Learning Problems Research Articles

Stochastic Gradient Descent with Pre-Conditioned Polyak Step-Size

Stochastic Gradient Descent (SGD) is one of the many iterative optimization methods that are widely used in solving machine learning problems. These methods display valuable properties and attract researchers and industrial machine learning engineers with their simplicity. However, one of the weaknesses of this type of methods is the necessity to tune learning rate (step-size) for every loss function and dataset combination to solve an optimization problem and get an efficient performance in a given time budget. Stochastic Gradient Descent with Polyak Step-size (SPS) is a method that offers an update rule that alleviates the need of fine-tuning the learning rate of an optimizer. In this paper, we propose an extension of SPS that employs preconditioning techniques, such as Hutchinson’s method, Adam, and AdaGrad, to improve its performance on badly scaled and/or ill-conditioned datasets.

Tutorial on Molecular Latent Space Simulators (LSSs): Spatially and Temporally Continuous Data-Driven Surrogate Dynamical Models of Molecular Systems.

The inherently serial nature and requirement for short integration time steps in the numerical integration of molecular dynamics (MD) calculations place strong limitations on the accessible simulation time scales and statistical uncertainties in sampling slowly relaxing dynamical modes and rare events. Molecular latent space simulators (LSSs) are a data-driven approach to learning a surrogate dynamical model of the molecular system from modest MD training trajectories that can generate synthetic trajectories at a fraction of the computational cost. The training data may comprise single long trajectories or multiple short, discontinuous trajectories collected over, for example, distributed computing resources. Provided the training data provide sufficient sampling of the relevant thermodynamic states and dynamical transitions to robustly learn the underlying microscopic propagator, an LSS furnishes a global model of the dynamics capable of producing temporally and spatially continuous molecular trajectories. Trained LSS models have produced simulation trajectories at up to 6 orders of magnitude lower cost than standard MD to enable dense sampling of molecular phase space and large reduction of the statistical errors in structural, thermodynamic, and kinetic observables. The LSS employs three deep learning architectures to solve three independent learning problems over the training data: (i) an encoding of the high-dimensional MD into a low-dimensional slow latent space using state-free reversible VAMPnets (SRVs), (ii) a propagator of the microscopic dynamics within the low-dimensional latent space using mixture density networks (MDNs), and (iii) a generative decoding of the low-dimensional latent coordinates back to the original high-dimensional molecular configuration space using conditional Wasserstein generative adversarial networks (cWGANs) or denoising diffusion probability models (DDPMs). In this software tutorial, we introduce the mathematical and numerical background and theory of LSS and present example applications of a user-friendly Python package software implementation to alanine dipeptide and a 28-residue beta-beta-alpha (BBA) protein within simple Google Colab notebooks.

The Contestation of National Adaptation Policies in Indonesia

Indonesia actively supports the United Nations Framework Convention on Climate Change to achieve the Global Goal on Adaptation by setting ambitious climate change adaptation pledges in its Nationally Determined Contributions. This paper explores the implementation of those adaptation commitments within a polycentric governance structure in Indonesia. There is a debate about whether polycentric institutions help or hinder the effectiveness of adaptation outcomes. The question remains, why is the implementation of climate change adaptation policies in Indonesia ineffective? This paper argues that the contestation between the National Development Planning Agency and the Ministry of Environment and Forestry, as two dominant ministries in adaptation, has beset the polycentric structure of adaptation governance in Indonesia and resulted in ineffective adaptation policy implementation. Moreover, the polycentric adaptation governance is ineffective because Indonesia is still facing problems in coordination, collaboration, policy learning, leadership, resource distribution, and dispersion of power.

