Learning Perspective Research Articles

Multi-Agent Interactive Localization: A Positive Transfer Learning Perspective

Multi-Agent Interactive Localization: A Positive Transfer Learning Perspective

Negatives Make a Positive: An Embarrassingly Simple Approach to Semi-Supervised Few-Shot Learning.

Semi-Supervised Few-Shot Learning (SSFSL) aims to train a classifier that can adapt to new tasks using limited labeled data and a fixed amount of unlabeled data. Various sophisticated methods have been proposed to tackle the challenges associated with this problem. In this paper, we present a simple but quite effective approach to predict accurate negative pseudo-labels of unlabeled data from an indirect learning perspective. We leverage these pseudo-labels to augment the support set, which is typically limited in few-shot tasks, e.g., 1-shot classification. In such label-constrained scenarios, our approach can offer highly accurate negative pseudo-labels. By iteratively excluding negative pseudo-labels one by one, we ultimately derive a positive pseudo-label for each unlabeled sample in our approach. The integration of negative and positive pseudo-labels complements the limited support set, resulting in significant accuracy improvements for SSFSL. Our approach can be implemented in just few lines of code by only using off-the-shelf operations, yet it outperforms state-of-the-art methods on four benchmark datasets. Furthermore, our approach exhibits good adaptability and generalization capabilities when used as a plug-and-play counterpart alongside existing SSFSL methods and when extended to generalized linear models.

Can lies be faked? Comparing low-stakes and high-stakes deception video datasets from a Machine Learning perspective

Despite the great impact of lies in human societies and a meager 54% human accuracy for Deception Detection (DD), Machine Learning systems that perform automated DD are still not viable for proper application in real-life settings due to data scarcity. Few publicly available DD datasets exist and the creation of new datasets is hindered by the conceptual distinction between low-stakes and high-stakes lies. Theoretically, the two kinds of lies are so distinct that a dataset of one kind could not be used for applications for the other kind. Even though it is easier to acquire data on low-stakes deception since it can be simulated (faked) in controlled settings, these lies do not hold the same significance or depth as genuine high-stakes lies, which are much harder to obtain and hold the practical interest of automated DD systems. To investigate whether this distinction holds true from a practical perspective, we design several experiments comparing a high-stakes DD dataset and a low-stakes DD dataset evaluating their results on a Deep Learning classifier working exclusively from video data. In our experiments, a network trained in low-stakes lies had better accuracy classifying high-stakes deception than low-stakes, although using low-stakes lies as an augmentation strategy for the high-stakes dataset decreased its accuracy.

Resolving the evolutionary paradox of consciousness

AbstractEvolutionary fitness threats and rewards are associated with subjectively unpleasant and pleasant sensations, respectively. Initially, these correlations appear explainable via adaptation by natural selection. But here I analyse the major metaphysical perspectives on consciousness – physicalism, dualism, and panpsychism – and conclude that none help to understand the adaptive-seeming correlations via adaptation. I also argue that a recently proposed explanation, the phenomenal powers view, has major problems that mean it cannot explain the adaptive-seeming correlations via adaptation either. So the mystery – call it the evolutionary paradox of consciousness – remains. Some have used this mystery to argue for non-naturalistic (e.g. theistic) explanations. But I propose a naturalistic, non-adaptive explanation of the adaptive-seeming correlations: namely, ‘sensational associative learning’ during development. In this perspective, pairing of particular sensations with unconditioned stimuli – fitness rewards or threats – cause the sensations themselves to come to be interpreted as good or bad, respectively. Sensations, like colours, that are not reliably paired with either fitness rewards or threats remain largely unvalenced. Sensational associative learning also provides explanations for adaptive-seeming structural aspects of sensations, such as the observation that sounds of different pitch are experienced as ordinal in correspondence to their wavelengths while the same is not true of colours of different hue. The sensational associative learning perspective appears compatible with physicalism, panpsychism, and dualism (though not epiphenomenalism).

