Redirigiendo al acceso original de articulo en 20 segundos...
Inicio  /  Algorithms  /  Vol: 14 Par: 2 (2021)  /  Artículo
ARTÍCULO
TITULO

Representing Deep Neural Networks Latent Space Geometries with Graphs

Carlos Lassance    
Vincent Gripon and Antonio Ortega    

Resumen

Deep Learning (DL) has attracted a lot of attention for its ability to reach state-of-the-art performance in many machine learning tasks. The core principle of DL methods consists of training composite architectures in an end-to-end fashion, where inputs are associated with outputs trained to optimize an objective function. Because of their compositional nature, DL architectures naturally exhibit several intermediate representations of the inputs, which belong to so-called latent spaces. When treated individually, these intermediate representations are most of the time unconstrained during the learning process, as it is unclear which properties should be favored. However, when processing a batch of inputs concurrently, the corresponding set of intermediate representations exhibit relations (what we call a geometry) on which desired properties can be sought. In this work, we show that it is possible to introduce constraints on these latent geometries to address various problems. In more detail, we propose to represent geometries by constructing similarity graphs from the intermediate representations obtained when processing a batch of inputs. By constraining these Latent Geometry Graphs (LGGs), we address the three following problems: (i) reproducing the behavior of a teacher architecture is achieved by mimicking its geometry, (ii) designing efficient embeddings for classification is achieved by targeting specific geometries, and (iii) robustness to deviations on inputs is achieved via enforcing smooth variation of geometry between consecutive latent spaces. Using standard vision benchmarks, we demonstrate the ability of the proposed geometry-based methods in solving the considered problems.

 Artículos similares

       
 
Edamkue Israel, Selvarajan Ramganesh, Akebe Luther King Abia and Chioma Blaise Chikere    
The marine environment possesses diverse and complex characteristics, representing a significant challenge for microbial survival. Therefore, bacteria must develop adaptive mechanisms to thrive in such environments. Quorum sensing (QS), a well-establishe... ver más

 
Norbert Fischer, Alexander Hartelt and Frank Puppe    
Digitization and transcription of historic documents offer new research opportunities for humanists and are the topics of many edition projects. However, manual work is still required for the main phases of layout recognition and the subsequent optical c... ver más
Revista: Algorithms

 
Guozhen Huang, Yichang Tang, Xi Chen, Mingsheng Chen and Yanlin Jiang    
Fossil fuel consumption has progressively increased alongside global population growth, representing the predominant energy consumption pattern for humanity. Unfortunately, this persistent reliance on fossil fuels has resulted in a substantial surge in p... ver más

 
Bikram Pratim Bhuyan, Vaishnavi Jaiswal and Amar Ramdane Cherif    
Investors at well-known firms are increasingly becoming interested in stock forecasting as they seek more effective methods to predict market behavior using behavioral finance tools. Accordingly, studies aimed at predicting stock performance are gaining ... ver más
Revista: Computers

 
Samuel R. Schrader and Eren Gultepe    
The evaluation of similarities between natural languages often relies on prior knowledge of the languages being studied. We describe three methods for building phylogenetic trees and clustering languages without the use of language-specific information. ... ver más
Revista: Informatics