Filters








253 Hits in 7.9 sec

Unsupervised Selection and Discriminative Estimation of Orthogonal Gaussian Mixture Models for Handwritten Digit Recognition

Xuefeng Chen, Xiabi Liu, Yunde Jia
2009 2009 10th International Conference on Document Analysis and Recognition  
We apply the proposed learning approach of OGMM to handwritten digit recognition.  ...  After the model selection is completed, a discriminative learning framework of Bayesian classifiers called Max-Min posterior pseudoprobabilities (MMP) is employed to estimate component parameters in OGMM  ...  Acknowledges The authors would like to thank Dr. Cheng-Lin Liu for providing the e-grg features of handwritten digits.  ... 
doi:10.1109/icdar.2009.44 dblp:conf/icdar/ChenLJ09 fatcat:oxwbehbnmvfhfc5qxw3ppv5boa

Statistical Modeling and Learning for Recognition-Based Handwritten Numeral String Segmentation

Yanjie Wang, Xiabi Liu, Yunde Jia
2009 2009 10th International Conference on Document Analysis and Recognition  
Based on this modeling, the recognition based segmentation is solved under the Max-Min posterior Pseudo-probabilities (MMP) framework of learning Bayesian classifiers.  ...  In the training phase, we use the MMP method to learn a posterior pseudo-probability measure function from positive samples and negative samples of numeral strings segmented correctly.  ...  Unknown parameters in the posterior pseudo-probability measure function are learned from the training data by the MMP method.  ... 
doi:10.1109/icdar.2009.25 dblp:conf/icdar/WangLJ09 fatcat:ffy2hpyn2nhaxnyyoqmudipn2a

Discriminative structure selection method of Gaussian Mixture Models with its application to handwritten digit recognition

Xuefeng Chen, Xiabi Liu, Yunde Jia
2011 Neurocomputing  
We introduce a GMM structure selection criterion based on a discriminative objective function called Soft target based Max-Min posterior Pseudo-probabilities (Soft-MMP).  ...  We evaluate the proposed GMM structure selection method through the experiments of handwritten digit recognition on the well-known CENPARMI and MNIST digit databases.  ...  The used discriminative learning criterion is SOFT target based Max-Min posterior Pseudo-probabilities (Soft-MMP) [20] .  ... 
doi:10.1016/j.neucom.2010.11.010 fatcat:j67jprdt3vgvfj7sevb6qkvtp4

Combining evolution strategy and gradient descent method for discriminative learning of bayesian classifiers

Xuefeng Chen, Xiabi Liu, Yunde Jia
2009 Proceedings of the 11th Annual conference on Genetic and evolutionary computation - GECCO '09  
posterior Pseudo-probabilities (Soft-MMP) learning framework.  ...  As a result, the efficiency and the effectiveness of the CMA-ES are improved. We apply the Soft-MMP with the proposed hybrid optimization approach to handwritten digit recognition.  ...  ACKNOWLEDGEMENTS The authors would like to thank Dr. Cheng-Lin Liu for providing the e-grg features of handwritten digits and the referees for their valuable comments and suggestions. This  ... 
doi:10.1145/1569901.1569972 dblp:conf/gecco/ChenLJ09 fatcat:sbyfev76qvbufiqj4ajykcvxxi

Objective Function Design for MCE-Based Combination of On-line and Off-line Character Recognizers for On-line Handwritten Japanese Text Recognition

Bilan Zhu, JinFeng Gao, Masaki Nakagawa
2011 2011 International Conference on Document Analysis and Recognition  
This paper describes effective object function design for combining on-line and off-line character recognizers for on-line handwritten Japanese text recognition.  ...  Moreover, we apply a genetic algorithm to estimate super parameters such as the number of clusters, initial learning rate and maximum learning times as well as the sigmoid function parameter for MCE optimization  ...  same parameters, secondly, we apply α, ω, ε, T to learn the weighting parameters by MCE to obtain a smallest MCE criterion L min MCE , and set 1-L min MCE as the fitness of each chromosome {G, α, ω, ε,  ... 
doi:10.1109/icdar.2011.125 dblp:conf/icdar/ZhuGN11 fatcat:ouqjit3ucvfkniotvfo4vxoo5e