Opponent-Awareness-Based Anti-Intelligent Jamming Channel Access Scheme: A Deep Reinforcement Learning Perspective

Opponent-Awareness-Based Anti-Intelligent Jamming Channel Access Scheme: A Deep Reinforcement Learning Perspective

Advancements in Privacy-Preserving Techniques for Federated Learning: A Machine Learning Perspective

Federated learning has emerged as a promising paradigm for collaborative machine learning while preserving data privacy. However, concerns about data privacy remain significant, particularly in scenarios where sensitive information is involved. This paper reviews recent advancements in privacy-preserving techniques for federated learning from a machine learning perspective. It categorizes and analyses state-of-the-art approaches within a unified framework, highlighting their strengths, limitations, and potential applications. By providing insights into the landscape of privacy-preserving federated learning, this review aims to guide researchers and practitioners in developing robust and privacy-conscious machine learning solutions for collaborative environments. The paper concludes with future research directions to address ongoing challenges and further enhance the effectiveness and scalability of privacy-preserving federated learning.

L2 Listening, Word Gain, and Learner Perspectives through Partial Dictation vs Reading-while-listening

Purpose: This study explored how the partial dictation of key words with L1 glossing (PDGL1) and reading-while-listening with the words boldfaced and underlined (RWLBU) while listening to audiovisual TED talks differently influence L2 listening, word gain, and learner perspectives.
 Methods: Fifty-six beginner learners took part in the study who were randomly assigned to the two groups. Data collection consisted of a pretest for selecting target words (TWs), a pretest for listening comprehension, two times of interventions, two immediate posttests for TWs, and a retrospective questionnaire.
 Results: The results showed that overall the PDGL1 group yielded significantly more immediate word recall and delayed retention in receptive and productive knowledge as well as listening comprehension, while RWLBU only contributed for immediate receptive word recall. Both groups showed positive perspectives on word learning and affective factors such as motivation and interest with no significant differences.
 Conclusion: The beneficial results from PDGL1 appeared to be attributed to language-focused partial dictation of key words with L1 glosses, which promotes learners to pay attention to form-meaning connections and cognitive processes. This could encourage beginner learners to be involved in better language performances, resulting in positive perspectives (Dankook University).

Critical Thinking In Learning Perspectives On Educational Philosophy

Students are encouraged to enhance their analytical and evaluative skills through critical thinking, an essential component of education. This publication aims to investigate the significance of critical thinking in the context of learning, with a particular emphasis on educational philosophy views. This article examines the philosophical perspectives underpinning critical thinking, defines the relationship between those perspectives and academic goals, and assesses the impact these perspectives have on pupils' intellectual and moral growth. A method of research that takes a qualitative approach. Examining the ideas of epistemology, logic, and methodology that form the basis of critical thinking is accomplished through applying a philosophical perspective. To provide theoretical contributions to the debate over the function of critical thinking in the educational process, the discussion encompasses perspectives from ancient and modern philosophers. Additionally, the practical consequences of an educational philosophy perspective on the creation of curricula and teaching methods that support the development of critical thinking in students are discussed in this article.

The Application of Contextual Integration in English Language Learning

In the process of social development, language plays a crucial role. It serves not only as a carrier of thought and culture but also as a fundamental tool for communication and exchange among individuals. In the journey of learning English, the integration of context can effectively aid students in understanding and mastering the language’s meaning, facilitating accurate translation of relevant sentences. From the perspective of contemporary English language learning, context integration can better elucidate the relationships within language contexts, encompassing factors such as time, space, and objects. Therefore, context integration holds significant practical importance in English language acquisition. It not only enhances learning efficiency but also enables students to accurately comprehend the meaning conveyed in English texts, thus averting translation errors and communication barriers. With this in mind, this paper offers a brief analysis of the application of context integration in English language learning, aiming to provide readers with valuable insights.

