检索结果-内蒙古大学图书馆

Word2Minecraft: Generating 3D Game Levels through Large Language Models

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Huang, Shuo Nasir, Muhammad Umair James, Steven Togelius, Julian Department of Computer Science and Engineering Tandon School of Engineering New York University New York10012 United States School of Computer Science and Applied Mathematics University of the Witwatersrand South Africa

We present Word2Minecraft, a system that leverages large language models to generate playable game levels in Minecraft based on structured stories. The system transforms narrative elements—such as protagonist goals, antagonist challenges, and environmental settings—into game levels with both spatial and gameplay constraints. We introduce a flexible framework that allows for the customization of story complexity, enabling dynamic level generation. The system employs a scaling algorithm to maintain spatial consistency while adapting key game elements. We evaluate Word2Minecraft using both metric-based and human-based methods. Our results show that GPT-4Turbo outperforms GPT-4o-Mini in most areas, including story coherence and objective enjoyment, while the latter excels in aesthetic appeal. We also demonstrate the system’s ability to generate levels with high map enjoyment, offering a promising step forward in the intersection of story generation and game design. We open-source the code at https://***/JMZkk/Word2Minecraft/tree/word2mc_v0 Copyright © 2025, The Authors. All rights reserved.

关键词：

Number of Subgraphs and Their Converses in Tournaments and New Digraph Polynomials

学校读者我要写书评

暂无评论

Journal of Graph Theory 2025年

作者： Ai, Jiangdong Gutin, Gregory Lei, Hui Yeo, Anders Zhou, Yacong School of Mathematical Sciences and LPMC Nankai University Tianjin China Department of Computer Science Royal Holloway University of London Surrey United Kingdom School of Statistics and Data Science LPMC and KLMDASR Nankai University Tianjin China Department of Mathematics and Computer Science University of Southern Denmark Odense Denmark Department of Mathematics and Applied Mathematics University of Johannesburg Auckland Park South Africa

An oriented graph (Formula presented.) is converse invariant if, for any tournament (Formula presented.), the number of copies of (Formula presented.) in (Formula presented.) is equal to that of its converse (Formula presented.). El Sahili and Ghazo Hanna [J. Graph Theory 102 (2023), 684-701] showed that any oriented graph (Formula presented.) with maximum degree at most 2 is converse invariant. They proposed a question: Can we characterize all converse invariant oriented graphs? In this paper, we introduce a digraph polynomial and employ it to give a necessary condition for an oriented graph to be converse invariant. This polynomial serves as a cornerstone in proving all the results presented in this paper. In particular, we characterize all orientations of trees with diameter at most 3 that are converse invariant. We also show that all orientations of regular graphs are not converse invariant if (Formula presented.) and (Formula presented.) have different degree sequences. In addition, in contrast to the findings of El Sahili and Ghazo Hanna, we prove that every connected graph (Formula presented.) with maximum degree at least 3, admits an orientation (Formula presented.) of (Formula presented.) such that (Formula presented.) is not converse invariant. We pose a new conjecture. © 2025 The Author(s). Journal of Graph Theory published by Wiley Periodicals LLC.

关键词： Polynomials

Normal approximation for number of edges in random intersection graphs

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Rybarczyk, Katarzyna Serafin, Grzegorz Faculty of Mathematics and Computer Science Adam Mickiewicz University Poznan Poland Faculty of Pure and Applied Mathematics Wroclaw University of Science and Technology Ul. Wybrzeze Wyspiańskiego 27 Wroclaw Poland

The random intersection graph model G(n,m,p) is considered. Due to substantial edge dependencies, studying even fundamental statistics such as the subgraph count is significantly more challenging than in the classical binomial model G(n,p). First, we establish normal approximation bound in both the Wasserstein and the Kolmogorov distances for a class of local statistics on G(n,m,p). Next, we apply these results to derive such bounds for the standardised number of edges, and determine the necessary and sufficient conditions for its asymptotic normality. We develop a new method that provides a combinatorial interpretation and facilitates the estimation of analytical expressions related to general distance bounds. In particular, this allows us to control the behaviour of central moments of subgraph existence indicators. The presented method can also be extended to count copies of subgraphs larger than a single *** Codes 05C80, 60F05, 05C82 © 2025, CC BY.