What we can do with one qubit in quantum machine learning: ten classical machine learning problems that can be solved with a single qubit

What we can do with one qubit in quantum machine learning: ten classical machine learning problems that can be solved with a single qubit

Project-Based Learning Tool Model that empowers the Metacognitive Skills of MAN Model Manado Students

Problems in learning that do not empower students' thinking abilities, especially metacognitive skills, are of high concern in the 21st century. The demands of life require humans to have some required skills and must be empowered to learn through appropriate learning tools. This research aims to develop project-style learning tools that enable students' metacognitive skills in biology subjects. The research method used is Borg & Gall model development research with the following steps: 1) needs analysis to obtain the required information, 2) literature study, 3) development of learning tools using a project model, and 4) expert validation, then continued with limited group testing on users. The results of the development in the form of ATP and the Phase E Biology Teaching Module, show that the learning tools that have been developed are suitable according to material experts and learning experts with very high assessment categories, namely 94% and 90.66%, 100% of teacher assessment results, so that the learning tools are suitable for use. This is also reinforced by data from the results of applying learning tools to students by measuring students' metacognitive skills reaching a score of 76 or the good category. Keywords: learning tools, PjBL, metacognitive skills

Exploring Design-Based Research as a framework for addressing pedagogical problems faced by higher education

Design-Based Research presents a pragmatic approach to addressing both pedagogical problems and innovations within a rigorous framework that aims to build transferable practice. This panel discussion will explore four examples of implementing DBR in various higher education contexts and draw participants into a discussion of how they might apply DBR to address identified pedagogical problems or innovations in their own contexts. Why DBR? This panel discussion is a focal point for exploring how design-based research (DBR) in higher education in 2024, can address pedagogical problems brought about by the global pandemic and the explosion of generative artificial intelligence use in assessment and learning (For examples). In a period of continuous disruption in higher education, flexible and ethical research strategies that embrace unexpected influences in teaching and learning is paramount (Cochrane et al., 2023), leading to a new generation of educational researchers embracing DBR. Furthermore, practical examples demonstrating students as partners in curriculum design to inform ‘what works’ in practice are essential for researchers to gain confidence in adopting research-teaching-practice methods of inquiry (Bakker, 2018; Cochrane & Munn, 2020). Understanding how the processes and outputs of DBR can meet these needs is crucial to alleviating both business and educational tensions about timing, funding, resources, and workload planning (Cochrane, 2022; McKenney & Reeves, 2019; Reeves & Lin, 2020). At the core of DBR is the opportunity to explore pedagogical problems or innovation through four phases: problem analysis, develop solutions to the identified problem, explore and evaluate the impact of the design in iterative interventions in real learning situations, leading to the development of transferable design principles and theory building (McKenney & Reeves, 2019) Various research methods can be applied to progressively ask participant perspectives, including students, teachers and learning designers, on appropriate approaches or technologies to solve an identified problem (McKenney & Reeves, 2019). The intended outcome of this panel discussion is to collaborate with participants in implementing DBR in higher education in response to specific pedagogical problems identified by the participants. Beginning with a brief background to DBR followed by four examples to stimulate discussion on its application in broader contexts to address identified problems in curriculum design. Relevance to Higher Education The relevance of debating the role of design-based research in higher education in 2024 is to stimulate discussion for how this pragmatic research approach informs ‘what works’ in situated learning contexts to develop cutting edge design principles and theory that can be transferrable (Galvin & Cochrane, 2023). The methodological framework of DBR invites students, teachers, and learning designers to collaborate from the conception of a design artefact or educational approach to solve a learning problem through to iterative stages of delivery. Following a new normal in higher education where change is inevitable, and fast-paced, research design that does not shy away from the progressive and ‘messy’ elements of educational settings to improve real time learning is essential. Finding a balance between naturalistic and interventionist research methods to explore this complexity is a way forward that DBR can provide. It is acknowledged that greater understanding for the value of DBR and how to plan practical stages with multiple participants is needed for teachers, researchers, and leaders in higher education to confidently adopt this approach (Haagen-Schützenhöfer et al., 2024). Opening discussion with peers models the collaborative ethos of DBR and opens potential connections and networking avenues to establish and generate knowledge for DBR in higher education. The Panel Members The panel will be comprised of project leaders from four DBR projects: Clinical reasoning development, interdisciplinary engineering education, biomechanics, performance anxiety. These projects were introduced in a concise paper for the ASCILITE 2023 conference (Cochrane et al., 2023) and the authors are in the process of collaborating on a full journal article analysing these DBR projects (Cochrane et al., 2024) that should be published in time for the ASCILITE 2024 conference.