Analisis efektivitas ship management untuk optimalisasi kinerja di era industry 5.0

In supporting the success of a project in a company, management issues often become an important factor in implementation in the field. Likewise, PT PAL Marine Service (PT PALMARS) operates in Surabaya, East Java. This company operates in the service sector and offers several services to support the company's progress, one of which is ship management. Therefore, to analyze the effectiveness of ship management in optimizing organizational performance in the Industrial Era 5.0, a case study was carried out at PT. PAL Marine Service (PT PALMARS) in Surabaya City, East Java. In this analysis, we will focus on four important perspectives that can influence the effectiveness of ship management through the balanced scorecard method. The independent variables consist of a customer perspective, an internal business process perspective, and a learning and growth perspective. The dependent variable is performance optimization. The sample or key research informants are the majority of lines involved in ship management activities: the Secretary, Marketing Department, Department of Accounting and Tax, Department of Finance and HR, M2KL Department, Operations, Engineering, and Procurement Departments, and the ship's crew. Data collection techniques used to measure organizational effectiveness are case study techniques, surveys/observations, data analysis, literature reviews, documentation, and interviews. The data analysis stages are data collection, reduction, display, and conclusion.

Machine Learning in Petrology: State-of-the-Art and Future Perspectives

Abstract This article reports on the state-of-the-art and future perspectives of machine learning (ML) in petrology. To achieve this goal, it first introduces the basics of ML, including definitions, core concepts, and applications. Then, it starts reviewing the state-of-the-art of ML in petrology. Established applications mainly concern the so-called data-driven discovery and involve specific tasks like clustering, dimensionality reduction, classification, and regression. Among them, clustering and dimensionality reduction have been demonstrated to be valuable for decoding the chemical record stored in igneous and metamorphic phases and to enhance data visualization, respectively. Classification and regression tasks find applications, for example, in petrotectonic discrimination and geo-thermobarometry, respectively. The main core of the manuscript consists of depicting emerging trends and the future directions of ML in petrological investigations. I propose a future scenario where ML methods will progressively integrate and support established petrological methods in automating time-consuming and repetitive tasks, improving current models, and boosting discovery. In this framework, promising applications include (1) the acquisition of new multimodal petrologic data; (2) the development of data fusion techniques, physics-informed ML models, and ML-supported numerical simulations; and (3) the continuous exploration of the ML potential in petrology. To boost the contribution of ML in petrology, our main challenges are: (1) to improve the ability of ML models to capture the complexity of petrologic processes, (2) progressively link ML algorithms with the physical and thermodynamic nature of the investigated problems, and (3) to start a collaborative effort among researchers coming from different disciplines, both in research and teaching.

Getting Students Interested in Psychological Measurement by Experiencing the Barnum Effect

Background: Fake psychological tests tend to be viewed as completely believable (Barnum effect), meaning psychometric properties cannot be judged subjectively. Experiencing this effect first-hand could help get students interested in the science of psychological assessment. Objective: In a blended learning perspective, we created materials to elicit a Barnum effect with a brief online fake test and tested the benefits of this procedure for student engagement. Method: Five cohorts of undergraduate psychology students (total N = 2,269) completed the fake test online prior to their first psychometrics course. We collected single-item ratings of perceived effectiveness for the test, willingness to use it, and course engagement (interest and enthusiasm for the course). We also retrieved grades. Results: Eighty-five percent of students rated the effectiveness of the fake test above the neutral point; 77% were willing to use it as future psychologists. Experiencing the effect was related to significantly increased interest and enthusiasm, and higher grades. Conclusion: Having students experience the Barnum effect for themselves is a useful and effortless approach to get them interested in psychometrics. Teaching Implications: Teachers of introductory courses related to psychological measurement should consider having all students experience the Barnum effect. All materials are provided.

Mapping the Research Landscape of Industry 5.0 from a Machine Learning and Big Data Analytics Perspective: A Bibliometric Approach

