Multi-representational Learning Research Articles

Fully unsupervised wear anomaly assessment of aero-bearings enhanced by multi-representation learning of deep features

The assessment of wear anomalies using vision-based techniques is crucial for automated inspections of machine health conditions. Given the diversity of worn anomalies and the time-consuming nature of labeling, recent advancements in unsupervised anomaly detection have focused on comparing extracted features with those of normal images. However, this approach faces significant challenges in constructing a comprehensive comparison sample database and extracting distinctive features from worn surfaces with varying degrees of severity. To overcome these limitations, we introduce a fully unsupervised anomaly assessment method that detects anomalies solely based on single-bearing images, without the need for unworn samples. This approach eliminates the requirement for a comparison sample database by constructing an implicit feature database from the single-bearing surfaces using pre-trained convolutional neural networks and dimensionality reduction strategies. Additionally, to address the limited differentiability of wear characteristics, we develop a feature refinement strategy that incorporates multi-representation learning and topographical distribution characteristics of worn surfaces. By integrating the statistics of the refined feature database with uncertainty distribution-based normal feature weights, we construct an anomaly distribution map. This map enables us to assess the anomaly degree of a single-bearing effectively. To validate our method, we conducted experiments using real aero-engine bearings. The results demonstrate that our proposed approach can accurately assess anomaly degrees without pre-prepared comparison samples, achieving a detection performance of 90%, which fulfills the requirements for wear anomaly assessment.

EXPLORING MULTI-REPRESENTATION LEARNING APPROACH IN MATHEMATICS ELEMENTARY SCHOOL STUDENTS

The purpose of this study was to analyze the effectiveness of the multi-representation learning approach in improving the understanding of mathematical concepts for elementary school students. The inclusion criteria in this review include studies that clearly apply a multi-representation approach in primary school settings and have relevant data to evaluate. Studies that do not provide adequate data or do not meet the inclusion criteria are excluded. Sources of information used to identify these studies include various academic databases such as ERIC, Springer, Elsevier, Ebschoohost, Taylor &; Francis, the ACM Technical Symposium on Computer Science Education and MDPI publishers. The last search was conducted last month before the writing of this abstract. Methods of evaluating bias risk in included studies include methodological quality assessment, data quality control, and evaluation of the accuracy of the statistical analysis used. The main findings of the study are presented through thematic and narrative analysis, with a total of 75 scientific studies included in this review. The characteristics of relevant studies will be discussed further in the analysis which includes research content, study design, type of participants, learning domain, and year of publication. The findings show that with various forms of representation students can improve their understanding of mathematics from informal to formal levels, in addition to the fact that multi-representation learning can deepen students' understanding of complex mathematical concepts.

Deep Multirepresentation Learning for Data Clustering.

Deep clustering incorporates embedding into clustering in order to find a lower-dimensional space suitable for clustering tasks. Conventional deep clustering methods aim to obtain a single global embedding subspace (aka latent space) for all the data clusters. In contrast, in this article, we propose a deep multirepresentation learning (DML) framework for data clustering whereby each difficult-to-cluster data group is associated with its own distinct optimized latent space and all the easy-to-cluster data groups are associated with a general common latent space. Autoencoders (AEs) are employed for generating cluster-specific and general latent spaces. To specialize each AE in its associated data cluster(s), we propose a novel and effective loss function which consists of weighted reconstruction and clustering losses of the data points, where higher weights are assigned to the samples more probable to belong to the corresponding cluster(s). Experimental results on benchmark datasets demonstrate that the proposed DML framework and loss function outperform state-of-the-art clustering approaches. In addition, the results show that the DML method significantly outperforms the SOTA on imbalanced datasets as a result of assigning an individual latent space to the difficult clusters.

