A transferable framework for cross-city forecasting via invariant structural knowledge learning

The study investigated learning of mathematical structure. Eighth grade students (N = 87) were assigned randomly to read either a programmed text on probability (experimental group) or one on prime numbers (control group). The subject matter structure of the probability text was mapped with the method of directed graphs. Structure in students' memories, cognitive structures, was investigated using a word association technique. Cognitive structure and achievement data were gathered at pretest, posttest, and retention test. The directed graphs provided an interpretable map of subject matter structure. Experimental students learned and retained the content structure but the control students did not. A comparison of word association, achievement, and attitude data indicated that learning of structure may differ from learning measured by achievement tests.

Bayesian networks are useful analytical models for designing the structure of knowledge in machine learning. Bayesian networks can represent probabilistic dependency relationships among the variables. One strategy of Bayesian Networks structure learning is the score and search technique. The authors present the Elephant Swarm Water Search Algorithm (ESWSA) as a novel approach to Bayesian network structure learning. In the algorithm; Deleting, Reversing, Inserting, and Moving are used to make the ESWSA for reaching the optimal structure solution. Mainly, water search strategy of elephants during drought periods is used in the ESWSA algorithm. The proposed method is compared with simulated annealing and greedy search using BDe score function. The authors have also investigated the confusion matrix performances of these techniques utilizing various benchmark data sets. As presented by the results of the evaluations, the proposed algorithm has better performance than the other algorithms and produces better scores and accuracy values.

Bayesian network (BN) is the main model for solving probabilistic inference problem. The current research of BN just focuses on structure learning or parameter learning. And it still lacked algorithm which can implement the consistence of structure building, parameter learning and knowledge inference, so that knowledge construction and application are not associated and cannot be verified. This paper proposes a novel approach of Bayesian network adaptive knowledge construction and inference which is based on genetic algorithm. This algorithm, which designs a new BN learning encodement, crossover and mutation operators with adjusted strategies and the fitness function with inferential error feedback, implements BN building in all processes of structure building, parameter learning, knowledge inference and feedback revise with samples supported. Results show that the new approach not only can optimize the structure and parameters learning synchronously, but also can revise inferential error adaptively, and has more satisfied and accurate inference result.

Background: With the discussion of fairness in Machine Learning (ML) gaining traction in recent years, the idea of viewing fairness through the causal lens has become prominent. The main idea behind this is that by looking at the causal structure underlying the data used for an ML model, we can see and evaluate more concisely which influences of the sensitive variables on the target variable are problematic and how they are problematic. Doing so allows not only a nuanced view of fairness and an informed choice of fairness measures but also more targeted approaches (such as path-specific bias mitigation) to handle fairness issues. Objectives: Mainly, two important points have hindered the practical use of the causal lens and causality-based bias mitigation. First, a classification of different graphical structures with different fairness implications involving a sensitive variable and a target variable is still missing, as is a discussion of how different contexts can shape our evaluation of fairness. Second, the construction of such graphical models is not trivial and error-prone. However, recent work showed that combining background knowledge and data-driven network structure learning may lead to more accurate graphs. In this work, we attempt to address and tackle these two practical shortcomings. Methods: Our first contribution is a classification and discussion of causal structures with different fairness implications and how contexts shape our assessment. Our second contribution is an advancement in learning more accurate graphs by adapting structure learning algorithms, and a detailed evaluation of graph correctness and subsequent fairness implications. Results: We show that when including background knowledge naturally available in fairness settings, graph learning becomes more accurate, which also has positive implications for accurate fairness assessments. Conclusions: Our work may pave the way for a broader adoption of causal ML fairness by providing concrete suggestions about the implications of causal structures and contexts, and learning more accurate graphs. We also address current limitations and highlight the need for stakeholder inclusion.

Low-level texture feature/knowledge is also of vital importance for characterizing the local structural pattern and global statistical properties, such as boundary, smoothness, regularity, and color contrast, which may not be well addressed by high-level deep features. In this paper, we aim to re-emphasize the low-level texture information in deep networks for semantic segmentation and related knowledge distillation tasks. To this end, we take full advantage of both structural and statistical texture knowledge and propose a novel Structural and Statistical Texture Knowledge Distillation (SSTKD) framework for semantic segmentation. Specifically, Contourlet Decomposition Module (CDM) is introduced to decompose the low-level features with iterative Laplacian pyramid and directional filter bank to mine the structural texture knowledge, and Texture Intensity Equalization Module (TIEM) is designed to extract and enhance the statistical texture knowledge with the corresponding Quantization Congruence Loss (QDL). Moreover, we propose the Co-occurrence TIEM (C-TIEM) and generic segmentation frameworks, namely STLNet++ and U-SSNet, to enable existing segmentation networks to harvest the structural and statistical texture information more effectively. Extensive experimental results on three segmentation tasks demonstrate the effectiveness of the proposed methods and their state-of-the-art performance on seven popular benchmark datasets, respectively.