Active graph based semi-supervised learning using image matching: Application to handwritten digit recognition

Hubert Cecotti
2016 Pattern Recognition Letters  
The robustness of the method is highlighted by the performance of handwritten digit recognition in different scripts.  ...  In this paper, we present a novel active learning strategy for the classification of handwritten digits.  ...  Acknowledgment The author would like to thank Prof. Ujjwal Bhattacharya for sharing the databases of Bangla, Devnagari, and Oriya.  ... 
doi:10.1016/j.patrec.2016.01.016 fatcat:dijkntjghbhd3erzofruiv6oni

Closed Loop Neural-Symbolic Learning via Integrating Neural Perception, Grammar Parsing, and Symbolic Reasoning [article]

Qing Li, Siyuan Huang, Yining Hong, Yixin Chen, Ying Nian Wu, Song-Chun Zhu
2020 arXiv   pre-print
The experiments are conducted on two weakly-supervised neural-symbolic tasks: (1) handwritten formula recognition on the newly introduced HWF dataset; (2) visual question answering on the CLEVR dataset  ...  Prior methods learn the neural-symbolic models using reinforcement learning (RL) approaches, which ignore the error propagation in the symbolic reasoning module and thus converge slowly with sparse rewards  ...  We thank Baoxiong Jia for helpful discussion on the generalized Earley Parser. This work reported herein is supported by ARO W911NF1810296, DARPA XAI N66001-17-2-4029, and ONR MURI N00014-16-1-2007.  ... 
arXiv:2006.06649v2 fatcat:nbr5yyhaqzbkjfdb7z2om2db34

Generative OpenMax for Multi-Class Open Set Classification [article]

ZongYuan Ge, Sergey Demyanov, Zetao Chen, Rahil Garnavi
2017 arXiv   pre-print
We validate the proposed method on two datasets of handwritten digits and characters, resulting in superior results over previous deep learning based method OpenMax Moreover, G-OpenMax provides a way to  ...  The proposed method, called Gener- ative OpenMax (G-OpenMax), extends OpenMax by employing generative adversarial networks (GANs) for novel category image synthesis.  ...  Those methods model the known class distribution using a parametric model from which the posterior probabilities of a test sample are computed.  ... 
arXiv:1707.07418v1 fatcat:hdd6satvqne77glzyqgdebcr4e

COMBINING MODEL-BASED AND DISCRIMINATIVE APPROACHES IN A MODULAR TWO-STAGE CLASSIFICATION SYSTEM: APPLICATION TO ISOLATED HANDWRITTEN DIGIT RECOGNITION [chapter]

Jonathan Milgram, Robert Sabourin, Mohamed Cheriet
2009 Series in Machine Perception and Artificial Intelligence  
In the first stage we estimate the posterior probabilities with a model-based approach and we re-estimate only the highest probabilities with appropriate Support Vector Classifiers (SVC) in the second  ...  Finally, the first experiments on the benchmark database MNIST have shown that our dynamic classification process allows to maintain the accuracy of SVCs, while decreasing complexity by a factor 8.7 and  ...  Experimental results To evaluate our method, we chose a classical pattern recognition problem: isolated handwritten digit recognition.  ... 
doi:10.1142/9789812834461_0011 dblp:series/smpai/MilgramSC09 fatcat:xn5xwtia7fhd3lwcd237zykgn4

Combining Model-based and Discriminative Approaches in a Modular Two-stage Classification System: Application to Isolated Handwritten Digit Recognition

