A Max-Margin Perspective on Sparse Representation-Based Classification (original) (raw)
Related papers
On robust face recognition via sparse coding: the good, the bad and the ugly
IET Biometrics, 2014
In the field of face recognition, sparse representation (SR) has received considerable attention during the past few years, with a focus on holistic descriptors in closed-set identification applications. The underlying assumption in such SR-based methods is that each class in the gallery has sufficient samples and the query lies on the subspace spanned by the gallery of the same class. Unfortunately, such an assumption is easily violated in the face verification scenario, where the task is to determine if two faces (where one or both have not been seen before) belong to the same person. In this study, the authors propose an alternative approach to SR-based face verification, where SR encoding is performed on local image patches rather than the entire face. The obtained sparse signals are pooled via averaging to form multiple region descriptors, which then form an overall face descriptor. Owing to the deliberate loss of spatial relations within each region (caused by averaging), the resulting descriptor is robust to misalignment and various image deformations. Within the proposed framework, they evaluate several SR encoding techniques: l 1 -minimisation, Sparse Autoencoder Neural Network (SANN) and an implicit probabilistic technique based on Gaussian mixture models. Thorough experiments on AR, FERET, exYaleB, BANCA and ChokePoint datasets show that the local SR approach obtains considerably better and more robust performance than several previous stateof-the-art holistic SR methods, on both the traditional closed-set identification task and the more applicable face verification task. The experiments also show that l 1 -minimisation-based encoding has a considerably higher computational cost when compared with SANN-based and probabilistic encoding, but leads to higher recognition rates.
On robust face recognition via sparse encoding : the good, the bad, and the ugly
Science Engineering Faculty, 2013
In the field of face recognition, Sparse Representation (SR) has received considerable attention during the past few years. Most of the relevant literature focuses on holistic descriptors in closed-set identification applications. The underlying assumption in SR-based methods is that each class in the gallery has sufficient samples and the query lies on the subspace spanned by the gallery of the same class. Unfortunately, such assumption is easily violated in the more challenging face verification scenario, where an algorithm is required to determine if two faces (where one or both have not been seen before) belong to the same person. In this paper, we first discuss why previous attempts with SR might not be applicable to verification problems. We then propose an alternative approach to face verification via SR. Specifically, we propose to use explicit SR encoding on local image patches rather than the entire face. The obtained sparse signals are pooled via averaging to form multiple region descriptors, which are then concatenated to form an overall face descriptor. Due to the deliberate loss spatial relations within each region (caused by averaging), the resulting descriptor is robust to misalignment & various image deformations. Within the proposed framework, we evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder Neural Network (SANN), and an implicit probabilistic technique based on Gaussian Mixture Models. Thorough experiments on AR, FERET, exYaleB, BANCA and ChokePoint datasets show that the proposed local SR approach obtains considerably better and more robust performance than several previous state-of-the-art holistic SR methods, in both verification and closed-set identification problems. The experiments also show that l1-minimisation based encoding has a considerably higher computational than the other techniques, but leads to higher recognition rates.
Block-Diagonal Sparse Representation by Learning a Linear Combination Dictionary for Recognition
arXiv (Cornell University), 2016
In a sparse representation based recognition scheme, it is critical to learn a desired dictionary, aiming both good representational power and discriminative performance. In this paper, we propose a new dictionary learning model for recognition applications, in which three strategies are adopted to achieve these two objectives simultaneously. First, a block-diagonal constraint is introduced into the model to eliminate the correlation between classes and enhance the discriminative performance. Second, a low-rank term is adopted to model the coherence within classes for refining the sparse representation of each class. Finally, instead of using the conventional over-complete dictionary, a specific dictionary constructed from the linear combination of the training samples is proposed to enhance the representational power of the dictionary and to improve the robustness of the sparse representation model. The proposed method is tested on several public datasets. The experimental results show the method outperforms most state-of-the-art methods.
Sparse Representation and Face Recognition
International Journal of Image, Graphics and Signal Processing, 2018
Now a days application of sparse representation are widely spreading in many fields such as face recognition. For this usage, defining a dictionary and choosing a proper recovery algorithm plays an important role for the method accuracy. In this paper, two type of dictionaries based on input face images, the method named SRC, and input extracted features, the method named MKD-SRC, are constructed. SRC fails for partial face recognition whereas MKD-SRC overcomes the problem. Three extension of MKD-SRC are introduced and their performance for comparison are presented. For recommending proper recovery algorithm, in this paper, we focus on three greedy algorithms, called MP, OMP, CoSaMP and another called Homotopy. Three standard data sets named AR, Extended Yale-B and Essex University are used to asses which recovery algorithm has an efficient response for proposed methods. The preferred recovery algorithm was chosen based on achieved accuracy and run time.
