Estimating accurate multi-class probabilities with support vector machines

By Jonathan Milgram, Mohamed Cheriet, Robert Sabourin. 2005

51 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yoshua Bengio
2. Kyunghyun Cho
3. Li Deng
4. Quoc V. Le
5. Jianfeng Gao
6. Mohit Bansal
7. Percy Liang
8. Trevor Darrell

In Computer Vision and Pattern Recognition

1. Yoshua Bengio
2. Luc Van Gool
3. Bernt Schiele
4. Trevor Darrell
5. Dacheng Tao
6. Ming-Hsuan Yang
7. Jitendra Malik
8. Xiaogang Wang

Citations in the corpus

• EMNIST: an extension of MNIST to handwritten letters (2017, methodology): “Another work of interest is that of Milgram et al., who made use of the digits in the NIST dataset as a classification task for comparing post-processing methods for multi-class probabilities with Support Vector Machines [13].”
• Deep Learners Benefit More from Out-of-Distribution Examples (2011, methodology): “…on NIST test digits only, along with the previous results from literature [Granger et al., 2007, Pérez-Cortes et al., 2000, Oliveira et al., 2002, Milgram et al., 2005] respectively based on ART, nearest neighbors, MLPs, and SVMs. setting is similar for the other two target classes (lower case…”
• Supervised learning of fuzzy ARTMAP neural networks through particle swarm optimization (2007, methodology): “Table 2 shows the generalization error obtained by the k-NN, MLP, SVM, and fuzzy ARTMAP classifiers after learning all the training data (15,000 patterns per class) of the NIST SD19 data set.”
• Convolutional Neural Network Committees for Handwritten Character Classification (2011, unknown): “88 [19] letters (52) 21.”
• Multi-column deep neural networks for image classification (2012, unknown): “88 [25] letters (52) 21.”
• Learning sparse models for a dynamic Bayesian network classifier of protein secondary structure (2011, methodology): “For this purpose, it is possible to provide the posterior distribution generated from the DBNs directly as input features to a structure prediction algorithm or convert the output of the SVM to a probability [37,38].”
• Active learning and discovery of object categories in the presence of unnameable instances (2015, background): “Furthermore, normalizing them afterwards often leads to poor results due to imbalanced scores [29].”