Over the past years, machine learning and big data analysis have emerged, starting as a scientific and fictional domain, very interesting but difficult to test, and becoming one of the most powerful tools that is part of Industry 5.0 and has a significant impact on sustainable, resilient manufacturing. This has garnered increasing attention within scholarly circles due to its applicability in various domains. The scope of the article is to perform an exhaustive bibliometric analysis of existing papers that belong to machine learning and big data, pointing out the capability from a scientific point of view, explaining the usability of applications, and identifying which is the actual in a continually changing domain. In this context, the present paper aims to discuss the research landscape associated with the use of machine learning and big data analysis in Industry 5.0 in terms of themes, authors, citations, preferred journals, research networks, and collaborations. The initial part of the analysis focuses on the latest trends and how researchers lend a helping hand to change preconceptions about machine learning. The annual growth rate is 123.69%, which is considerable for such a short period, and it requires a comprehensive analysis to check the boom of articles in this domain. Further, the exploration investigates affiliated academic institutions, influential publications, journals, key contributors, and most delineative authors. To accomplish this, a dataset has been created containing researchers’ papers extracted from the ISI Web of Science database using keywords associated with machine learning and big data, starting in 2016 and ending in 2023. The paper incorporates graphs, which describe the most relevant authors, academic institutions, annual publications, country collaborations, and the most used words. The paper ends with a review of the globally most cited documents, describing the importance of machine learning and big data in Industry 5.0.

The school is also samba: art teaching as an experience and construction of a decolonial learning process

This article discusses a teaching experience, in the teaching of Art, with students in the 9th year of Elementary School II, at the Escola Municipal Professor Waldson José Bastos Pinheiro, in the community of Vale Dourado, in the neighborhood of Nossa Senhora da Presentação, a school from the public network in Natal (RN). It analyzes Samba as an object of study from the perspective of decolonial learning, identifying how African dances influenced Brazilian culture and, in particular, the theme in question. It problematizes the political-social ills inherited from colonialist exploitation in Brazil and points to the legitimization of cultural manifestations of African origin as actions of resistance and construction of knowledge and learning. The theoretical basis with which we work is based on Freirean pedagogy (FREIRE, 1998); as well as, from the perspective of resuming the “triangular approach” (BARBOSA, 2010) as an epistemological basis. For reflections regarding the teaching of dance practice, Vieira (2007) and Marques (2003) are used. In the historical-social contextualization of the theme in an anthropological and sociological perspective of the Brazilian body, the African rhythm and diversity, the materiality of this rhythm and its relationships, the basis is sought in Sodré (1998).

DiffSED: Sound Event Detection with Denoising Diffusion

Sound Event Detection (SED) aims to predict the temporal boundaries of all the events of interest and their class labels, given an unconstrained audio sample. Taking either the split-and-classify (i.e., frame-level) strategy or the more principled event-level modeling approach, all existing methods consider the SED problem from the discriminative learning perspective. In this work, we reformulate the SED problem by taking a generative learning perspective. Specifically, we aim to generate sound temporal boundaries from noisy proposals in a denoising diffusion process, conditioned on a target audio sample. During training, our model learns to reverse the noising process by converting noisy latent queries to the ground-truth versions in the elegant Transformer decoder framework. Doing so enables the model generate accurate event boundaries from even noisy queries during inference. Extensive experiments on the Urban-SED and EPIC-Sounds datasets demonstrate that our model significantly outperforms existing alternatives, with 40+% faster convergence in training. Code: https://github.com/Surrey-UPLab/DiffSED

Protein 3D Graph Structure Learning for Robust Structure-Based Protein Property Prediction

Protein structure-based property prediction has emerged as a promising approach for various biological tasks, such as protein function prediction and sub-cellular location estimation. The existing methods highly rely on experimental protein structure data and fail in scenarios where these data are unavailable. Predicted protein structures from AI tools (e.g., AlphaFold2) were utilized as alternatives. However, we observed that current practices, which simply employ accurately predicted structures during inference, suffer from notable degradation in prediction accuracy. While similar phenomena have been extensively studied in general fields (e.g., Computer Vision) as model robustness, their impact on protein property prediction remains unexplored. In this paper, we first investigate the reason behind the performance decrease when utilizing predicted structures, attributing it to the structure embedding bias from the perspective of structure representation learning. To study this problem, we identify a Protein 3D Graph Structure Learning Problem for Robust Protein Property Prediction (PGSL-RP3), collect benchmark datasets, and present a protein Structure embedding Alignment Optimization framework (SAO) to mitigate the problem of structure embedding bias between the predicted and experimental protein structures. Extensive experiments have shown that our framework is model-agnostic and effective in improving the property prediction of both predicted structures and experimental structures.