Peningkatan Kemampuan Pemecahan Masalah Matematis Siswa Melalui Model Pembelajaran Dircursus Multy Reprecentation Pada Mata Pelajaran Matematika di Sekolah Dasar

Primary research related to the implementation of Discourse Multi Representation (DMR) to improve students' mathematical problem solving abilities has been carried out a lot. These studies need to be identified, their developments analyzed, and general conclusions drawn so that they can be used as suggestions for the progress of researchers and practitioners in the field of education. This study aims to look at research trends by identifying and classifying all research results related to the application of the Discourse Multi Representation (DMR) model to improve students' mathematical problem solving skills in Indonesia in 2018-2022. This study uses the Systematic Literature Review (SLR) method. There are 7 related articles analyzed in this study. The results of this SLR show that the most research on the implementation of the Discourse Multi Representation (DMR) learning model on students' mathematical problem solving abilities was mostly carried out in 2021 and the most research subjects were elementary students. The most widely used research method is the quantitative method and in general the results show that the Discursus Multy Reprecentation (DMR) model is able to have a positive impact or improve students' mathematical problem solving skills compared to other learning models. The results of this study can be a consideration for researchers in conducting research and for teachers in carrying out learning in the classroom

Trust-worth multi-representation learning for audio classification with uncertainty estimation

Multi-view learning has been explored for audio classification tasks, exploiting different representations of audio signals, ranging from MFCC, CQT, to raw signals. The quality of each view may vary for different audio signals, and the appropriate uncertainty quantification for each view has not been fully explored. In this work, we explore a trusted multi-view learning framework for classification tasks in order to fully incorporate different views. Our framework consists of three parallel branches of Transformer architectures (Gammatone spectrogram, log-Mel and CQT) and they are combined using the uncertainty estimation of different branch. In addition to computing the classification probabilities, the uncertainty of each representation can also be obtained using the framework. We firstly calculate the evidence based on feature vectors to obtain the probabilities and the uncertainty of classification problems for Gammatone, log-Mel and CQT branch. By integrating the confidence from each of the different representations using the Dempster–Shafer theory, the classification framework can provide higher accuracy and confidence. To demonstrate the effectiveness of the proposed framework, we conduct the experiments on the GTZAN dataset. The obtained results show that our method can reach the accuracy of 83.0%, which significantly outperforms single representation-based methods while providing uncertainty estimation for different views.

The impact of multiple representations on students' understanding of vector field concepts: Implementation of simulations and sketching activities into lecture-based recitations in undergraduate physics.

Multiple external representations (e.g., diagrams, equations) and their interpretations play a central role in science and science learning as research has shown that they can substantially facilitate the learning and understanding of science concepts. Therefore, multiple and particularly visual representations are a core element of university physics. In electrodynamics, which students encounter already at the beginning of their studies, vector fields are a central representation typically used in two forms: the algebraic representation as a formula and the visual representation depicted by a vector field diagram. While the former is valuable for quantitative calculations, vector field diagrams are beneficial for showing many properties of a field at a glance. However, benefiting from the mutual complementarity of both representations requires representational competencies aiming at referring different representations to each other. Yet, previous study results revealed several student problems particularly regarding the conceptual understanding of vector calculus concepts. Against this background, we have developed research-based, multi-representational learning tasks that focus on the visual interpretation of vector field diagrams aiming at enhancing a broad, mathematical as well as conceptual, understanding of vector calculus concepts. Following current trends in education research and considering cognitive psychology, the tasks incorporate sketching activities and interactive (computer-based) simulations to enhance multi-representational learning. In this article, we assess the impact of the learning tasks in a field study by implementing them into lecture-based recitations in a first-year electrodynamics course at the University of Göttingen. For this, a within- and between-subjects design is used comparing a multi-representational intervention group and a control group working on traditional calculation-based tasks. To analyze the impact of multiple representations, students' performance in a vector calculus test as well as their perceived cognitive load during task processing is compared between the groups. Moreover, analyses offer guidance for further design of multi-representational learning tasks in field-related physics topics.

