A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Filters
Solo or Ensemble? Choosing a CNN Architecture for Melanoma Classification
2019
2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
Preserved Fulltext
Whenever possible, the best approach for melanoma classification is still to create ensembles of multiple models. ...
For small ensembles, we found a slight advantage on the second approach but found that random choice was also competitive. ...
Avila is partially funded by Google LARA 2018. E. Valle is partially funded by a CNPq PQ-2 grant (311905/2017-0). ...
doi:10.1109/cvprw.2019.00336
dblp:conf/cvpr/PerezAV19
fatcat:bphj5vjdxzbftjrni3gw7izw3m
Solo or Ensemble? Choosing a CNN Architecture for Melanoma Classification
[article]
2019
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Whenever possible, the best approach for melanoma classification is still to create ensembles of multiple models. ...
For small ensembles, we found a slight advantage on the second approach but found that random choice was also competitive. ...
Avila is partially funded by Google LARA 2018. E. Valle is partially funded by a CNPq PQ-2 grant (311905/2017-0). ...
arXiv:1904.12724v1
fatcat:il2m2o7ub5c7hnlehhl7oqga7y
An Empirical Evaluation of Current Convolutional Architectures' Ability to Manage Nuisance Location and Scale Variability
[article]
2016
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
very effective at marginalizing nuisance variability. ...
We also quantify the effects of context on the overall classification task and its impact on the performance of CNNs, and propose improved sampling techniques for heuristic proposal schemes that improve ...
We gratefully acknowledge NVIDIA Corporation for donating a K40 GPU that was used in support of some of the experiments. ...
arXiv:1505.06795v2
fatcat:wisopx623fdgbhmhz5jaufv2ha
An Empirical Evaluation of Current Convolutional Architectures' Ability to Manage Nuisance Location and Scale Variability
2016
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2017; you can also visit the original URL.
The file type is application/pdf.
very effective at marginalizing nuisance variability. ...
We also quantify the effects of context on the overall classification task and its impact on the performance of CNNs, and propose improved sampling techniques for heuristic proposal schemes that improve ...
We gratefully acknowledge NVIDIA Corporation for donating a K40 GPU that was used in support of some of the experiments. ...
doi:10.1109/cvpr.2016.481
dblp:conf/cvpr/KarianakisDS16
fatcat:i3sdwsrsb5gxxgo77xcnviioye
End-to-End Training of Hybrid CNN-CRF Models for Stereo
[article]
2017
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We propose a novel and principled hybrid CNN+CRF model for stereo estimation. ...
For inference, we apply a recently proposed highly parallel dual block descent algorithm which only needs a small fixed number of iterations to compute a high-quality approximate minimizer. ...
This requires robustness to all kinds of visual nuisances as well as a good prior model of the 3D environment. ...
arXiv:1611.10229v2
fatcat:kmqnpznsxjdvxknu5gccc4wz24
What Makes for Good Views for Contrastive Learning?
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2020 pre-print arXiv:2005.10243v1 fatcat:l5gkrfrkgnawno76auoy4evaga
As a by-product, we achieve a new state-of-the-art accuracy on unsupervised pre-training for ImageNet classification (73% top-1 linear readout with a ResNet-50). ...
Despite its success, the influence of different view choices has been less studied. ...
Yonglong is grateful to Zhoutong Zhang for encouragement and feedback on experimental design. ...
arXiv:2005.10243v3
fatcat:36l33k5k7vhp3gylsyrigifyyq
An adversarial approach for the robust classification of pneumonia from chest radiographs
2020
Proceedings of the ACM Conference on Health, Inference, and Learning
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Specically, we demonstrate improved out-of-hospital generalization performance of a pneumonia classier by training a model that is invariant to the view position of chest radiographs (anterior-posterior ...
In order for these models to be safely deployed, we would like to ensure that they do not use confounding variables to make their classication, and that they will work well even when tested on images from ...
Su-In Lee's lab for their valuable general feedback on the project. ...
doi:10.1145/3368555.3384458
dblp:conf/chil/JanizekEDL20
fatcat:bovemdso7jddndlceujn6znuk4
AutoBayes: Automated Bayesian Graph Exploration for Nuisance-Robust Inference
2021
IEEE Access
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
We benchmark the framework on several public datasets, and provide analysis of its capability for subject-transfer learning with/without variational modeling and adversarial training. ...