Towards Evidential and Class Separable Open Set Object Detection

Detecting in open-world scenarios poses a formidable challenge for models intended for real-world deployment. The advanced closed set object detectors achieve impressive performance under the closed set setting, but often produce overconfident misprediction on unknown objects due to the lack of supervision. In this paper, we propose a novel Evidential Object Detector (EOD) to formulate the Open Set Object Detection (OSOD) problem from the perspective of Evidential Deep Learning (EDL) theory, which quantifies classification uncertainty by placing the Dirichlet Prior over the categorical distribution parameters. The task-specific customized evidential framework, equipped with meticulously designed model architecture and loss function, effectively bridges the gap between EDL theory and detection tasks. Moreover, we utilize contrastive learning as an implicit means of evidential regularization and to encourage the class separation in the latent space. Alongside, we innovatively model the background uncertainty to further improve the unknown discovery ability. Extensive experiments on benchmark datasets demonstrate the outperformance of the proposed method over existing ones.

Unlocking the Power of Open Set: A New Perspective for Open-Set Noisy Label Learning

Learning from noisy data has attracted much attention, where most methods focus on closed-set label noise. However, a more common scenario in the real world is the presence of both open-set and closed-set noise. Existing methods typically identify and handle these two types of label noise separately by designing a specific strategy for each type. However, in many real-world scenarios, it would be challenging to identify open-set examples, especially when the dataset has been severely corrupted. Unlike the previous works, we explore how models behave when faced with open-set examples, and find that a part of open-set examples gradually get integrated into certain known classes, which is beneficial for the separation among known classes. Motivated by the phenomenon, we propose a novel two-step contrastive learning method CECL (Class Expansion Contrastive Learning) which aims to deal with both types of label noise by exploiting the useful information of open-set examples. Specifically, we incorporate some open-set examples into closed-set classes to enhance performance while treating others as delimiters to improve representative ability. Extensive experiments on synthetic and real-world datasets with diverse label noise demonstrate the effectiveness of CECL.

Rethinking Causal Relationships Learning in Graph Neural Networks

Graph Neural Networks (GNNs) demonstrate their significance by effectively modeling complex interrelationships within graph-structured data. To enhance the credibility and robustness of GNNs, it becomes exceptionally crucial to bolster their ability to capture causal relationships. However, despite recent advancements that have indeed strengthened GNNs from a causal learning perspective, conducting an in-depth analysis specifically targeting the causal modeling prowess of GNNs remains an unresolved issue. In order to comprehensively analyze various GNN models from a causal learning perspective, we constructed an artificially synthesized dataset with known and controllable causal relationships between data and labels. The rationality of the generated data is further ensured through theoretical foundations. Drawing insights from analyses conducted using our dataset, we introduce a lightweight and highly adaptable GNN module designed to strengthen GNNs' causal learning capabilities across a diverse range of tasks. Through a series of experiments conducted on both synthetic datasets and other real-world datasets, we empirically validate the effectiveness of the proposed module. The codes are available at https://github.com/yaoyao-yaoyao-cell/CRCG.

A multimodal model for predicting feedback position and type during conversation

This study investigates conversational feedback, that is, a listener's reaction in response to a speaker, a phenomenon which occurs in all natural interactions. Feedback depends on the main speaker's productions and in return supports the elaboration of the interaction. As a consequence, feedback production has a direct impact on the quality of the interaction.This paper examines all types of feedback, from generic to specific feedback, the latter of which has received less attention in the literature. We also present a fine-grained labeling system introducing two sub-types of specific feedback: positive/negative and given/new. Following a literature review on linguistic and machine learning perspectives highlighting the main issues in feedback prediction, we present a model based on a set of multimodal features which predicts the possible position of feedback and its type. This computational model makes it possible to precisely identify the different features in the speaker's production (morpho-syntactic, prosodic and mimo-gestural) which play a role in triggering feedback from the listener; the model also evaluates their relative importance.The main contribution of this study is twofold: we sought to improve 1/ the model's performance in comparison with other approaches relying on a small set of features, and 2/ the model's interpretability, in particular by investigating feature importance. By integrating all the different modalities as well as high-level features, our model is uniquely positioned to be applied to French corpora.