Pattern Recognition of Handwritten Digits MNIST Dataset
Department of Electrical and Computer Engineering, Mississippi State University, 2021
MNIST, Modified National Institute of Standards and Technology, is the largest database of handwritten numbers used in deep learning, and machine learning. In this project, a hands-on experience of applying machine learning and pattern recognition techniques is given to a real-world data set such as MNIST. Multiple building blocks have been proposed and analyzed to improve the speed and the accuracy of the Convolutional Neural Networks (CNN). Two networks have been used with the same data. In network I, a three-layer MLP with ReLU and dropout resulting in fast training process with over all accuracy 95% during training and 94% for testing. Network II on the other hand, a stack of CNN, RelU, and Max pooling shows slower training process with better accuracy than network I and overall, 99% accuracy for training, and 98.9% for testing. Another modification on network II improved overall accuracy during the training to 99.82% and accuracy for testing to 99.25%. this modification will be shown in the report. The building blocks for the project will be discussed briefly with the results and figures. Python code is also provided for this project. This project may be used for as a guidance for new students or engineers who aiming to understand pattern recognition.
Classification based on sparse representation and euclidean distance
2012
In this paper, A novel classification approach based on sparse representation framework is proposed. The method finds the minimum Euclidian distance between an input patch (pattern) and atoms (templates) of a learnt-base dictionary for different classes to perform the classification task. A mathematical approach is developed to map the sparse representation vector to Euclidian distances. We show that the highest coefficient of the sparse vector is not necessarily a suitable indicator to classify input patches, and it results in classification errors. The K-SVD dictionary learning method is utilized to separately create class specific sub-dictionaries. The proposed algorithm is compared with the conventional sparse representation classification (SRC) framework to evaluate its performance. Our experimental results demonstrate a higher accuracy with a lower computational time.
Joint learning and dictionary construction for pattern recognition
2008 IEEE Conference on Computer Vision and Pattern Recognition, 2008
We propose a joint representation and classification framework that achieves the dual goal of finding the most discriminative sparse overcomplete encoding and optimal classifier parameters. Formulating an optimization problem that combines the objective function of the classification with the representation error of both labeled and unlabeled data, constrained by sparsity, we propose an algorithm that alternates between solving for subsets of parameters, whilst preserving the sparsity. The method is then evaluated over two important classification problems in computer vision: object categorization of natural images using the Caltech 101 database and face recognition using the Extended Yale B face database. The results show that the proposed method is competitive against other recently proposed sparse overcomplete counterparts and considerably outperforms many recently proposed face recognition techniques when the number training samples is small.
Indian Journal of Pharmaceutical Sciences
Over the past 20 y, face recognition has drawn the attention of many researchers in various fields, such as security, psychology and engineering. On the other hand, with the advancement of technology, interactions between humans and computers will also increase. One of the key steps in this interaction is face recognition [1]. Face recognition is one of the important tools for identification in the field of biometrics and various algorithms have been proposed in relation to it. Although most of these have experienced significant advances in various applications and research areas, they still face major challenges such as lightness, gesture, expression and occlusion. There are a lot of ways in this area, but one important point that should always be considered is which feature contains very important information for identification? According to the geometry and appearance of the face, where fixed filter banks such as down sampling, Fourier, wavelet and Gabor are not used, which are suitable tools for the analysis of static signals like texture, but instead methods that adaptively extract the facial features based on the given pictures (e.g. Eigen face [2] , Fisher faces [3] and Laplacian faces [4]) are used (fig. 1). The face recognition can be performed by using these features and designing a classifier such as the nearest-neighbor (NN), the nearest-subspace (NS) or SVM. Considering the process of designing the face recognition algorithm, it is seen that the performance of the algorithm is dependent on 2 parts,
Classification based on sparse representation and Euclidian distance
2012 Visual Communications and Image Processing, 2012
In this paper, A novel classification approach based on sparse representation framework is proposed. The method finds the minimum Euclidian distance between an input patch (pattern) and atoms (templates) of a learnt-base dictionary for different classes to perform the classification task. A mathematical approach is developed to map the sparse representation vector to Euclidian distances. We show that the highest coefficient of the sparse vector is not necessarily a suitable indicator to classify input patches, and it results in classification errors. The K-SVD dictionary learning method is utilized to separately create class specific sub-dictionaries. The proposed algorithm is compared with the conventional sparse representation classification (SRC) framework to evaluate its performance. Our experimental results demonstrate a higher accuracy with a lower computational time.
Sparse representation for image classification by using feature dictionary
SPIE Proceedings, 2013
Sparse coding technique is usually applied for feature representation. To learn discriminative features for visual recognition, a dictionary learning method, called Paired Discriminative K-SVD (PD-KSVD), is presented in this paper. Firstly, to reduce the reconstruction error of positive class while increasing the errors of negative classes, the scheme inverted signal is applied to the negative training samples. Then, the class-specific sub-dictionaries are learned from pairs of positive and negative classes to jointly achieve high discrimination and low reconstruction errors for sparse coding. Multiple sub-dictionaries are concatenated with respect to the same negative class so that the non-zero sparse coefficients can be discriminatively distributed to improve classification accuracy. Last, sparse coefficients are solved via the concatenated sub-dictionaries and used to train the classifier. Compared to the existing dictionary learning methods, PD-KSVD method achieves superior performance in a variety of visual recognition tasks on several publicly available datasets.