The main goal of this paper is the presentation of a new GReedy knowledge Acquisition Procedure (GRAP) for rapid prototyping of knowledge structures (KS) or spaces. The classical knowledge acquisition method for this [2] is even for domain experts cognitive demanding and computational complex. GRAP interactively generates an online knowledge acquisition schedule so that experts only have to provide simple nonredundant judgements about the (learning / cognitive) precedence in pairs of (learning / cognitive) objects. From these data GRAP generates a Hasse diagram of the surmise relation from which the knowledge structures and optimal user-adaptive learning paths can be derived. In a case-study we developed with three expert software engineers a knowledge structure and optimal learning paths for 23 software design patterns within a few hours.

This study investigates effects of form-focused instructions on the learning of English verb complementation, especially ‘verb+NP+to infinitive’ construction, by Korean EFL learners. The seven target verbs were chosen from monotranstive and ditransitive constructions. 150 college students were randomly divided into two experimental groups, which received negative feedback plus explicit rule presentation (NF) and input enhancement plus meaning-oriented rule search (IE) respectively, and one control group. A pretest and a posttest were conducted to measure learners’ receptive and productive knowledge of the target structure through a grammaticality judgment task (GJT) and a picture description task (PDT). The results of the posttest revealed that experimental groups outperformed the control group in both tasks. In the grammaticality judgment task, the NF group showed better learning in rejecting incorrect structures as well as accepting target structures when compared to the IE group. The NF group also outperformed in reducing error tokens in the picture description task. The findings indicate that two types of form-focused instructions implemented in the study have advantages in facilitating learning of the target structure. (174)

Knowledge Tracing (KT) aims to model students' evolving knowledge states and predict their future performance. Traditional sequential models often struggle with long-range dependencies due to inefficient memory retention and high computational costs, limiting their effectiveness in capturing students' continuous learning trajectories. Meanwhile, graph-based KT models effectively represent knowledge concept relationships but are not designed for sequential modeling. To address these challenges, we propose DualFlowKT, a novel model that employs a dual-path architecture, integrating Graph Neural Networks (GNNs) for structured knowledge modeling and the Mamba state space model for efficient long-sequence learning. Specifically, the graph-based path extracts relational dependencies among knowledge concepts, while the sequential path leverages Mamba's selective state-space modeling to effectively capture long-range learning patterns with enhanced memory efficiency and reduced computational overhead. To further enhance input representations, we introduce a Feature Enhancement Module (FEM) that extracts critical learning features such as concept continuity, learning rate, and adaptive difficulty. Additionally, a gated fusion mechanism dynamically integrates knowledge structure representations from GNNs with sequential dependencies captured by Mamba, allowing the model to adaptively balance structural and temporal learning information. Extensive experiments on multiple KT benchmark datasets demonstrate that DualFlowKT outperforms existing methods, particularly in handling long-sequence interactions and complex knowledge structures. Furthermore, the model provides deeper insights into students' evolving knowledge states, supporting personalized education systems.

