sysivirta, Franziska Niemeyer, Veera Kurki, Inkalilja Pirinen, Laura Savolainen, Ceren Seref-Vujaklilja, Andrew Erickson. Spatial concordance between genome and proteome in oncology: A systematic review [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1041.

Active transfer learning for audiogram estimation.

Computational audiology (CA) has grown over the last few years with the improvement of computing power and the growth of machine learning (ML) models. There are today several audiogram databases which have been used to improve the accuracy of CA models as well as reduce testing time and diagnostic complexity. However, these CA models have mainly been trained on single populations. This study integrated contextual and prior knowledge from audiogram databases of multiple populations as informative priors to estimate audiograms more precisely using two mechanisms: (1) a mapping function drawn from feature-based homogeneous Transfer Learning (TL) also known as Domain Adaptation (DA) and (2) Active Learning (Uncertainty Sampling) using a stream-based query mechanism. Simulations of the Active Transfer Learning (ATL) model were tested against a traditional adaptive staircase method akin to the Hughson-Westlake (HW) method for the left ear at frequencies kHz, resulting in accuracy and reliability improvements. ATL improved HW tests from a mean of 41.3 sound stimuli presentations and reliability of dB down to dB. Integrating multiple databases also resulted in classifying the audiograms into 18 phenotypes, which means that with increasing data-driven CA, higher precision is achievable, and a possible re-conceptualisation of the notion of phenotype classifications might be required. The study contributes to CA in identifying an ATL mechanism to leverage existing audiogram databases and CA models across different population groups. Further studies can be done for other psychophysical phenomena using ATL.

Tahsin al-Kitabah and al-Khat al-Araby Learning Innovations in the Digital Era

The emergence of various kinds of technological tools, digital media, and social media are making learning easier, by using online correction systems both in learning Tahsin al-Kitabah and Khat Araby. Apart from that, there are many applications related to khat learning, such as kitabahku, Mashq, Ana Mukhtariful khat, Kelk, and other applications that not only support students in learning Khat but also support the development of their creativity in creating work. This requires educators to utilize digital media as a medium to support learning. The implementation of Tahsin al-Kitabah learning using the "Abajadun" method and also learning Khat Araby using Manhaj Hamidi have their respective advantages. The Abajadun method emphasizes understanding the keys of mufrodat and murokkab using the keys of alif, ba', jim, and dal. Meanwhile, Manhaj Hamidi uses a tiered system from easy types of khat to difficult types of khat. And one of the advantages is the Khat Diploma as a standardization of the Sanad system. Both methods are innovative learning methods based on digital media because the learning process also uses digital technology as a learning support. The results show an increased effectiveness of khat learning. Even though in practice, there are several problems, efforts to improve and improve the quality of learning must still be made. Meanwhile, the use of digital-based applications such as Kitabahku and also digital-based guidebooks is a learning model that uses a Learning Visualization approach because students are greatly helped in the learning process by understanding the writing of Tahsin al-Kitabah and Khat Araby. Third, using the Mashq, Ana Mukhtarif al-Khat, and Kelk applications, is one of the strategies and approaches using the Active Learning model, where students are also actively involved in learning using these media both for practice and the work process. So, they can master calligraphy from basic to advanced level and be able to work well using manual or digital media.

Efforts to increase student participation, achievement motivation, and learning outcomes through the active learning model type everyone is a teacher here in the subject of sociology class XI of SMA Negeri 2 Menyuke

This research was conducted in an effort to improve student participation, achievement motivation, and learning outcomes through the Active Learning Model Type Everyone is A teacher Here in the subject of sociology for grade XI students at SMA Negeri 2 Menyuke. This research is generally motivated by the fact that the learning outcomes of students in this subject at SMA Negeri 2 Menyuke are still low. Many efforts have been made, but they have not provided maximum results. The use of the everyone is A teacher here (ETH) learning model in this research is an innovative effort in sociology learning at SMA Negeri 2 Menyuke. The method used is Classroom Action Research, with the research subjects being grade XI IPS 1 students at SMA Negeri 2 Menyuke. In addition to observation to observe the implementation of the action and student participation in learning, the instruments used in this research are also questionnaires to collect data on achievement motivation and learning outcome tests to measure the success of learning using this method. There are two cycles used in this research, and it is concluded after the second cycle is implemented. The results of the research show that the use of everyone is a teacher here (ETH) can improve student participation, achievement motivation, and learning outcomes in sociology for grade XI IPS 1 students at SMA Negeri 2 Menyuke. The research results recommend that sociology teachers use this method by varying it with various relevant media.

