1 code implementation • 25 Jan 2023 • Artur Jordao, George Correa de Araujo, Helena de Almeida Maia, Helio Pedrini
In this work, we investigate LTH and pruning at initialization from the lens of layer pruning.
no code implementations • 10 Aug 2021 • Artur Jordao, Helio Pedrini
However, studies have shown the potential of pruning as a form of regularization, which reduces overfitting and improves generalization.
2 code implementations • 23 Apr 2020 • Artur Jordao, Fernando Akio, Maiko Lie, William Robson Schwartz
Motivated by this, we propose a NAS approach to efficiently design accurate and low-cost convolutional architectures and demonstrate that an efficient strategy for designing these architectures is to learn the depth stage-by-stage.
2 code implementations • 5 Oct 2019 • Artur Jordao, Maiko Lie, Victor Hugo Cunha de Melo, William Robson Schwartz
Dimensionality reduction plays an important role in computer vision problems since it reduces computational cost and is often capable of yielding more discriminative data representation.
1 code implementation • 17 Oct 2018 • Artur Jordao, Ricardo Kloss, Fernando Yamada, William Robson Schwartz
Finally, we show that the proposed method achieves the highest FLOPs reduction and the smallest drop in accuracy when compared to state-of-the-art pruning approaches.
1 code implementation • 13 Jun 2018 • Artur Jordao, Antonio C. Nazare Jr., Jessica Sena, William Robson Schwartz
Inspired by this, we conduct an extensive set of experiments that analyze different sample generation processes and validation protocols to indicate the vulnerable points in human activity recognition based on wearable sensor data.
no code implementations • 8 Jun 2018 • Jessica Sena, Artur Jordao, William Robson Schwartz
We propose a novel method called Content-Based Spatial Consensus (CSBC), which, in addition to relying on spatial consensus, considers the content of the detection windows to learn a weighted-fusion of pedestrian detectors.
no code implementations • 7 Nov 2017 • Artur Jordao, Ricardo Kloss, William Robson Schwartz
To demonstrate the robustness and accuracy of the LHN, we evaluate it using four different networks architectures in five publicly available HAR datasets based on wearable sensors, which vary in the sampling rate and number of activities.