Institution profile

Hiroshima City University

Academic institutionasia · jp
Official website
Research library5linked papers
Opportunities0open roles
Selected work

Representative Papers

Automatic Extraction of Road Networks by Using Teacher–Student Adaptive Structural Deep Belief Network and Its Application to Landslide Disaster

Nov 04, 2025IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

To address the challenge of rapid post-disaster road accessibility assessment (e.g., after landslides), this paper proposes an adaptive Deep Belief Network (DBN) method based on a teacher–student learning framework for fully automatic road network extraction and passability classification from aerial/satellite imagery. The method innovatively integrates Restricted Boltzmann Machines (RBMs), a neuron birth–death algorithm, and layer-wise generative strategies to dynamically optimize both network architecture and parameters, while incorporating lightweight design for real-time inference on edge devices. Evaluated on a benchmark dataset comprising satellite/aerial imagery from seven major cities, the approach achieves a substantial improvement in mean road detection accuracy—from 40.0% to 89.0%. It was successfully deployed to analyze pre- and post-rainfall disaster satellite imagery in Japan, accurately identifying traversable road segments. The method significantly enhances robustness, generalization capability, and practical deployability under complex, real-world conditions.

4 citationsRead paper

Efficient Pattern Matching for Unordered Term Tree Patterns under Generalized Height-Constrained Bindings

Jul 16, 2026

This study addresses the matching problem for unordered tree patterns under generalized height constraints, relaxing the conventional restriction that variables may only bind to leaf nodes by allowing their child ports to map to any non-root node in the binding tree. To this end, we introduce a more general variable-binding model with height constraints and present, for the first time, a polynomial-time matching algorithm for this setting. By reducing the problem to a graph isomorphism subproblem and integrating efficient tree pattern matching techniques, the proposed method demonstrates superior performance both theoretically and empirically, confirming its practical efficiency and feasibility.

0 citationsRead paper

Efficient Pattern Matching in Unordered Term Tree Patterns with Height Constraints

Jul 02, 2026

This work addresses the problem of efficiently matching highly constrained patterns in unordered child trees—such as abstract syntax trees or chemical structures—by introducing a novel formalism called "unordered item tree patterns." For the first time, this framework incorporates variables that bound both the length of the pattern backbone and the height of subtrees, enabling a precise characterization of the matching task. Leveraging tools from graph theory and combinatorial optimization, the authors devise a polynomial-time algorithm with a time complexity of \(O(N \cdot \max\{nD^{3/2}, \mathcal{S}\})\), where \(N\) denotes the size of the input tree, \(n\) and \(D\) relate to pattern structure, and \(\mathcal{S}\) captures the solution space size. Empirical evaluation on real-world datasets demonstrates the algorithm’s efficiency and scalability, confirming its suitability for large-scale pattern matching in unordered tree structures.

0 citationsRead paper
Recent publications

Latest Papers

Efficient Pattern Matching for Unordered Term Tree Patterns under Generalized Height-Constrained Bindings

Jul 16, 2026

This study addresses the matching problem for unordered tree patterns under generalized height constraints, relaxing the conventional restriction that variables may only bind to leaf nodes by allowing their child ports to map to any non-root node in the binding tree. To this end, we introduce a more general variable-binding model with height constraints and present, for the first time, a polynomial-time matching algorithm for this setting. By reducing the problem to a graph isomorphism subproblem and integrating efficient tree pattern matching techniques, the proposed method demonstrates superior performance both theoretically and empirically, confirming its practical efficiency and feasibility.

0 citationsRead paper

Efficient Pattern Matching in Unordered Term Tree Patterns with Height Constraints

Jul 02, 2026

This work addresses the problem of efficiently matching highly constrained patterns in unordered child trees—such as abstract syntax trees or chemical structures—by introducing a novel formalism called "unordered item tree patterns." For the first time, this framework incorporates variables that bound both the length of the pattern backbone and the height of subtrees, enabling a precise characterization of the matching task. Leveraging tools from graph theory and combinatorial optimization, the authors devise a polynomial-time algorithm with a time complexity of \(O(N \cdot \max\{nD^{3/2}, \mathcal{S}\})\), where \(N\) denotes the size of the input tree, \(n\) and \(D\) relate to pattern structure, and \(\mathcal{S}\) captures the solution space size. Empirical evaluation on real-world datasets demonstrates the algorithm’s efficiency and scalability, confirming its suitability for large-scale pattern matching in unordered tree structures.

0 citationsRead paper

Automatic Extraction of Road Networks by Using Teacher–Student Adaptive Structural Deep Belief Network and Its Application to Landslide Disaster

Nov 04, 2025IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

To address the challenge of rapid post-disaster road accessibility assessment (e.g., after landslides), this paper proposes an adaptive Deep Belief Network (DBN) method based on a teacher–student learning framework for fully automatic road network extraction and passability classification from aerial/satellite imagery. The method innovatively integrates Restricted Boltzmann Machines (RBMs), a neuron birth–death algorithm, and layer-wise generative strategies to dynamically optimize both network architecture and parameters, while incorporating lightweight design for real-time inference on edge devices. Evaluated on a benchmark dataset comprising satellite/aerial imagery from seven major cities, the approach achieves a substantial improvement in mean road detection accuracy—from 40.0% to 89.0%. It was successfully deployed to analyze pre- and post-rainfall disaster satellite imagery in Japan, accurately identifying traversable road segments. The method significantly enhances robustness, generalization capability, and practical deployability under complex, real-world conditions.

4 citationsRead paper