The Effect of Storytelling Learning Using Role Playing Model Assisted by Audiovisual Media on Students Written and Storytelling Skills of Students Grade V Elementary School in Pringsurat Sub-District Temanggung District

The Role Playing learning approach aims to make learning more meaningful for students. The Role Playing learning approach, combined with multimedia resources, stands out as a chosen and viable strategy for overcoming emerging challenges. In this study, experimental research with a quantitative approach was applied to assess the impact of using storytelling learning models paired with various media on storytelling skills and writing stories in a fifth-grade elementary school class. Based on the assessment process that has been carried out, the results in this study are an assessment of typical differences in comprehension tests between experimental and control groups supporting the acceptance of H1 versus H0. With a bidirectional significance probability value of 0.005, H1 is accepted, and H0 is rejected because 0.000 is less than 0.05. Therefore, this study ensures that the Role Playing learning model has a significant influence on students' storytelling skills. Similarly, the evaluation of typical differences in comprehension tests between the experimental group and the control group also favored the acceptance of H1 over H0. With a bidirectional significance probability value of 0.005, H1 is accepted, and H0 is rejected because 0.000 is less than 0.05. In summary, role playing has a significant influence on students' storytelling skills. Assessment of typical differences in comprehension tests between the experimental group and the control group led to acceptance of H1 versus H0. With a bidirectional significance probability value of 0.005, H1 is accepted, and H0 is rejected because 0.000 is less than 0.05. In conclusion, Role Playing significantly affects students' ability to tell stories and write stories. Keywords: role playing, storytelling skills, story writing skills, audio-visual media.

Project-Based Learning Model Increases Student Creativity and Learning Outcomes in Pancasila and Citizenship Education Learning

The phenomenon of a lost generation results from less than optimal education services for children during the pandemic season. It causes low student learning outcomes. This research aims to analyze the project-based learning model on students' creativity and learning outcomes in learning Pancasila and citizenship education in high school. This research is quasi-experimental or quasi-experimental with a pretest-posttest control group design. The population in this study was 386 class XI students. The sample selection for the experimental and control groups used a random sampling technique, totaling 70 students. Data collection methods use questionnaires and tests. Data collection instruments use questionnaires and test questions. The analysis test uses the independent sample T-test and descriptive analysis. The research results showed that there was a significant difference in student creativity between the experimental class and the control class. There is a significant difference between the experimental and control classes' post-test results. It was concluded that applying the project-based learning model significantly influenced the creativity and learning outcomes of class XI MIPA-1 students in high school. This research implies that teachers can apply active and reflective learning models, such as project-based learning, to make learning more meaningful.

Deep Unsupervised Active Learning on Learnable Graphs.

Recently, deep learning has been successfully applied to unsupervised active learning. However, the current method attempts to learn a nonlinear transformation via an auto-encoder while ignoring the sample relation, leaving huge room to design more effective representation learning mechanisms for unsupervised active learning. In this brief, we propose a novel deep unsupervised active learning model via learnable graphs, named ALLGs. ALLG benefits from learning optimal graph structures to acquire better sample representation and select representative samples. To make the learned graph structure more stable and effective, we take into account k -nearest neighbor graph as a priori and learn a relation propagation graph structure. We also incorporate shortcut connections among different layers, which can alleviate the well-known over-smoothing problem to some extent. To the best of our knowledge, this is the first attempt to leverage graph structure learning for unsupervised active learning. Extensive experiments performed on six datasets demonstrate the efficacy of our method.