Design Thinking

In today’s increasingly competitive Higher Education (HE) market, academic jobs are more precarious than ever. The way academics do their day-to-day job is under increasing scrutiny, with their outputs measured through the rigid lens of performativity and financial profit. Professional Learning (PL) is an institutional approach aimed towards achieving and sustaining high quality teaching and innovation among academic staff. The significance of PL to developing academics’ skills and fostering their professional identity has been confirmed by peer reviewed research. However, how to design and deliver PL that is meaningful to academic workforce, while empowering academics and supporting them in finding solutions to their teaching and learning problems, remains up for debate. This paper offers a conceptual way forward; arguing for a complete re-think of the predominant way of doing PL in universities in favour of placing Design Thinking (DT) at the heart of the matter. Grounded within a robust theoretical approach and scaffolded with a practical toolkit, DT can enable a nurturing environment to engage and empower academics, building their confidence and equipping them with a methodology for problem solving into the future.

Problematika Pembelajaran Fikih Mawaris pada Mahasiswa Berlatar Belakang Pendidikan SMA di Program Studi Pendidikan Agama Islam Universitas Negeri Padang

The problem in learning can obstruct and slow down the learning process, which can become an obstacle to achieving learning objectives. This study aims to identify the problems faced by students with a high school background in the Islamic Religious Education study program in learning Fiqh Mawaris, as well as the efforts made by both students and Islamic Law lecturers to address these issues. The method used is descriptive qualitative with informants including the head of the Islamic Religious Studies Department, two Islamic Law lecturers, and ten students from the Islamic Religious Education program. Data were collected through observation, interviews, and documentation, and then analyzed using data reduction, data presentation, and conclusion drawing techniques. Data validity was tested using source triangulation and technique triangulation. This study found three main problems in the Fiqh Mawaris learning process among students with a high school background: 1) difficulty in understanding the concepts of inheritance classification and distribution, 2) difficulty in analyzing inheritance distribution cases, and 3) challenges faced by students due to their high school educational background. The efforts made by the students to overcome these problems include: 1) writing and memorizing Fiqh Mawaris concepts, 2) seeking additional learning resources, and 3) discussing and studying together. Meanwhile, the lecturers addressed these problems by: 1) delivering material in stages, 2) using discussion and question-and-answer methods, and 3) conducting evaluations and providing feedback to students. The problems in Fiqh Mawaris learning for students with a high school background can be overcome through the active efforts of students seeking additional understanding and through a more structured and interactive teaching approach from the lecturers.

Learning properties of quantum states without the IID assumption

We develop a framework for learning properties of quantum states beyond the assumption of independent and identically distributed (i.i.d.) input states. We prove that, given any learning problem (under reasonable assumptions), an algorithm designed for i.i.d. input states can be adapted to handle input states of any nature, albeit at the expense of a polynomial increase in training data size (aka sample complexity). Importantly, this polynomial increase in sample complexity can be substantially improved to polylogarithmic if the learning algorithm in question only requires non-adaptive, single-copy measurements. Among other applications, this allows us to generalize the classical shadow framework to the non-i.i.d. setting while only incurring a comparatively small loss in sample efficiency. We leverage permutation invariance and randomized single-copy measurements to derive a new quantum de Finetti theorem that mainly addresses measurement outcome statistics and, in turn, scales much more favorably in Hilbert space dimension.