关键词： Graph theory

NEURO-SYMBOLIC AI FOR ANALYTICAL SOLUTIONS OF DIFFERENTIAL EQUATIONS

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Oikonomou, Orestis Lingsch, Levi Grund, Dana Mishra, Siddhartha Kissas, Georgios Department of Computer Science University of Zurich Switzerland Seminar for Applied Mathematics Switzerland ETH Zurich Switzerland ETH AI Center Switzerland

Analytical solutions of differential equations offer exact insights into fundamental behaviors of physical processes. Their application, however, is limited as finding these solutions is difficult. To overcome this limitation, we combine two key insights. First, constructing an analytical solution requires a composition of foundational solution components. Second, iterative solvers define parameterized function spaces with constraint-based updates. Our approach merges compositional differential equation solution techniques with iterative refinement by using formal grammars, building a rich space of candidate solutions that are embedded into a low-dimensional (continuous) latent manifold for probabilistic exploration. This integration unifies numerical and symbolic differential equation solvers via a neuro-symbolic AI framework to find analytical solutions of a wide variety of differential equations. By systematically constructing candidate expressions and applying constraint-based refinement, we overcome longstanding barriers to extract such closed-form solutions. We illustrate advantages over commercial solvers, symbolic methods, and approximate neural networks on a diverse set of problems, demonstrating both generality and accuracy. © 2025, CC BY-NC-SA.

关键词： Iterative methods

Availability of Perfect Decomposition in Statistical Linkage Learning for Unitation-based Function Concatenations

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Prusik, Michal Frej, Bartosz Przewozniczek, Michal W. Faculty of Pure and Applied Mathematics Wroclaw University of Science and Technology Wroclaw Poland Departament of Systems and Computer Networks Wroclaw University of Science and Technology Wroclaw Poland

Statistical Linkage Learning (SLL) is a part of many state-of-the-art optimizers. The purpose of SLL is to discover variable interdependencies. It has been shown that the effectiveness of SLL-using optimizers is highly dependent on the quality of SLL-based problem decomposition. Thus, understanding what kind of problems are hard or easy to decompose by SLL is important for practice. In this work, we analytically estimate the size of a population sufficient for obtaining a perfect decomposition in case of concatenations of certain unitation-based functions. The experimental study confirms the accuracy of the proposed estimate. Finally, using the proposed estimate, we identify those problem types that may be considered hard for SLL-using optimizers. © 2025, CC BY-NC-ND.

关键词： Adversarial machine learning

On the Effectiveness of Random Weights in Graph Neural Networks

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Bui, Thu Schönlieb, Carola-Bibiane Ribeiro, Bruno Bevilacqua, Beatrice Eliasof, Moshe Department of Computer Science Purdue University West LafayetteIN United States Department of Applied Mathematics University of Cambridge Cambridge United Kingdom

Graph Neural Networks (GNNs) have achieved remarkable success across diverse tasks on graph-structured data, primarily through the use of learned weights in message passing layers. In this paper, we demonstrate that random weights can be surprisingly effective, achieving performance comparable to end-to-end training counterparts, across various tasks and datasets. Specifically, we show that by replacing learnable weights with random weights, GNNs can retain strong predictive power, while significantly reducing training time by up to 6× and memory usage by up to 3×. Moreover, the random weights combined with our construction yield random graph propagation operators, which we show to reduce the problem of feature rank collapse in GNNs. These understandings and empirical results highlight random weights as a lightweight and efficient alternative, offering a compelling perspective on the design and training of GNN architectures. © 2025, CC BY.

关键词： Graph neural networks

Optimizing Fog Computing Performance: A Machine Learning-Based Task Scheduling Approach Using Fuzzy Logic and Clustering 1

学校读者我要写书评

暂无评论

Optimizing Fog Computing Performance: A Machine Learning-Bas...

