Locally interacting dynamical systems, such as epidemic spread, rumor
propagation through crowd, and forest fire, exhibit complex global dynamics
originated from local, relatively simple, and often stochastic interactions
between dynamic elements. Their temporal evolution is often driven by
transitions between a finite number of discrete states. Despite significant
advancements in predictive modeling through deep learning, such interactions
among many elements have rarely explored as a specific domain for predictive
modeling. We present Attentive Recurrent Neural Cellular Automata (AR-NCA), to
effectively discover unknown local state transition rules by associating the
temporal information between neighboring cells in a permutation-invariant
manner. AR-NCA exhibits the superior generalizability across various system
configurations (i.e., spatial distribution of states), data efficiency and
robustness in extremely data-limited scenarios even in the presence of
stochastic interactions, and scalability through spatial dimension-independent
prediction.

局部相互作用动力系统，如流行病传播、谣言传播和森林火灾，表现出由动态元素之间的局部、相对简单和通常随机的相互作用产生的复杂全局动态。我们提出了 AR-NCA（Attention Recurrent Neural Cellular Automata），通过以置换 - 不变的方式在相邻细胞之间关联时间信息，从而有效发现未知的局部状态转换规则。AR-NCA 在各种系统配置（即状态的空间分布）、极度数据有限的情况下，甚至在存在随机相互作用的情况下，显示出卓越的泛化能力、数据效率和鲁棒性，以及空间维度无关的可扩展性。

学习局部交互离散动力系统：迈向高效数据和可扩展预测

Learning Locally Interacting Discrete Dynamical Systems: Towards  Data-Efficient and Scalable Prediction

The forests are significant assets for every country. When it gets destroyed,
it may negatively impact the environment, and forest fire is one of the primary
causes. Fire weather indices are widely used to measure fire danger and are
used to issue bushfire warnings. It can also be used to predict the demand for
emergency management resources. Sensor networks have grown in popularity in
data collection and processing capabilities for a variety of applications in
industries such as medical, environmental monitoring, home automation etc.
Semantic sensor networks can collect various climatic circumstances like wind
speed, temperature, and relative humidity. However, estimating fire weather
indices is challenging due to the various issues involved in processing the
data streams generated by the sensors. Hence, the importance of forest fire
detection has increased day by day. The underlying Semantic Sensor Network
(SSN) ontologies are built to allow developers to create rules for calculating
fire weather indices and also the convert dataset into Resource Description
Framework (RDF). This research describes the various steps involved in
developing rules for calculating fire weather indices. Besides, this work
presents a Web-based mapping interface to help users visualize the changes in
fire weather indices over time. With the help of the inference rule, it
designed a decision support system using the SSN ontology and query on it
through SPARQL. The proposed fire management system acts according to the
situation, supports reasoning and the general semantics of the open-world
followed by all the ontologies

本研究介绍了利用语义传感器网络来计算森林火灾的火险指数以及通过 SPARQL 查询 SSN 本体构建一个基于推理的决策支持系统，帮助用户可视化地监测火险指数变化的步骤。