Redirigiendo al acceso original de articulo en 16 segundos...
Inicio  /  Algorithms  /  Vol: 16 Par: 7 (2023)  /  Artículo
ARTÍCULO
TITULO

Integrated Gradients for Feature Assessment in Point Cloud-Based Data Sets

Markus Schwegler    
Christoph Müller and Alexander Reiterer    

Resumen

Integrated gradients is an explainable AI technique that aims to explain the relationship between a model?s predictions in terms of its features. Adapting this technique to point clouds and semantic segmentation models allows a class-wise attribution of the predictions with respect to the input features. This allows better insight into how a model reached a prediction. Furthermore, it allows a quantitative analysis of how much each feature contributes to a prediction. To obtain these attributions, a baseline with high entropy is generated and interpolated with the point cloud to be visualized. These interpolated point clouds are then run through the network and their gradients are collected. By observing the change in gradients during each iteration an attribution can be found for each input feature. These can then be projected back onto the original point cloud and compared to the predictions and input point cloud. These attributions are generated using RandLA-Net due to it being an efficient semantic segmentation model that uses comparatively few parameters, therefore keeping the number of gradients that must be stored at a reasonable level. The attribution was run on the public Semantic3D dataset and the SVGEO large-scale urban dataset.

 Artículos similares

       
 
Lorenzo Pasculli, Viviana Piermattei, Alice Madonia, Gabriele Bruzzone, Massimo Caccia, Roberta Ferretti, Angelo Odetti and Marco Marcelli    
The Arctic region is greatly affected by climate change, with evident alterations in both physical and biological processes: temperatures are changing at a rate that is twice the global average and phytoplankton productivity is directly affected by ice m... ver más

 
Chang Lin, Wu Chen and Haifeng Zhou    
To visually detect sea-surface targets, the objects of interest must be effectively and rapidly isolated from the background of sea-surface images. In contrast to traditional image detection methods, which employ a single visual feature, this paper propo... ver más

 
Tarandeep S. Kalra, Xiangyu Li, John C. Warner, Wayne R. Geyer and Hui Wu    
The numerical simulation of estuarine dynamics requires accurate prediction for the transport of tracers, such as temperature and salinity. During the simulation of these processes, all the numerical models introduce two kinds of tracer mixing: (1) by pa... ver más

 
Digna T. Rueda-Roa, Tal Ezer and Frank E. Muller-Karger    
The southern Caribbean Sea experiences strong coastal upwelling between December and April due to the seasonal strengthening of the trade winds. A second upwelling was recently detected in the southeastern Caribbean during June?August, when local coastal... ver más

 
Su Yean Teh, Michael Turtora, Donald L. DeAngelis, Jiang Jiang, Leonard Pearlstine, Thomas J. Smith III and Hock Lye Koh    
Global climate change poses challenges to areas such as low-lying coastal zones, where sea level rise (SLR) and storm-surge overwash events can have long-term effects on vegetation and on soil and groundwater salinities, posing risks of habitat loss crit... ver más