2^nd International Conference on Artificial Intelligence and IoT (AIIoT 2024)

Accepted Papers

Design of Proprietary Frameworks for Neural Models: Methodology and Best Practices

José Gabriel Carrasco Ramirez, CEO at Quarks Advantage, Jersey City, New Jersey. United States of America

ABSTRACT

The creation of proprietary frameworks for the development of neural models is essential to meet specific needs that generic frameworks cannot address. This article examines the key stages in the design of these frameworks and offers best practices for their effective implementation. It explores everything from needs identification and resource assessment to architectural design and implementation. Additionally, it emphasizes the importance of user-centered design and continuous evaluation to ensure the framework's usability and adaptability to changing needs.

Keywords

Proprietary frameworks, neural models, artificial intelligence, framework design, model optimization, user-centered design, continuous evaluation, scalability, performance optimization, data management, model training, regulatory compliance, explainable AI (XAI), agile methodology, security and privacy.

Multi-classification of CAD Entities: Leveraging the Entity-as-node Approach with Graph Neural Networks

Sheela Raju Kurupathi¹, Park Dongryul¹, Sebastian Bosse¹, and Peter Eisert^{1, 2}, ¹Fraunhofer Heinrich Hertz Institute (HHI), ²Humboldt-Universit¨at zu Berlin

ABSTRACT

The construction industry faces challenges in extracting and interpreting semantic information from CAD floor plans and related data. Graph Neural Networks (GNNs) have emerged as a potential solution, preserving the structural integrity of CAD drawings without rasterization. Accurate identification of structural symbols, such as walls, doors, windows, etc. is vital for generalizing floor plans. This paper investigates GNN methods to enhance the classification of these symbols in CAD floorplans, proposing an entity-as-node graph representation. We evaluate various preprocessing strategies and GNN architectures, including Graph Attention Networks (GAT), GATv2, Generalized Aggregation Networks (GEN), Principal Neighborhood Aggregation (PNA), and Unified Message Passing (UniMP) on the CubiCasa5K dataset. Our results show that these methods significantly outperform current state-of-the-art approaches, demonstrating their effectiveness in CAD floor plan entity classification.

Keywords

BIM, CAD, Floor Plans, GNN, Entity-as-Node, Multi-Classification.