Zobrazeno 1 - 10
of 272
pro vyhledávání: '"Carvalho, Marcus"'
Multitude of deep learning models have been proposed for node classification in graphs. However, they tend to perform poorly under labeled-data scarcity. Although Few-shot learning for graphs has been introduced to overcome this problem, the existing
Externí odkaz:
http://arxiv.org/abs/2406.04825
Continual learning is a process that involves training learning agents to sequentially master a stream of tasks or classes without revisiting past data. The challenge lies in leveraging previously acquired knowledge to learn new tasks efficiently, wh
Externí odkaz:
http://arxiv.org/abs/2402.12490
Catastrophic forgetting has been a significant problem hindering the deployment of deep learning algorithms in the continual learning setting. Numerous methods have been proposed to address the catastrophic forgetting problem where an agent loses its
Externí odkaz:
http://arxiv.org/abs/2209.02112
Graph Neural Networks (GNNs) have become the backbone for a myriad of tasks pertaining to graphs and similar topological data structures. While many works have been established in domains related to node and graph classification/regression tasks, the
Externí odkaz:
http://arxiv.org/abs/2209.01556
Autor:
Weng, Weiwei, Pratama, Mahardhika, Za'in, Choiru, De Carvalho, Marcus, Appan, Rakaraddi, Ashfahani, Andri, Yee, Edward Yapp Kien
Publikováno v:
IEEE Transactions on Neural Networks and Learning Systems, 2022
A cross domain multistream classification is a challenging problem calling for fast domain adaptations to handle different but related streams in never-ending and rapidly changing environments. Notwithstanding that existing multistream classifiers as
Externí odkaz:
http://arxiv.org/abs/2209.01548
We consider the problem of online unsupervised cross-domain adaptation, where two independent but related data streams with different feature spaces -- a fully labeled source stream and an unlabeled target stream -- are learned together. Unique chara
Externí odkaz:
http://arxiv.org/abs/2110.01326
Autor:
Carvalho, Marcus Rodrigo
Este trabalho descreve uma abordagem do método primal-dual barreira logarítmica (MPDBL) associado ao método de Newton modificado para a resolução do problema de fluxo de potência para sistemas de distribuição radial. Também foi realizado um
Orientadora: Thais Regina Pavez
A feira livre de domingo na cidade de Marília foi deslocada de lugar por interferência da igreja Universal do Reino de Deus e com a mediação do poder público. Ao pesquisar os elementos que construíram esse a
A feira livre de domingo na cidade de Marília foi deslocada de lugar por interferência da igreja Universal do Reino de Deus e com a mediação do poder público. Ao pesquisar os elementos que construíram esse a
Externí odkaz:
http://hdl.handle.net/11449/190858
Transferring knowledge across many streaming processes remains an uncharted territory in the existing literature and features unique characteristics: no labelled instance of the target domain, covariate shift of source and target domain, different pe
Externí odkaz:
http://arxiv.org/abs/1910.03434