Deep learning on small datasets without pre-training using cosine loss

B Barz, J Denzler - Proceedings of the IEEE/CVF winter …, 2020 - openaccess.thecvf.com
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …
被引用次数：199 相关文章

Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler - arXiv preprint arXiv:1901.09054, 2019 - arxiv.org
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …
[PDF] researchgate.net

[PDF][PDF] Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler - researchgate.net
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …

Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler - 2020 IEEE Winter Conference on Applications of …, 2020 - computer.org
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …

Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler - arXiv e-prints, 2019 - ui.adsabs.harvard.edu
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …

Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler - … IEEE Winter Conference on Applications of …, 2020 - ieeexplore.ieee.org
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …
[PDF] researchgate.net

[PDF][PDF] Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler - researchgate.net
Two things seem to be indisputable in the contemporary deep learning discourse: 1. The
categorical cross-entropy loss after softmax activation is the method of choice for …

[引用][C] Deep Learning on Small Datasets without Pre-Training using Cosine Loss

B Barz, J Denzler