Auto-Bayes also enables learning disentangled representations, where the latent variable is split into multiple pieces to impose various relationships with the nuisance variation and task labels. ...
Although A-CVAE in Fig. 1 (b) may offer nuisance-robust performance through adversarial disentanglement of S from latent Z , there is no guarantee that such a model can perform well across different datasets ...
doi:10.1109/access.2021.3064530
fatcat:hfwenaojunegfbytlcvc73z2e4
AutoBayes: Automated Bayesian Graph Exploration for Nuisance-Robust Inference
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We benchmark the framework on several public datasets, where we have access to subject and class labels during training, and provide analysis of its capability for subject-transfer learning with/without ...
Learning data representations that capture task-related features, but are invariant to nuisance variations remains a key challenge in machine learning. ...
Although A-CVAE in Fig. 1 (b) may offer nuisance-robust performance through adversarial disentanglement of S from latent Z, there is no guarantee that such a model can perform well across different datasets ...
arXiv:2007.01255v2
fatcat:su2j7qcsfndsnbxp4nb2dnv2c4
An Adversarial Approach for the Robust Classification of Pneumonia from Chest Radiographs
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Specifically, we demonstrate improved out-of-hospital generalization performance of a pneumonia classifier by training a model that is invariant to the view position of chest radiographs (anterior-posterior ...
In order for these models to be safely deployed, we would like to ensure that they do not use confounding variables to make their classification, and that they will work well even when tested on images ...
Su-In Lee's lab for their valuable general feedback on the project. ...
arXiv:2001.04051v1
fatcat:sgpos7poz5ewdostefqxhdubbi
End-to-End Training of Hybrid CNN-CRF Models for Stereo
2017
2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
We propose a novel method for stereo estimation, combining advantages of convolutional neural networks (CNNs) and optimization-based approaches. ...
The trained hybrid model with shallow CNNs is comparable to state-of-the-art deep models in both time and performance. ...
This requires robustness to all kinds of visual nuisances as well as a good prior model of the 3D environment. ...
doi:10.1109/cvpr.2017.159
dblp:conf/cvpr/KnobelreiterRSP17
fatcat:pve22663nvbsplidi2nwsrdbda
Field Convolutions for Surface CNNs
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
The result is a rich notion of convolution which we call field convolution, well-suited for CNNs on surfaces. ...
We present a novel surface convolution operator acting on vector fields that is based on a simple observation: instead of combining neighboring features with respect to a single coordinate parameterization ...
The result is a rich notion of convolution which we call field convolution (FC), well-suited for CNNs on surfaces. ...
arXiv:2104.03916v2
fatcat:ltt2urfjfndklj2jvvqzymvw2y
Representation Based Complexity Measures for Predicting Generalization in Deep Learning
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
An implementation of our solution is available at https://github.com/parthnatekar/pgdl. ...
Deep Neural Networks can generalize despite being significantly overparametrized. ...
Acknowledgements We'd like to thank the organizers of the NeurIPS 2020 Competition on Predicting Generalization in Deep Learning for hosting this competition and for providing a platform for us to test ...
arXiv:2012.02775v1
fatcat:6kniioe4rfhmvj3t7qh4yrvkpa
Deep Imbalanced Learning for Face Recognition and Attribute Prediction
[article]
2019
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We show that it is easy to deploy angular margins between the cluster distributions on a hypersphere manifold. ...
We further demonstrate that more discriminative deep representation can be learned by enforcing a deep network to maintain inter-cluster margins both within and between classes. ...
These margins are designed angular, which translates well to the inner product based similarity metric on a unit hypersphere. ...
arXiv:1806.00194v2
fatcat:3qbovwkh7bfknaudlcjclui4gi
Learning Deep Representation for Imbalanced Classification
2016
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2016; you can also visit the original URL.
The file type is application/pdf.
We further demonstrate that more discriminative deep representation can be learned by enforcing a deep network to maintain both intercluster and inter-class margins. ...
We show that the margins can be easily deployed in standard deep learning framework through quintuplet instance sampling and the associated triple-header hinge loss. ...
This work is partially supported by SenseTime Group Limited and the Hong Kong Innovation and Technology Support Programme. ...
doi:10.1109/cvpr.2016.580
dblp:conf/cvpr/HuangLLT16
fatcat:pr23evghqzcoboup7gbfrq63uq
« Previous
Showing results 1 — 15 out of 632 results