UPAYA MENINGKATKAN AKTIVITAS BELAJAR IPA DENGAN MODEL PEMBELAJARAN ACTIVE LEARNING TIPE PEER LESSON PADA SISWA KELAS V SDN 017 RANAH SINGKUANG

This research was motivated by the low learning activity of fifth-grade students of UPT SD Negeri 017 Singkuang Ranch. This study aims to increase the learning activities of fifth-grade students of UPT SD Negeri 017 Singkuang Ranah, which consists of 25 students. The research used was Classroom Action Research (PTK), carried out in two learning cycles. Each cycle consists of two meetings and four learning stages: planning, implementing, observing, and reflecting. Data collection techniques in the form of documentation, observation, and tests. In contrast, the data analysis technique uses qualitative analysis and quantitative analysis. Based on the data analysis, the results of the learning activities of class V UPT SD Negeri 017 Ranah Singkuang in the pre-action were obtained with a learning completeness percentage of 63.04%. In Cycle 1 Meeting I, with a learning completeness percentage of 67.61%, and in Cycle 1 Meeting II, there was an increase with a learning completeness percentage of 72%. In cycle two, the first meeting also increased with the rate of learning activities at 75.62%. In cycle 2, the second meeting experienced another increase, with the speed of learning completeness at 84.75%. Thus, applying the peer lesson type active learning model can increase the learning activities of class V UPT SD Negeri 017 Ranah Singkuang.Keywords : Model Active learning type peer lesson, Activities learning and Elementary School Students

Meningkatkan Pemahaman Konsep Matematika Siswa Sekolah Dasar melalui Pembelajaran Role Playing Berbantuan Uang Lipat (Uang Nilai Tempat)

Based on the results of pre-research observations at SD Negeri Depok 1 Sleman, it is known that some students have not reached the minimum completeness criteria (KKM) that have been determined by the school, namely 80. This research aims to determine the increase in understanding of Mathematics concepts for class II students at SD Negeri Depok 1 Sleman regarding addition. and reduction through role playing learning assisted by folding money. The research method used is classroom action research (PTK). Data collection for this research used test question sheets and teacher observation sheets. The data analysis method used is descriptive qualitative and and quantitative descriptive. The results obtained showed that teacher activity in cycle I was 66.67% (good) and increased in cycle II to 86.67% (very good). The average understanding of Mathematics concepts in cycle I obtained a score of 67 and increased in cycle II to 76.89. Completeness of the test results for understanding Mathematics concepts in cycle I was 57.14% and increased in cycle II to 78.57%. The results of this research indicate that the role playing learning model assisted by folding money media can improve students' understanding of mathematics concepts.

MicroCycle: An Integrated and Automated Platform to Accelerate Drug Discovery.

We herein describe the development and application of a modular technology platform which incorporates recent advances in plate-based microscale chemistry, automated purification, in situ quantification, and robotic liquid handling to enable rapid access to high-quality chemical matter already formatted for assays. In using microscale chemistry and thus consuming minimal chemical matter, the platform is not only efficient but also follows green chemistry principles. By reorienting existing high-throughput assay technology, the platform can generate a full package of relevant data on each set of compounds in every learning cycle. The multiparameter exploration of chemical and property space is hereby driven by active learning models. The enhanced compound optimization process is generating knowledge for drug discovery projects in a time frame never before possible.