1st International Conference on Advances in computer science, Electrical, Electronics, and Communication Technologies, CE2CT 2025

作者： Dbritto, Ceres Lopes, Joel Oli, Sanjay Joshi, Maulik S Singh, Aniket Rayal, Ashish St Louis United States Redmond United States Dayananda Sagar College of Engineering Department of Mathematics Bangalore560078 India Gujarat Power Engineering and Research Institute GTU Department of Humanity and Science Gujarat India Veer Madho Singh Bhandari Uttarakhand Technical University Department of Computer Science and Engineering Pithoragarh Campus India School of Applied and Life Sciences Uttaranchal University Department of Mathematics Dehradun248007 India

ISBN: (纸本)9798331518578

The ability to deliver results promptly has become crucial for analytical frameworks due to the swift increase in data volumes. Complex algorithms are needed to schedule data processing activities on distributed compute clusters efficiently. Because it is impractical to create and fine-tune a scheduling policy for every task, current systems rely on basic, generic heuristics and neglect workload peculiarities. Effective task scheduling is essential for optimising the performance of fog computing systems. Fog nodes are dynamic, heterogeneous, and resource-constrained, and traditional task scheduling techniques, which were largely created for centralised cloud settings, frequently cannot accommodate these characteristics. We provide a cutting-edge machine learning (ML) based solution that adjusts job assignment to the evolving needs of the Fog environment in order to get beyond these restrictions. Fuzzy logic, a reliable unsupervised learning method, is used with the K-Means clustering algorithm in our method to effectively group. We have used Python to model and assess our suggested method. The simulation findings validate the effectiveness of our scheduling method, demonstrating notable improvements in response time, reduced network usage, and execution time reduction in comparison to other ML and non-ML based scheduling techniques available in the Python framework. © 2025 IEEE.

关键词： Fog computing

Investigating the Impact of Language-Adaptive Fine-Tuning on Sentiment Analysis in Hausa Language Using AfriBERTa

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Sani, Sani Abdullahi Muhammad, Shamsuddeen Hassan Jarvis, Devon School of Computer Science and Applied Mathematics University of the Witwatersrand Johannesburg South Africa Department of Computing Imperial College London United Kingdom

Sentiment analysis (SA) plays a vital role in Natural Language Processing (NLP) by identifying sentiments expressed in text. Although significant advances have been made in SA for widely spoken languages, low-resource languages such as Hausa face unique challenges, primarily due to a lack of digital resources. This study investigates the effectiveness of Language-Adaptive Fine-Tuning (LAFT) to improve SA performance in Hausa. We first curate a diverse, unlabeled corpus to expand the model’s linguistic capabilities, followed by applying LAFT to adapt AfriBERTa specifically to the nuances of the Hausa language. The adapted model is then fine-tuned on the labeled NaijaSenti sentiment dataset to evaluate its performance. Our findings demonstrate that LAFT gives modest improvements, which may be attributed to the use of formal Hausa text rather than informal social media data. Nevertheless, the pre-trained AfriBERTa model significantly outperformed models not specifically trained on Hausa, highlighting the importance of using pre-trained models in low-resource contexts. This research emphasizes the necessity for diverse data sources to advance NLP applications for low-resource African languages. We published the code and the dataset to encourage further research and facilitate reproducibility in low-resource NLP here SA for LowRes Language Copyright © 2025, The Authors. All rights reserved.

关键词： Natural language processing systems

Global well-posedness and polynomial stability for the chemotaxis-fluid systems in Marcinkiewicz spaces

学校读者我要写书评

暂无评论

Computational and applied mathematics 2025年第6期44卷 1-16页

作者： Xuan, Pham Truong Van, Nguyen Thi Sac, Le The Thang Long Institute of Mathematics and Applied Sciences (TIMAS) Thang Long University Nghiem Xuan Yem Vietnam Faculty of Computer Science and Engineering Thuyloi University Dong Da Vietnam

In this paper, we study the global existence, uniqueness and polynomial stability of mild solutions for the Keller–Segel–Navier–Stokes system in the framework of Marcinkiewicz spaces, i.e., weak- $$L^p$$ spaces $$L^{p,\infty }(\mathbb {R}^d)$$ , where $$d \geqslant 4$$ . In our strategy, we first combine the $$L^{p,d_1}-L^{q,d_2}$$ heat estimates and Yamazaki-type estimates of the heat semigroup with fixed point arguments to obtain the global well-posedness of mild solutions by using a fixed point lemma. Then, we establish a polynomial stability of such solutions.

关键词：