Multi-unit building address geocoding: An approach without indoor location reference data

Article Computer Science, Information Systems

Deep Learning for Toponym Resolution: Geocoding Based on Pairs of Toponyms

Jacques Fize et al.

Summary: Geocoding aims to assign geographic coordinates to place names mentioned within documents, facing challenges like referent ambiguity and database completeness. The proposed geocoding approach models pairs of toponyms utilizing a deep neural network. Experiment results show that models trained with co-occurrence data achieve higher geocoding accuracy.

ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION (2021)

Add to Collection

Proceedings Paper Computer Science, Artificial Intelligence

House-GAN plus plus : Generative Adversarial Layout Refinement Network towards Intelligent Computational Agent for Professional Architects

Nelson Nauata et al.

Summary: This paper presents a generative adversarial network for automated floorplan generation, utilizing a combination of graph-constrained relational GAN and conditional GAN to iteratively refine layouts. Through meta-optimization techniques, the generator's performance is improved, resulting in significant enhancements over current state-of-the-art methods and even competitive with floorplans designed by professional architects.

2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 (2021)

Add to Collection

Article Computer Science, Information Systems

SDTR: Soft Decision Tree Regressor for Tabular Data

Haoran Luo et al.

Summary: Traditional methods are favored by researchers in handling tabular data, while non-differentiable and uninterpretable models are inconvenient for neural network ensembles. The Soft Decision Tree Regressor (SDTR) is a hierarchical differentiable neural network model that outperforms traditional methods and uninterpretable models.

IEEE ACCESS (2021)

Add to Collection

Article Geography, Physical