Peningkatan Kemampuan Kognitif Materi Kalor Peserta Didik Kelas XI Melalui Implementasi Media Multi Representasi Berbasis Instagram

One of the learning strategies that can be applied using the e-Learning approaches in physics learning is multi-representational learning by utilizing Instagram social media Instagram is a digital social media platform it many attractive features and designs that can help display and share certain videos and photos. Using Instagram in this research is used as a learning resource in the network. Therefore, the research conducted aims to improve students' cognitive abilities on heat material through Instagram-based multi-representational media. The population in this study were students of XI MIA class at SMA Negeri 7 Ambon, consisting of 45 students. The sample in this study was class XI MIA 2, totaling 22 people obtained using a random sampling technique. Results of research on the initial abilities of students are below the minimum completeness criteria with an average value of 28.90, where 22 (100%) students are in failed the qualification. Then, after being given treatment in the teaching and learning process taking place, an increase was obtained in the final test results with an average value of 80,83. Apart from these results, it is also proven through the N-Gain test where students experience an increase with an average of 0,73 and are included in high qualifications. Based on the research results, it can be concluded that the learning process by utilizing Instagram-based multi-representational media can improve students' cognitive abilities on heat material.

Analysis Multi Representation Ability and Learning Outcomes of Optical Materials

The purpose of this study was to describe students' ability of multi-representation in solving physics problems on optical material and students' learning outcomes in Physics Department, Universitas Negeri Surabaya. This study is quantitative descriptive research. The sample of this research is one class of students in the physics education study program and one class of students in the physics study program who took program courses in the odd semester of 2021/2022. The research data are multi-representation abilities and learning outcomes of optical materials. The test instrument used is based on multi-representation, which has been validated theoretically and is declared valid. The results of the analysis show that students' multi-representation abilities that are quite good are verbal and visual representation abilities, each with a percentage, while the ability to very good representation is a mathematical representation and symbolic. There is no very significant difference between the multi-representation abilities of students in physics education study programs and students of physics study programs. It is hoped that in the future, the results of the study can become information and references for teachers and lecturers in carrying out physics learning on optical material.

Driver Maneuver Identification with Multi-Representation Learning and Meta Model Update Designs

Driver maneuver identification (DMI), i.e., the task of predicting the driver maneuver classes given sensor measurements (such as IMU sensors from mobile devices), can serve as a key enabler for many ubiquitous driver behavior analysis applications. Despite the prior studies, two major technical challenges remain before an effective DMI system can be deployed: (i) latent complex maneuver behavior feature relations for accuracy enhancement, and (ii) inconsistency and uncertainty in model adaptation given dynamic data collection settings (e.g., given different urban environments, drivers, and mobile devices). To address the aforementioned challenges, we propose MetaDMI, a novel and adaptive driver maneuver identification framework based on multi-representation learning and meta model update, with the case study on the inertial measurement unit (IMU) sensor measurements (i.e., accelerometer and gyroscope). Specifically, we first extract multiple feature representations for each driver maneuver record in the forms of graph, spectral, and time sequence. Next, MetaDMI processes them with a novel multi-representation learning network, extracting complex patterns and feature relations from the driver maneuvers. Finally, to further enhance the adaptivity of our DMI model to external impacts with dynamic data collection, we have designed a regularized meta learning-based training method to regularize the knowledge transfers across the source and target datasets (e.g., across cities/devices, and few-shot initialization) for consistent and robust identification performance. We have conducted extensive experimental studies upon our MetaDMI prototype based on two datasets (one is collected on our own) and shown that our approach outperforms other baseline approaches for DMI in terms of accuracy and adaptivity.

Using Augmented Reality to Enhance Students’ Representational Fluency: The Case of Linear Functions

Using multiple representations is advocated and emphasized in mathematics and science education. However, many students have difficulty connecting multiple representations of linear functions. Augmented Reality (AR) may affect these teaching and learning difficulties by offering dynamically linked representations. Inspired by this, our study aims to develop, implement, and evaluate an AR-based multi-representational learning environment (MRLE) with three representations of linear functions. The data were collected from 82 seventh graders from two high-performing classes in an urban area in China, through a pre-test, a post-questionnaire, and follow-up interviews. The results reveal that students were satisfied with the AR-based MRLE, which assisted in enhancing their understanding of the real-life, symbolic, and graphical representations and connections among them. Regarding students’ interactions with multiple representations, apparent differences in learning sequences and preferences existed among students in terms of their representational learning profile. In sum, learning in the AR-based MRLE is a complex interaction process between the mathematics content, forms of representations, digital features, and students’ representational learning profile.

