50,944 Hits in 4.2 sec

Training Binary Classifiers as Data Structure Invariants

Facundo Molina, Renzo Degiovanni, Pablo Ponzio, German Regis, Nazareno Aguirre, Marcelo Frias
2019 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE)  
The technique is based on building an artificial neural network, more precisely a binary classifier, and training it to identify valid and invalid instances of a data structure.  ...  The obtained classifier can then be used in place of the data structure's invariant, in order to attempt to identify (in)correct behaviors in programs manipulating the structure.  ...  The problem we deal with in this paper, namely the approximation of a class invariant to classify valid vs. invalid data structure objects, clearly falls in the category of binary classification: we want  ... 
doi:10.1109/icse.2019.00084 dblp:conf/icse/MolinaDPRAF19 fatcat:aznovd4qlbadla7vbbvdax7ske

A cascaded ensemble classifier for object segmentation in high resolution polarimetric SAR data

Marc Jager, Andreas Reigber, Olaf Hellwich
2014 2014 IEEE Geoscience and Remote Sensing Symposium  
and across the object category as a whole.  ...  The classifier structure is based on a combination of techniques developed for related problems in computer vision: the cascade architecture helps breaking down the problem into manageable stages while  ...  The classifier was trained and evaluated on distinct regions of the same data acquisition.  ... 
doi:10.1109/igarss.2014.6946603 dblp:conf/igarss/JagerRH14 fatcat:5khxfyvmvjhbvkgupcj5z43p2y

Diabetic Retinopathy and Age Related Macular De-Generation Diseases Screening Using Local Binary Patterns Approach

Bhargavi K.V
2017 International Journal for Research in Applied Science and Engineering Technology  
Local binary patterns (LBP) is used as a main texture descriptor that provide generalizations to the gray scale and rotation invariant texture classification method.  ...  For each experiment support vector machine and neural classifiers were used.  ...  learning given labelled training data.  ... 
doi:10.22214/ijraset.2017.9102 fatcat:bq57j3cxdjedxbbna3xpusmbam

HDL Based Illumination Invariant High Performance Face Detection System for Mobile Applications

T. Suguna, Y. Mahesh
2013 i-manager s Journal on Communication Engineering and Systems  
First we introduce illumination invariant Local Structure Features for face detection.  ...  Secondly we introduceanefficient face detection classifier for rapid detection to render high performance face detection rate.The Classifier structure is much simpler because we use only single stage classifier  ...  Fig. 2 .Fig. 3 . 23 Example.Converting binary patterns to binary numbers using MCT Example of the illumination invariance of the Modified MCT transformed test imagesFigure 2shows an example of structure  ... 
doi:10.26634/jcs.2.2.2244 fatcat:2tuatpfu4zgsxjb35y4wnnleey

A hierarchical classifier using new support vector machines for automatic target recognition

David Casasent, Yu-Chiang Wang
2005 Neural Networks  
Using this hierarchical SVRDM classifier with magnitude Fourier transform (jFTj) features, which provide shift-invariance, initial test results on infrared (IR) data are excellent. q  ...  A binary hierarchical classifier is proposed for automatic target recognition.  ...  Hierarchical classifier structures (balanced binary hierarchy design) using 25!25 pixel jFTj features (a) with preprocessed data (b) without using preprocessed data.  ... 
doi:10.1016/j.neunet.2005.06.033 pmid:16087318 fatcat:o6mercsj3jgp7apzau4zu56ife

Counterfactual Generative Networks [article]

Axel Sauer, Andreas Geiger
2021 arXiv   pre-print
Lastly, our generative model can be trained efficiently on a single GPU, exploiting common pre-trained models as inductive biases.  ...  In this work, we take a step towards more robust and interpretable classifiers that explicitly expose the task's causal structure.  ...  We then generate counterfactual data to train three classifiers: • Shape classifier (SC): invariant wrt. object color and background color • Object color classifier (OCC): invariant wrt. object shape and  ... 
arXiv:2101.06046v1 fatcat:apykwyzkevee5hiii23lo2dpae

Fourier Descriptor based Isolated Marathi Handwritten Numeral Recognition

G. G. Rajput, S. M. Mali
2010 International Journal of Computer Applications  
These classifiers are trained with 64 dimensional Fourier Descriptors (FD) of training samples.  ...  Fourier Descriptors that describe the shape of Marathi handwritten numerals are used as feature. 64 dimensional Fourier Descriptors represents the shape of numerals, invariant to rotation, scale and translation  ...  Given a training database of M data: {xm| m=1,...  ... 
doi:10.5120/724-1017 fatcat:vixrinsb3zcozl2i5n7r6le77m

Circular Blurred Shape Model for Multiclass Symbol Recognition

S Escalera, A Fornés, O Pujol, J Lladós, P Radeva
2011 IEEE Transactions on Systems Man and Cybernetics Part B (Cybernetics)  
In the case of multiclass categorization, the new feature space is learned using a set of binary classifiers which are embedded in an error-correcting output code design.  ...  The feature extraction is performed by capturing the spatial arrangement of significant object characteristics in a correlogram structure.  ...  The first stage (namely, the training) should learn to distinguish among the target object and background (i.e., learning a binary classifier).  ... 
doi:10.1109/tsmcb.2010.2060481 pmid:20729173 fatcat:nlgz36csoffh7bzbtyqdo5yrpq