A dual-ways feature fusion mechanism enhancing active learning based on TextCNN

Active Learning (AL) is a technique being widely employed to minimize the time and labor costs in the task of annotating data. By querying and extracting the specific instances to train the model, the relevant task’s performance is improved maximally within limited iterations. However, rare work was conducted to fully fuse features from different hierarchies to enhance the effectiveness of active learning. Inspired by the thought of information compensation in many famous deep learning models (such as ResNet, etc.), this work proposes a novel TextCNN-based Two ways Active Learning model (TCTWAL) to extract task-relevant texts. TextCNN takes the advantage of little hyper-parameter tuning and static vectors and achieves excellent results on various natural language processing (NLP) tasks, which are also beneficial to human-computer interaction (HCI) and the AL relevant tasks. In the process of the proposed AL model, the candidate texts are measured from both global and local features by the proposed AL framework TCTWAL depending on the modified TextCNN. Besides, the query strategy is strongly enhanced by maximum normalized log-probability (MNLP), which is sensitive to detecting the longer sentences. Additionally, the selected instances are characterized by general global information and abundant local features simultaneously. To validate the effectiveness of the proposed model, extensive experiments are conducted on three widely used text corpus, and the results are compared with with eight manual designed instance query strategies. The results show that our method outperforms the planned baselines in terms of accuracy, macro precision, macro recall, and macro F1 score. Especially, to the classification results on AG’s News corpus, the improvements of the four indicators after 39 iterations are 40.50%, 45.25%, 48.91%, and 45.25%, respectively.

Using dropout based active learning and surrogate models in the inverse viscoelastic parameter identification of human brain tissue.

Inverse mechanical parameter identification enables the characterization of ultrasoft materials, for which it is difficult to achieve homogeneous deformation states. However, this usually involves high computational costs that are mainly determined by the complexity of the forward model. While simulation methods like finite element models can capture nearly arbitrary geometries and implement involved constitutive equations, they are also computationally expensive. Machine learning models, such as neural networks, can help mitigate this problem when they are used as surrogate models replacing the complex high fidelity models. Thereby, they serve as a reduced order model after an initial training phase, where they learn the relation of in- and outputs of the high fidelity model. The generation of the required training data is computationally expensive due to the necessary simulation runs. Here, active learning techniques enable the selection of the "most rewarding" training points in terms of estimated gained accuracy for the trained model. In this work, we present a recurrent neural network that can well approximate the output of a viscoelastic finite element simulation while significantly speeding up the evaluation times. Additionally, we use Monte-Carlo dropout based active learning to identify highly informative training data. Finally, we showcase the potential of the developed pipeline by identifying viscoelastic material parameters for human brain tissue.

Strategi Pengembangan Kurikulum di SMP Islam Terpadu Pondok Pesantren Al Madinah Boyolali

Curriculum is an important tool in education. To improve the quality of education, curriculum development must be carried out comprehensively. This research aims to describe the curriculum development at Al Madinah Boyolali Integrated Islamic Middle School. The research method used in this research is a qualitative descriptive method, with a research approach in the form of a case study. The results of the research show that the curriculum at the Al Madinah Boyolali Integrated Islamic Middle School is an integrity curriculum, meaning an educational unit-level curriculum that combines the education service curriculum with the Saudi Arabian curriculum with superior tahfizhul Qur'an and Arabic language programs. The model of teaching and learning activities adopted at the Al Madinah Boyolali Integrated Islamic Middle School consists of two combined systems, namely the Salafiyah education system and the modern education system, namely that teaching and learning activities adhere to the school system in general. Four of the existing components include objectives, curriculum content, strategy, and evaluation. The objective component specifications relate to the four core competencies of graduates that must be mastered or achieved by the students of Al Madinah Boyolali Integrated Islamic Middle School, namely the core competency of monotheism, the core competency of having noble morals, the core competency of being intelligent, and the core competency of being competitive. The content components of the Al Madinah Boyolali Integrated Islamic Middle School curriculum are organized in three forms of activities, namely curricular, co-curricular, and extracurricular activities. The learning strategy component refers to the teaching approaches methods and equipment used in teaching. The curriculum evaluation component is intended to assess a curriculum as an educational program and determine the efficiency, effectiveness, relevance, and productivity of the program in achieving educational goals.

Virtual Active Learning to Maximize Knowledge Acquisition in Nursing Students: A Comparative Study.