Pengembangan Media Video Animasi Pembelajaran Berbasis Pendekatan Inkuiri pada Mata Pelajaran Bahasa Inggris Kelas VIII SMP

The development of learning video media is carried out as an effort to solve student learning problems. This Learning Media can be used as an alternative to reduce problems in learning. This study aims to produce products in English class VIII subjects and determine the quality of validity and practicality of the products developed so that they are feasible to use. This study is a type of Research and Development (R&D) development or by using a 4-D development model (four-D model). This Model consists of 4 stages of development include Define, Design, Develop, Disseminate. The product validity test was conducted by 3 validators, namely 2 media validators and 1 material validator. Practicalities test was conducted on 31 students of Class VIII SMP Negeri 1 Junjung Sirih. The results of this study explain that the average percentage of media validators I and II respectively is 97.5% and 96.25% with the category of "Very Valid". From the assessment of material experts, an average percentage of 94.54% was obtained with the category of "Very Valid". For the assessment of practicality from students, a percentage of 93.22% was obtained with the product category "Very Practical" Based on these calculations, it can be concluded that the learning animation video media based on the inquiry approach in English subjects developed "Very Feasible" is used in learning

The Impact Of-Problematic Learning Styles-On Learning Outcomes Of Cultural Diversity In My Country 1 Sub-Diameter 1 Unity

The Impact of Problematic Learning Styles on Learning Outcomes in My Country Cultural Diversity Teachers and Teaching Education Dissertation Pacuan University. 2566 This study is a quasi-experimental study using traditional and problem-based learning models to investigate how cultural diversity sub-factors affect learning outcomes in My Country State School in Bokor IV C 78 Problem-based learning models have more requirements than traditional control class learning problems based on the experimental class model compared to the model. Because the results of both samples are lower than the L table, they have a high N value, so to find out whether the sample <0.148 is distributed in normal class groups, the researcher conducted an additional normality test using the Liliefors test hypothesis test of 7.42843 is greater than 1.9944 in the table, therefore the alternative hypothesis is accepted and the null hypothesis H0 is rejected Based on the above findings, the problem-based learning model is more effective than the traditional approach. diversity in my country.

A review of multimodal deep learning methods for genomic-enabled prediction in plant breeding.

Deep learning methods have been applied when working to enhance the prediction accuracy of traditional statistical methods in the field of plant breeding. Although deep learning seems to be a promising approach for genomic prediction, it has proven to have some limitations, since its conventional methods fail to leverage all available information. Multimodal deep learning methods aim to improve the predictive power of their unimodal counterparts by introducing several modalities (sources) of input information. In this review, we introduce some theoretical basic concepts of multimodal deep learning and provide a list of the most widely used neural network architectures in deep learning, as well as the available strategies to fuse data from different modalities. We mention some of the available computational resources for the practical implementation of multimodal deep learning problems. We finally performed a review of applications of multimodal deep learning to genomic selection in plant breeding and other related fields. We present a meta-picture of the practical performance of multimodal deep learning methods to highlight how these tools can help address complex problems in the field of plant breeding. We discussed some relevant considerations that researchers should keep in mind when applying multimodal deep learning methods. Multimodal deep learning holds significant potential for various fields, including genomic selection. While multimodal deep learning displays enhanced prediction capabilities over unimodal deep learning and other machine learning methods, it demands more computational resources. Multimodal deep learning effectively captures intermodal interactions, especially when integrating data from different sources. To apply multimodal deep learning in genomic selection, suitable architectures and fusion strategies must be chosen. It is relevant to keep in mind that multimodal deep learning, like unimodal deep learning, is a powerful tool but should be carefully applied. Given its predictive edge over traditional methods, multimodal deep learning is valuable in addressing challenges in plant breeding and food security amid a growing global population.

Synapses learn to utilize stochastic pre-synaptic release for the prediction of postsynaptic dynamics.