Recognition of Aerial Insulator Image Based on Structural Model and the Optimal Entropy Threshold Segmentation

Yongjie Zhai, Di Wang, Yu Guo, Muliu Zhang, Yang Liu
2017 DEStech Transactions on Engineering and Technology Research  
Finally, the moment invariants of the insulator and background are calculated to train classifiers based on Adaboost Algorithm, and then a strong classifier is created.  ...  Those insulator simulation images are used to build the training sample set. Then the insulator image is grayed and enhanced.  ...  After image being preprocessed, we extract invariant moment eigenvalue of insulator and background regions as training data. Finally we train a classifier using Adaboost algorithm.  ... 
doi:10.12783/dtetr/iceta2016/7036 fatcat:ftmmsnphmba5pnttwwbjve7ati

Shape Quantization and Recognition with Randomized Trees

Yali Amit, Donald Geman
1997 Neural Computation  
We explore a new approach to shape recognition based on a virtually infinite family of binary features (queries) of the image data, designed to accommodate prior information about shape invariance and  ...  The important attributes of the queries are a natural partial ordering corresponding to increasing structure and complexity; semi-invariance, meaning that most shapes of a given class will answer the same  ...  Recall that the entropy values are estimated from training data and that Q m is binary.  ... 
doi:10.1162/neco.1997.9.7.1545 fatcat:nchexof4mndo7nr4kclbjiknje

Joint Real-time Object Detection and Pose Estimation Using Probabilistic Boosting Network

Jingdan Zhang, Shaohua Kevin Zhou, Leonard McMillan, Dorin Comaniciu
2007 2007 IEEE Conference on Computer Vision and Pattern Recognition  
We implement PBN using a graph-structured network that alternates the two tasks of foreground/background discrimination and pose estimation for rejecting negatives as quickly as possible.  ...  Grounded on the law of total probability, PBN integrates evidence from two building blocks, namely a multiclass boosting classifier for pose estimation and a boosted detection cascade for object detection  ...  This can be achieved by pooling together all data from different poses as positives to train a poseinvariant classifier as a pre-filter.  ... 
doi:10.1109/cvpr.2007.383275 dblp:conf/cvpr/ZhangZMC07 fatcat:j3ahhvx46ngc5feyiy2ipzrqgq

A General Framework for Fast 3D Object Detection and Localization Using an Uncalibrated Camera

Andres Solis Montero, Jochen Lang, Robert Laganiere
2015 2015 IEEE Winter Conference on Applications of Computer Vision  
Our algorithm exploits the specific structure of various binary descriptors in order to boost feature matching by conserving descriptor properties (e.g., rotational and scale invariance, robustness to  ...  We develop our algorithm using a feature-based method based on two novel naive Bayes classifiers for viewpoint and feature matching.  ...  Acknowledgment This research was partly funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) under the Engage program with You.i Labs as the industrial partner.  ... 
doi:10.1109/wacv.2015.122 dblp:conf/wacv/MonteroLL15 fatcat:w5z62pk5und65naxc26xvwr5bq

Invariant Descriptors and Classifiers Combination for Recognition of Isolated Printed Tifinagh Characters

2013 International Journal of Advanced Computer Science and Applications  
The Legendre moments, Zernike moments and Hu moments are used as descriptors in the features extraction phase due to their invariance to translation, rotation and scaling changes.  ...  In the classification phase, the neural network, the multiclass SVM (Support Vector Machine) and the nearest neighbour classifiers are combined together.  ...  Each SVM binary classifier is trained using a matrix of training data, where each row corresponds to features extracted as an observation from a class.  ... 
doi:10.14569/specialissue.2013.030205 fatcat:5crsll2erbbi5nryqyxddoxq7e

Classifying Similarity and Defect Fabric Textures based on GLCM and Binary Pattern Schemes

R. Obula Konda Reddy, B. Eswara Reddy, E. Keshava Reddy
2013 International Journal of Information Engineering and Electronic Business  
Then a new rotation-invariant, scale invariant steerable decomposition filter is applied to filter the four orientation sub bands of the image.  ...  Finally the texture is classified by different classifiers (PNN, K-NN and SVM) and the classification performance of each classifier is compared.  ...  They showed that the number of training images required can be drastically reduced (to as few as three) by synthesizing additional training data using photometric stereo.  ... 
doi:10.5815/ijieeb.2013.05.04 fatcat:52kvef5mtbcnzjuantx3mjff7y

Heterogeneous Domain Adaptation for Multiple Classes

Joey Tianyi Zhou, Ivor W. Tsang, Sinno Jialin Pan, Mingkui Tan
2014 International Conference on Artificial Intelligence and Statistics  
We cast this learning task as a compressed sensing problem, where each binary classifier induced from multiple classes can be deemed as a measurement sensor.  ...  Therefore, to guarantee reconstruction performance, we construct sufficiently many binary classifiers based on the error correcting output coding.  ...  The reason is that the size of labeled training data is too limited to train a precise and stable classifier in the target domain.  ... 
dblp:conf/aistats/ZhouTPT14 fatcat:rv4nym27ozaevd6rxycpkrs4dq
« Previous Showing results 1 — 15 out of 50,944 results