Implementation of Multi-representation Learning on Climate Change Integrated Dynamic Fluid to Improve Student’s Problem-Solving Skills

This research aims to analyze the implementation of multi-representation learning on climate change topics integrated fluid dynamics to improve students' problem-solving skills and analyze student responses to multi-representation learning on climate change topics integrated fluid dynamics. The type of this research is quantitative research with a pre-experimental that uses a one-group pretest-posttest design. This research was conducted in XI MIPA-1 at Senior High School 4 Sidoarjo with 36 students. Data collection techniques in this research used tests with test instruments in the form of description questions and questionnaires (responses) on a Likert scale. The data analysis technique is N-gain for the description test instrument and descriptive for student response questionnaires. The results of this research obtained the score of N-gain is 0.565 with a category of the medium. The results of student responses carried out using a Likert scale questionnaire for ten statements had positive and very positive results with a percentage of more than 77%. Thus, it was concluded that learning multi-representation could improve students' problem-solving skills with the topic of climate change integrated with fluid dynamics and elicit positive and very positive responses from students.

KEMAMPUAN MULTIREPRESENTASI VERBAL DAN TABEL MAHASISWA PENDIDIKAN FISIKA SEMESTER 5 PADA MATERI KARAKTERISTIK GELOMBANG ELEKTROMAGNETIK

The purpose of writing this article is to determine the effectiveness of verbal and table multi- representation learning for 5th semester physics students on the characteristics of electromagnetic waves. The background of this article is that there are many physics students who have low verbal and table multirepresentation abilities, so this research needs to be done in order to find out the multi-presentation ability of physics students which can be seen from the results of their answers when solving problems on the characteristics of electromagnetic waves. The data collection process is carried out with a survey method of knowledge tests using a questionnaire on the google form application. The method used is a quantitative method, namely by collecting data, then analyzing it. The population of this study were 50 physics education students while the sample was fifth semester students who were studying the characteristics of electromagnetic waves. With the data collection method above, the results obtained are the results of the multi-representation in the form of a table which is higher than the multi-representation in the form of a verbal. Therefore, this multi-representation learning strategy in the form of a table can be a solution to increase student knowledge on the material characteristics of electromagnetic waves.Key words: multipresentation, learning, electromagnetic waves

Mono- and Multi-Representational Learning of the Covariational Aspect of Functional Thinking

Using multiple external representations is advocated for learning in STEM education. This learning approach assumes that multiple external representations promote richer mental representations and a deeper understanding of the concept. In mathematics, the concept of function is a prototypical content area in which multiple representations are used. However, there are hardly any experimental studies investigating the effect of learning functional thinking with multiple representations compared to learning with only one form of representation. Therefore, this article reports on a quasi-experimental intervention study with students from Grade 7, using three measurement time points. The study compared the multi-representational learning of functional thinking with both tables and graphs with mono-representational learning with either tables or graphs. The results show that multi-representational learning led to advantages in learning qualitative functional thinking. However, in quantitative functional thinking, learning with both graphs and tables did not result in higher learning gains than learning exclusively with graphs. Furthermore, students were better able to transfer their knowledge from graphs to tables than vice versa. The results also indicate that multi-representational learning requires more time than mono-representational learning but can lead to higher learning gains. In sum, the results show that the effect of learning with representations is a complex interaction process between learning content and the forms of representation.