Jonathan Milgram, Robert Sabourin, Mohamed Cheriet
2005 ELCVIA Electronic Letters on Computer Vision and Image Analysis  
In the first stage we estimate the posterior probabilities with a model-based approach and we re-estimate only the highest probabilities with appropriate Support Vector Classifiers (SVC) in the second  ...  Finally, the first experiments on the benchmark database MNIST have shown that our dynamic classification process allows to maintain the accuracy of SVCs, while decreasing complexity by a factor 8.7 and  ...  Experimental results To evaluate our method, we chose a classical pattern recognition problem: isolated handwritten digit recognition.  ... 
doi:10.5565/rev/elcvia.92 fatcat:nganv5b33zdsljnx7nyxdpcqkq

Convex Optimizations for Distance Metric Learning and Pattern Classification [Applications Corner

Kilian Weinberger, Fei Sha, Lawrence Saul
2010 IEEE Signal Processing Magazine  
EXPERIMENTAL RESULTS We experimented with LMNN classification on problems in face recognition and handwritten digit recognition.  ...  For digit recognition, we experimented with images from an extensively benchmarked data set of handwritten digits (available at http://yann.lecun.com/ exdb/mnist).  ...  [SP] define a valid distance metric; second, because linear inequalities in the elements of this matrix can ensure that inputs are correctly labeled by kNN classification or Gaussian mixture modeling.  ... 
doi:10.1109/msp.2010.936013 fatcat:ffx2slovabbelmoil7nz2iptia

Online EM for Functional Data [article]

Florian Maire, Eric Moulines, Sidonie Lefebvre
2016 arXiv   pre-print
Our sequential inference framework is significantly more computationally efficient than equivalent batch learning algorithms, especially when the missing data is high-dimensional.  ...  A novel approach to perform unsupervised sequential learning for functional data is proposed.  ...  Acknowledgments This work has been supported by the ONERA, the French Aerospace Lab and the DGA, the French Procurement Agency.  ... 
arXiv:1604.00570v1 fatcat:44kzhojuwralnodhvivhxzdoju

Optimally combining a cascade of classifiers

Kumar Chellapilla, Michael Shilman, Patrice Simard, Kazem Taghva, Xiaofan Lin
2006 Document Recognition and Retrieval XIII  
The effectiveness of the approach is demonstrated on handwritten character recognition by finding a) the fastest possible combination given an upper bound on classification error, and also b) the most  ...  The search procedure only updates the rejection thresholds (one for each constituent classier) in the cascade, consequently no new classifiers are added and no training is necessary.  ...  One such network achieved the best known error rate of 0.4% [7] for handwritten digit recognition on (MNIST).  ... 
doi:10.1117/12.643669 dblp:conf/drr/ChellapillaSS06 fatcat:jjswwl7py5az7pa5jboamhkqpq

Online EM for functional data

Florian Maire, Eric Moulines, Sidonie Lefebvre
2017 Computational Statistics & Data Analysis  
The designed sequential inference framework is significantly more computationally efficient than equivalent batch learning algorithms, especially when the missing data is high-dimensional.  ...  A novel approach to perform unsupervised sequential learning for functional data is proposed.  ...  Florian Maire would like to thank the ONERA-The French Aerospace Lab and the DGA-The French Procurement Agency.  ... 
doi:10.1016/j.csda.2017.01.006 fatcat:iwf4xqxjknb6nnjo6gby3ysqui

A multi-class structured dictionary learning method using discriminant atom selection [article]

R.E. Rolón, L.E. Di Persia, R.D. Spies, H.L. Rufiner
2018 arXiv   pre-print
Our method was tested with a widely used database for handwritten digit recognition and compared with three state-of-the-art classification methods.  ...  The results show that our method significantly outperforms the other three achieving good recognition rates and additionally, reducing the computational cost of the classifier.  ...  The method was tested with a widely used database for handwritten digit recognition and compared with three state-of-the-art classification methods.  ... 
arXiv:1812.01389v1 fatcat:ulsl5l6qnnazplypw4jld3ltne
« Previous Showing results 1 — 15 out of 253 results