Synapses in the brain are highly noisy, which leads to a large trial-by-trial variability. Given how costly synapses are in terms of energy consumption these high levels of noise are surprising. Here we propose that synapses use noise to represent uncertainties about the somatic activity of the postsynaptic neuron. To show this, we developed a mathematical framework, in which the synapse as a whole interacts with the soma of the postsynaptic neuron in a similar way to an agent that is situated and behaves in an uncertain, dynamic environment. This framework suggests that synapses use an implicit internal model of the somatic membrane dynamics that is being updated by a synaptic learning rule, which resembles experimentally well-established LTP/LTD mechanisms. In addition, this approach entails that a synapse utilizes its inherently noisy synaptic release to also encode its uncertainty about the state of the somatic potential. Although each synapse strives for predicting the somatic dynamics of its postsynaptic neuron, we show that the emergent dynamics of many synapses in a neuronal network resolve different learning problems such as pattern classification or closed-loop control in a dynamic environment. Hereby, synapses coordinate themselves to represent and utilize uncertainties on the network level in behaviorally ambiguous situations.

Deephive: A Reinforcement Learning Approach for Automated Discovery of Swarm-Based Optimization Policies

We present an approach for designing swarm-based optimizers for the global optimization of expensive black-box functions. In the proposed approach, the problem of finding efficient optimizers is framed as a reinforcement learning problem, where the goal is to find optimization policies that require a few function evaluations to converge to the global optimum. The state of each particle within the swarm is defined as its current position and function value within a design space, and the particles learn to take favorable actions that maximize the reward, which is based on the final value of the objective function. The proposed approach is tested on 50 benchmark optimization functions and compared to the performance of other global optimization strategies. Furthermore, the generalization capabilities of the trained particles on the four categories of optimization benchmark functions are investigated. The results show superior performance compared to the other optimizers, desired scaling when the dimension of the functions is varied, and acceptable performance even when applied to unseen functions. On a broader scale, the results show promise for the rapid development of domain-specific optimizers.

POSSIBILITIES OF USING PROBLEMED LEARNING IN TEACHING GEOMETRIC CONCEPTS IN PRIMARY GRADES

In this article, an idea is given about the possibilities of using problem-based learning in the formation of geometric concepts in elementary school mathematics education. Problematic education includes such categories as problem, learning problem, problem situation, problem, problem question, problem assignment. Teaching concepts like straight-line, straight-line section, broken line, length of broken line, perimeter of a triangle, perimeter of a rectangle, perimeter of a polygon, which are studied in primary grades, using problem-based learning, allows students to quickly and effectively master these concepts, as well as to be able to apply the acquired knowledge in practice gives.

Weakly supervised label learning flows

Supervised learning usually requires a large amount of labelled data. However, attaining ground-truth labels is costly for many tasks. Alternatively, weakly supervised methods learn with cheap weak signals that only approximately label some data. Many existing weakly supervised learning methods learn a deterministic function that estimates labels given the input data and weak signals. In this paper, we develop label learning flows (LLF), a general framework for weakly supervised learning problems. Our method is a generative model based on normalizing flows. The main idea of LLF is to optimize the conditional likelihoods of all possible labelings of the data within a constrained space defined by weak signals. We develop a training method for LLF that trains the conditional flow inversely and avoids estimating the labels. Once a model is trained, we can make predictions with a sampling algorithm. We apply LLF to three weakly supervised learning problems. Experiment results show that our method outperforms many baselines we compare against.

Kinematic matrix: One-shot human action recognition using kinematic data structure

One-shot action recognition, which refers to recognizing human-performed actions using only a single training example, holds significant promise in advancing video analysis, particularly in domains requiring rapid adaptation to new actions. However, existing algorithms for one-shot action recognition face multiple challenges, including high computational complexity, limited accuracy, and difficulties in generalization to unseen actions. To address these issues, we propose a novel kinematic-based skeleton representation that effectively reduces computational demands while enhancing recognition performance. This representation leverages skeleton locations, velocities, and accelerations to formulate the one-shot action recognition task as a metric learning problem, where a model projects kinematic data into an embedding space. In this space, actions are distinguished based on Euclidean distances, facilitating efficient nearest-neighbour searches among activity reference samples. Our approach not only reduces computational complexity but also achieves higher accuracy and better generalization compared to existing methods. Specifically, our model achieved a validation accuracy of 78.5%, outperforming state-of-the-art methods by 8.66% under comparable training conditions. These findings underscore the potential of our method for practical applications in real-time action recognition systems.