Scholars spanning the social sciences and humanities wrestle with the complex and often contested meanings of race, racism, and discrimination. In all of this enterprise, sociologists rightly retain a special claim to illuminating processes of group boundary maintenance, systems of racial inequality and supporting ideologies, and attendant patterns of intergroup behavior (Jackman 1994; Lamont 2000). Mainstream sociological research, however, has focused principally on the structural manifestations of race, racism, and discrimination, particularly as they characterize black-white relations (Wilson 1978). Sociologists have made signal contributions to the understanding of modern ghetto joblessness and poverty (Wilson 1996), of racial residential segregation (Massey and Denton 1993), and of fundamental disparities in accumulated wealth (Oliver and Shapiro 1995). In some critical respects this work has expanded to include multiracial and multiethnic comparisons with respect to both key economic (Lichter and Oliver 2000; Smith 2001; Waldinger 1996) and residential outcomes (Charles 2001; Emerson, Yancey, and Chai 2001). To a surprising degree, however, the micro social processes necessarily embedded in these structural analyses are still largely unaddressed. Yet the basic social processes invoked by the terms race, racism, and discrimination are quintessentially social psychological phenomena; sociologists ignore or downplay this basic insight at the discipline's peril. These concepts concern the meanings of social groupings and how those meanings come to guide patterns of relations among individuals recognized as members of particular groups. They immediately entail the labeling and social learning of group categories, identity, feelings, beliefs, and related cognitive structures. These factors, in turn, are expressed in lines of interaction and behavior that flow from, reinforce and reconstitute, or come to transform those social categorizations. In addition, such categorizations have direct implications for the structure and basic conditions of social organization. That is, race,1 racism,2 and discrimination3 are also, and perhaps most fundamentally, bases and mechanisms of hierarchical differentiation that shape the ordering of social relations as well as the allocation of life experiences and life chances (Zuberi 2001a).

Recognition of multi-interval rules in dataset with continuous-valued attributes

The purpose of the present study was to further clarify the role of stimulus exposure time as a variable which affects responses to projective material. Presentation of the standard Rorschach cards over a series of exposure times, ranging from .03 sec. to unlimited exposure, constituted a systematically controlled range of perceptual conditions with stimuli known to have rich and useful personality correlates. In an earlier tachistoscopic smdy Stein (1949) demonstrated different patterns of response as a function of exposure time of Rorschach cards; however, the small sample used (17 Ss who responded under ascending exposure order) limited the generality of his results, and no evaluation of perceptual modes, independent of the experimental conditions, was reported. Stein's study explored the clinical usefulness of a tachistoscopic Rorschach technique by applying it to the analysis of several individual cases. A later study by Stein and Meer (1954) pointed to the usefulness of the tachistoscopic Rorschach technique in discriminating between Ss of high and low creative ability. In the view of chis author, the experimental design used in the present study invites consideration of what perceptual processes may accompany the physically manipulated conditions of minimal and sustained exposure of the stimulus. When an external stimulus is exposed very briefly, the opportunity for the perceiver to base his reactions on an interaction with the stimulus is relatively limited, and he is thereby forced to rely largely upon subjective factors and his immediate, brief perception of the stimulus. Some of the contributing factors which the perceiver may bring to the experimental situation are those of cognitive structure, personality structure, motivational states, and past learning. As exposure time of the external stimulus is increased, more time is available during which interaction between perceiver and stimulus can occur. As there is increased opportunity to investigate the stimulus at slower tachistoscopic speeds, more information can be obtained from the sc~mulus, and there is more time for corrections in perception. While the experimental situation thus varies systematically in the opportunity for interaction, it does not of course follow that all Ss will respond equally to the increased possibilities for interaction with the stimulus at longer exposure times. The present article will be limited to a report and discussion of the more formal aspects of Ror'This article is based upon a doctoral dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the School of Education of New York University, 1956. The author is indebted to Professor John 1. Sullivan for his direction of the research. An abridged account of the study was presented at the Sixty-Sixth Annual Convention of the American Psychological Association, 1958.

Transfer of abstract structural knowledge aids new concept learning in humans and artificial neural networks

The World in a Book: Al-Nuwayri an the Islamic Encyclopedic Tradition

The increasing spread of Internet technology has highlighted the need for a better understanding of the fundamental issues concerning human users in a virtual space. Despite the great degree of navigational freedom, however, not all hypermedia users have the capability to locate information or assimilate internal knowledge. Research findings suggest that this type of problem could be solved if users were able to hold a cognitive overview of the hypermedia structure. How a learner can acquire the correct structural knowledge of online information has become an important factor in learning performance in a hypermedia environment. Variables that might influence learners' abilities in structuring a cognitive overview, such as users' cognitive styles and the different ways of representing information, should be carefully taken into account. The results of this study show that the interactions between information representation approaches and learners' cognitive styles have significant effects on learners' performance in terms of structural knowledge and feelings of disorientation. Learners' performance could decline if a representational approach that contradicts their cognitive style is used. Finally, the results of the present study may apply only when the learner's knowledge level is in the introductory stage. It is not clear how and what type of cognitive styles, as well as information representation approaches, will affect the performance of advanced and expert learners.

Salespeople knowledge search behavior and sales performance: An investigation of printing equipment industry