Multi-Representation Discourse Learning Model As An Alternative Solution To Improve Student Learning Outcomes In Science Lessons

The purpose of this writing is to apply the Model of Multi-Representation Discourse Learning Model as an Alternative Solution to Improve Student Learning Outcomes in Science Subjects. The subject in this writing is grade 4 students at SD Islam Putradarma Tambun Selatan. The author found some facts of problems in the observation, when viewed from the daily replay of the students' science, from the number of 45 out of 2 classes and only 10 students who were able to achieve the KKM score. The author uses the Multi Representation Discourse Model as one of the solutions to improve the learning process of previous students' science.The Multi Representation Discourse Model is a lesson designed by teachers in order to generate discussion through problem presentation, assignments, and student training. this learning students communicate unofficially so that students are able to express ideas casually according to their own language.so that students can be more active in the learning process. Predicted results with multi-representation (DMR) discursion learning model can improve learning outcomes. There is an influence of multi-representation discursion learning model on student learning outcomes, with indicators of cognitive learning outcomes C1 (Knowledge). C2 (Understanding), C3 (Application), C4 (Analysis), C5 (Synthesis), C6 (Evaluation). In the process of teaching and learning, this cognitive aspect is the most prominent and can be seen directly from the test results.

MENINGKATKAN KEMAMPUAN BERPIKIR KREATIF MATEMATIS MELALUI MODEL GENERATIVE MULTI-REPRESENTATION LEARNING

The ability to think creatively is a characteristic of 21st century learning. It takes learning innovation to accommodate increased creative thinking in learning materials. One alternative learning model that is thought to improve creative thinking is the Generative Multi-Representation Learning (GMRL) model. This study aims to analyze the effect of the GMRL model on mathematical creative thinking skills. The research was conducted in class X IPA at a public high school in Tangerang Regency. The method used was a quasi-experimental design with a post-test only randomize control group. The research sample was 66 students, 33 as the experimental group and 33 as the control group. Determination of the sample using cluster random sampling technique. Data collection on mathematical creative thinking skills using test instruments. In general, the research findings reveal that students' mathematical creative thinking skills taught using the GMRL model are higher than the conventional model. These skills include fluency, flexibility, originality and elaboration. The fluency indicator is more dominant than the other three indicators. The conclusion of this study is that the GMRL model is more effective in increasing mathematical creative thinking skills than the conventional model.

DEVELOPING A MULTIREPRESENTATION LEARNING MODEL BASED ON LOCAL WISDOM TO TRANSFORM CHARACTER FOR STUDENTS OF 5-6 YEARS OLD

This research aimed to: (1) develop a multirepresentation learning model, (2) reveal the effectiveness of the developed multirepresentation learning model, and (3) transform character for students of 5-6 years old. The data were collected using a validation sheet, questionnaire of teacher’s responses, and observation. The data were analyzed using the independent sample t test with a significance level of 0,05. The result of the assessment by the material expert validator was “Good”. The research finding revealed that the multirepresentation learning model in terms of the content expediency, the language, the presentation, and the graphic component were categorized “Very Good”. There was a difference in the nation spirit character between the experimental class and the control class after the teacher used the multirepresentation learning model with p < 0.05, and there was also a significant increase, with p = 0.000.

Characterizing the growth of one student’s mathematical understanding in a multi-representational learning environment

The purpose of this study was to characterize the growth of one student’s mathematical understanding and use of different representations about a geometric transformation, dilation. We accomplished this purpose by using the Pirie-Kieren model jointly with the Semiotic Representation Theory as a lens. Elif, a 10th- grade student, was purposefully chosen as the case for this study because of the growth of mathematical understanding about dilation she exhibited over time. Elif participated in task-based interviews before, during and after participating in a variety of transformation lessons where she used multiple representations, including physical and virtual manipulatives. The results revealed that Elif was able to progress in her mathematical understanding from informal levels to the formal levels in the Pirie-Kieren model as she performed treatments and conversions, movements involving different registers of representations. The results also showed numerous examples of Elif’s mathematical understanding based on folding back activities, complementary aspects of acting and expressing, and interventions. Using the two theories together provides a powerful and holistic approach to a deeper understanding of mathematical learning by characterizing and articulating the growth of mathematical understanding and the way of mathematical thinking.