Nursing students need to acquire knowledge through active methods that promote critical thinking and decision making. The purpose of this study is to analyze whether there are differences in the acquisition of knowledge by nursing students between active face-to-face or virtual teaching methods. In this comparative study, nursing students enrolled in the psychology course were divided into two groups: a face-to-face group that received active teaching methods and a virtual group. The virtual group was exposed to the Effective Learning Strategy (ELS), which included seminars based on video content through the Virtual Campus and answering questions using the H5P tool. In addition, participants engaged in reflection tasks on the content. Covariate data were collected, and knowledge tests were administered to both groups before and after the course. After three months, subjects were re-evaluated with a final exam to assess content retention. A total of 280 students were randomized. No differences were found in students' scores at the end of the knowledge test or in their final grades in the subject. Having study habits (b = 0.12, p = 0.03) and social support from relevant people (b = 0.09; p = 0.03) were associated with better post-intervention scores, and inversely with social support from friends (b = -0.12, p < 0.01). Final grades were inversely associated with digital safety literacy (b = -0.101, p = 0.01). No factors were associated with the scores of each group separately. The ELS virtual active learning model is as effective as face-to-face active learning methods for teaching psychology to first-year nursing students. This study was not registered.

Active machine learning model for the dynamic simulation and growth mechanisms of carbon on metal surface

Substrate-catalyzed growth offers a highly promising approach for the controlled synthesis of carbon nanostructures. However, the growth mechanisms on dynamic catalytic surfaces and the development of more general design strategies remain ongoing challenges. Here we show how an active machine-learning model effectively reveals the microscopic processes involved in substrate-catalyzed growth. Utilizing a synergistic approach of molecular dynamics and time-stamped force-biased Monte Carlo methods, augmented by the Gaussian Approximation Potential, we perform fully dynamic simulations of graphene growth on Cu(111). Our findings accurately replicate essential subprocesses–from the preferred diffusion of carbon monomer/dimer, chain or ring formations to edge-passivated Cu-aided graphene growth and bond breaks by ion impacts. Extending our simulations to carbon deposition on metal surfaces like Cu(111), Cr(110), Ti(001), and oxygen-contaminated Cu(111), our results align closely with experimental observations, providing a practical and efficient approach for designing metallic or alloy substrates to achieve desired carbon nanostructures and explore further reaction possibilities.

Research on Teaching Methods of Integrating Watercolor Techniques and Information Technology in Higher Education Art Education

Abstract Watercolor painting is a visual and creative discipline, which is closely related to the development of digital technology, which also affects the methods and means of basic art education in China. In this paper, we detect the trajectory of gesture movement after preprocessing the gesture data collected from students in watercolor painting using a Leap Motion sensor. We use Euclidean distance and nearest neighbor laws to identify and classify gesture movements in order to recognize students’ gesture movements in art education. Then, the virtual interactive system is designed with Unity 3D software, and finally, the art teaching method combining watercolor techniques and information technology is established. The results of teaching practice show that the level of students’ watercolor technique skills in all dimensions under innovative teaching is much higher than that of students under traditional teaching, and the significance values are all less than 0.05, which indicates that the use of VR technology has a very obvious impact on the teaching of watercolor painting. Meanwhile, the scores of students’ watercolor painting works were also improved with the assistance of VR technology. This study can provide a case reference for the subsequent research on the application of augmented reality technology and related teaching resources to art education and can enrich the construction of a new model of art learning activities in colleges and universities with technical support.

Active Blended Learning Model for Teaching IoT Application Development Course: A Case Study

Abstract—Traditional University learning approaches in Software Engineering, focusing on students listening to lecturers, then individually performing tasks is an isolating experience, a long way from the inherently collaborative approach in the workforce. Active learning focuses on student-centered learning in a much more authentic experience. This paper focuses on experiences in developing a course in the transnational education context involving a partnership between Indian and Australian universities. It highlights the design of a technical Software Engineering course for Internet of Things (IoT) application development with blended and active learning strategies and focuses on insights gained from implementing this curriculum for two different cohorts at an Indian university. Keywords—Active Learning, Blended Learning, IoT, Student Engagement, Embedded System, Software Engineering.

Accelerated design of nickel-cobalt based catalysts for CO2 hydrogenation with human-in-the-loop active machine learning.