대학 집단상담 분야의 연구 동향 분석

Objectives The purpose of this study is to examine the research trends on group counseling in universities, and to provide implications for future research. Methods For this study, 150 domestic journal papers on university group counseling published from 2000 to April 2024 were analyzed for basic elements, research topics, research subjects, research methods, research types, and types of group counseling. Results The research results are as follows. First, as a result of analysis as a research topic, it was classified into counseling theory, career development, self-development, interpersonal relationships, emotions, subjective well-being and learning development. Second, most of the subjects of the study were general college students, and in addition, there were college students in specific majors such as education and nursing. And college students who had disability problems, psychology, and learning problems were the subjects. Third, as a result of analysis by research method, it was classified into quantitative research, qualitative research, and literature research. Quantitative research included methods such as difference test, variance analysis, and nonparametric test, and qualitative research included content analysis, phenomenological research, grounded theory, and FGI. Fourth, research types were classified into counseling effect research, process research, development research, meta research, and research trend survey. Among them, counseling effect research accounted for the largest proportion. Fifth, as a result of analyzing the types of group counseling programs, face-to-face counseling was the most common in the form of counseling, structured counseling was the most common in the counseling structure, and counselor-centered counseling was the most common in group leaders. Conclusions It was discussed the importance of applying various perspectives and methods when studying group counseling research in universities. In addition, implications for the development and implementation of an effective university group counseling program for college students were presented.

The Investigation of Progress and Application in the Multi-Armed Bandit Algorithm

This article delves deeply into the development process and practical applications of the multi-armed bandit algorithm in the current digital era. With the continuous popularity of online advertising and online learning, information has grown explosively, making decision optimization crucial. The multi-armed bandit algorithm, as a sequential decision-making model, encompasses common algorithms such as the greedy algorithm, -greedy algorithm, UCB algorithm, and Thompson sampling. Its main role is to seek the best balance between exploration and exploitation to solve the fundamental problems in reinforcement learning. The article introduces an internationally released datasets, namely MovieLens, and elaborates in detail a series of indicators for evaluation, including the average number of friends per user, the average number of listened-to artists per user, the average number of movie rating times, the average number of tags added by users, content diversity indicators, and statistics on the differences in click-through rates of recommendations for different types of movies. In addition, the article also presents the specific methods of literature collection, screening, analysis, and review. Its purpose is to understand the multi-armed bandit algorithm more deeply and provide strong guidance for the future development and wide application of this algorithm in various fields.

An Empirical Study of Self-Supervised Learning with Wasserstein Distance

In this study, we consider the problem of self-supervised learning (SSL) utilizing the 1-Wasserstein distance on a tree structure (a.k.a., Tree-Wasserstein distance (TWD)), where TWD is defined as the L1 distance between two tree-embedded vectors. In SSL methods, the cosine similarity is often utilized as an objective function; however, it has not been well studied when utilizing the Wasserstein distance. Training the Wasserstein distance is numerically challenging. Thus, this study empirically investigates a strategy for optimizing the SSL with the Wasserstein distance and finds a stable training procedure. More specifically, we evaluate the combination of two types of TWD (total variation and ClusterTree) and several probability models, including the softmax function, the ArcFace probability model, and simplicial embedding. We propose a simple yet effective Jeffrey divergence-based regularization method to stabilize optimization. Through empirical experiments on STL10, CIFAR10, CIFAR100, and SVHN, we find that a simple combination of the softmax function and TWD can obtain significantly lower results than the standard SimCLR. Moreover, a simple combination of TWD and SimSiam fails to train the model. We find that the model performance depends on the combination of TWD and probability model, and that the Jeffrey divergence regularization helps in model training. Finally, we show that the appropriate combination of the TWD and probability model outperforms cosine similarity-based representation learning.