Low-Rank Regularized Multi-Representation Learning for Fashion Compatibility Prediction

The currently flourishing fashion-oriented community websites and the continuous pursuit of fashion have attracted the increased research interest of the fashion analysis community. Many studies show that predicting the compatibility of fashion outfits is a nontrivial task due to the difficulty in capturing the implicit patterns affecting fashion compatibility prediction and the complex relationships presented by raw data. To address these problems, in this paper, we propose a transductive low-rank hypergraph regularizer multiple-representation learning framework (LHMRL), whereby we formulate the processes of feature representation and fashion compatibility prediction in a joint framework. Specifically, we first introduce a low-rank regularized multiple-representation learning framework, in which the lowest-rank multiple representations of samples can be learned to characterize samples from different perspectives. In this framework, we maximize the total difference among multiple representations based on Grassmann manifold theory and incorporate a common hypergraph regularizer to naturally encode the complex relationships between fashion items and an outfit. To enhance the representation ability of our model, we then develop a supervised learning term by exploiting two types of supervision information from labeled data. Experiments on a publicly available large-scale dataset demonstrate the effectiveness of our proposed model over the state-of-the-art methods.

Penerapan Model Pembelajaran Diskursus Multy Reprecentacy (DMR) untuk Meningkatkan Pemahaman dan Prestasi Belajar Siswa


 The research problem is the low activity in the classroom in the learning process, especially in the process of activity and achievement which is still rigid. The purpose of this study was to find out by applying the Discourse Multy Representace learning model that could improve student activity and achievement abilities. This research was conducted at SMP N 1 Hinai in the 2020/2021 academic year. With the subject of 36 students. This research was formed in the CAR research. In data collection, the data obtained are student learning activity observation sheets and student learning outcomes tests. For data analysis used quantitative and qualitative data. Based on data analysis, the result of student activity (cycle I was 38.88%), while in cycle II it was (86.21%). On student learning outcomes from the pretest data obtained a score of 62.7 with 12 students who completed and 24 students who did not complete. In the first cycle posttest obtained a value of (73.19%) with a total of 8 students who completed from 28 students who did not complete. And in the second cycle posttest with a value of (94.16) with 35 students who completed and 1 student who did not complete. This shows the success of the research conducted. It can be concluded that the application of collaborating the Discourse Multi-Representation learning model with the concept sentence on the students has succeeded with what was expected. 
 
 Keywords: Multi-Representation Discourse; improve; Understanding and learning achievement; Students

Multi-representational learning for Offline Signature Verification using Multi-Loss Snapshot Ensemble of CNNs

Offline Signature Verification (OSV) is a challenging pattern recognition task, especially in the presence of skilled forgeries that are not available during training. This study aims to tackle its challenges and meet the substantial need for generalization for OSV by examining different loss functions for Convolutional Neural Network (CNN). We adopt our new approach to OSV by asking two questions: 1. which classification loss provides more generalization for feature learning in OSV?, and 2. How integration of different losses into a unified multi-loss function lead to an improved learning framework?These questions are studied based on analysis of three loss functions, including cross entropy, Cauchy-Schwarz divergence, and hinge loss. According to complementary features of these losses, we combine them into a dynamic multi-loss function and propose a novel ensemble framework for simultaneous use of them in CNN. Our proposed Multi-Loss Snapshot Ensemble (MLSE) consists of several sequential trials. In each trial, a dominant loss function is selected from the multi-loss set, and the remaining losses act as a regularizer. Different trials learn diverse representations for each input based on signature identification task. This multi-representation set is then employed for the verification task. An ensemble of SVMs is trained on these representations, and their decisions are finally combined according to the selection of most generalizable SVM for each user.We conducted two sets of experiments based on two different protocols of OSV, i.e., writer-dependent and writer-independent on three signature datasets: GPDS-Synthetic, MCYT, and UT-SIG. Based on the writer-dependent OSV protocol, On UT-SIG, we achieved 6.17% Equal Error Rate (EER) which showed substantial improvement over the best EER in the literature, 9.61%. Our method surpassed state-of-the-arts by 2.5% on GPDS-Synthetic, achieving 6.13%. Our result on MCYT was also comparable to the best previous results. The second set of experiments examined the robustness of our proposed method to the arrival of new users enrolled in the OSV system based on the writer-independent protocol. The results also confirmed that our proposed system efficiently performed the verification of new users enrolled in the OSV system.