Thermo-catalytic conversion of CO2 into more valuable compounds, such as methane, is an attractive strategy for energy storage in chemical bonds and creating a carbon-based circular economy. However, designing heterogeneous catalysts remains a challenging, time- and resource-consuming task. Herein, we present an interpretable, human-in-the-loop active machine learning framework to efficiently plan catalytic experiments, execute them in an automated set-up, and estimate the effect of experimental variables on the catalytic activity. A dataset with 48 catalytic activity tests was compiled from a design space of Ni-Co/Al2O3 catalysts with over 50 million potential combinations in only eight iterations. This small dataset was found sufficient to predict CO2 conversion, methane selectivity, and methane space-time yield with remarkable accuracy (R 2 > 0.9) for untested catalysts and reaction conditions. New experiments and catalysts were selected with this methodology, leading to experimental conditions that improved the methane space-time yield by nearly 50% in comparison to the previously obtained maximum in the dataset. Interpretation of the model predictions unveiled the effect of each catalyst descriptor and reaction condition on the outcome. Particularly, the strong predicted inverse trend between the calcination temperature and the catalytic activity was validated experimentally, and characterization implied an underlying structure-performance relationship. Finally, it is demonstrated that the deployed active learning model is excellently suited to predict and fit kinetic trends with a minimal amount of data. This data-driven framework is a first step to faster, model-based, and interpretable design of catalysts and holds promise for broader applications across catalytic processes.

AI-Driven Dynamics: ChatGPT Transforming ELT Teacher-Student Interactions

In the rapidly evolving educational domain, where artificial intelligence (AI) is pivotal, English Language Teaching (ELT) witnesses profound transformations. This study, focusing on integrating ChatGPT, an AI-driven language model, illuminates its substantial impact on the dynamics between teachers and students within ELT. Pertinent to this research is the inquiry into how ChatGPT reshapes teacher roles, responsibilities, and pedagogical practices, an area previously overlooked in the realm of AI in education. Employing a theoretical lens, the study scrutinises active learning models, technology integration theories, and paradigms of teacher professional development. Such analysis is crucial in comprehending teachers' shift from traditional methodologies to more facilitative and supportive roles, emphasising the urgency of this investigation for active learning's progression. Utilising diverse analytical approaches, ChatGPT propels pedagogies encouraging active learning, notably blended, collaborative, and project-based learning, while accentuating the necessity of professional development and institutional support for teachers' successful transition to these novel roles. These findings offer pivotal insights for the higher education sector, underlining the imperative of embracing AI's transformative potential, particularly tools like ChatGPT. By delving into ChatGPT's influence on teacher-student interaction and pedagogical methods, the study carves a significant path for future exploration and application in technology-enhanced ELT.

An Integrated Active Deep Learning Approach for Image Classification from Unlabeled Data with Minimal Supervision

The integration of active learning (AL) and deep learning (DL) presents a promising avenue for enhancing the efficiency and performance of deep learning classifiers. This article introduces an approach that seamlessly integrates AL principles into the training process of DL models to build robust image classifiers. The proposed approach employs a unique methodology to select high-confidence unlabeled data points for immediate labeling, reducing the need for human annotation and minimizing annotation costs. Specifically, by combining uncertainty sampling with the pseudo-labeling of confident data, the proposed approach expands the training set efficiently. The proposed approach uses a hybrid active deep learning model that selects the most informative data points that need labeling based on an uncertainty measure. Then, it iteratively retrains a deep neural network classifier on the newly labeled samples. The model achieves high accuracy with fewer manually labeled samples than traditional supervised deep learning by selecting the most informative samples for labeling and retraining in a loop. Experiments on various image classification datasets demonstrate that the proposed model outperforms conventional approaches in terms of classification accuracy and reduced human annotation requirements. The proposed model achieved accuracy of 98.9% and 99.3% for the Cross-Age Celebrity and Caltech Image datasets compared to the conventional approach, which achieved 92.3% and 74.3%, respectively. In summary, this work presents a promising unified active deep learning approach to minimize the human effort in manually labeling data while maximizing classification accuracy by strategically labeling only the most valuable samples for the model.