Efficient nonlinear classification via low-rank regularised least squares
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2013
- Type: Text , Journal article
- Relation: Neural Computing and Applications Vol. 22, no. 7-8(2013), p. 1279-1289
- Full Text: false
- Reviewed:
- Description: We revisit the classical technique of regularised least squares (RLS) for nonlinear classification in this paper. Specifically, we focus on a low-rank formulation of the RLS, which has linear time complexity in the size of data set only, independent of both the number of classes and number of features. This makes low-rank RLS particularly suitable for problems with large data and moderate feature dimensions. Moreover, we have proposed a general theorem for obtaining the closed-form estimation of prediction values on a holdout validation set given the low-rank RLS classifier trained on the whole training data. It is thus possible to obtain an error estimate for each parameter setting without retraining and greatly accelerate the process of cross-validation for parameter selection. Experimental results on several large-scale benchmark data sets have shown that low-rank RLS achieves comparable classification performance while being much more efficient than standard kernel SVM for nonlinear classification. The improvement in efficiency is more evident for data sets with higher dimensions.
Learning sparse kernel classifiers for multi-instance classification
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2013
- Type: Text , Journal article
- Relation: IEEE Transactions on Neural Networks and Learning Systems Vol. 24, no. 9 (2013), p. 1377-1389
- Full Text: false
- Reviewed:
- Description: We propose a direct approach to learning sparse kernel classifiers for multi-instance (MI) classification to improve efficiency while maintaining predictive accuracy. The proposed method builds on a convex formulation for MI classification by considering the average score of individual instances for bag-level prediction. In contrast, existing formulations used the maximum score of individual instances in each bag, which leads to nonconvex optimization problems. Based on the convex MI framework, we formulate a sparse kernel learning algorithm by imposing additional constraints on the objective function to enforce the maximum number of expansions allowed in the prediction function. The formulated sparse learning problem for the MI classification is convex with respect to the classifier weights. Therefore, we can employ an effective optimization strategy to solve the optimization problem that involves the joint learning of both the classifier and the expansion vectors. In addition, the proposed formulation can explicitly control the complexity of the prediction model while still maintaining competitive predictive performance. Experimental results on benchmark data sets demonstrate that our proposed approach is effective in building very sparse kernel classifiers while achieving comparable performance to the state-of-the-art MI classifiers.
Optimizing cepstral features for audio classification
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2013
- Type: Text , Conference paper
- Relation: International Joint Conference on Artificial Intelligence p. 1330-1336
- Full Text: false
- Reviewed:
- Description: Cepstral features have been widely used in audio applications. Domain knowledge has played an important role in designing different types of cepstral features proposed in the literature. In this paper, we present a novel approach for learning optimized cepstral features directly from audio data to better discriminate between different categories of signals in classification tasks. We employ multi-layer feedforward neural networks to model the cepstral feature extraction process. The network weights are initialized to replicate a reference cepstral feature like the mel frequency cepstral coefficient. We then propose a embedded approach that integrates feature learning with the training of a support vector machine (SVM) classifier. A single optimization problem is formulated where the feature and classifier variables are optimized simultaneously so as to refine the initial features and minimize the classification risk. Experimental results have demonstrated the effectiveness of the proposed feature learning approach, outperforming competing methods by a large margin on benchmark data.
Building sparse support vector machines for multi-instance classification
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2011
- Type: Text , Conference paper
- Relation: European Conference on Machine Learning Knowledge Discovery in Databases (ECML PKDD) p. 471-486
- Full Text: false
- Reviewed:
- Description: We propose a direct approach to learning sparse Support Vector Machine (SVM) prediction models for Multi-Instance (MI) classification. The proposed sparse SVM is based on a “label-mean” formulation of MI classification which takes the average of predictions of individual instances for bag-level prediction. This leads to a convex optimization problem, which is essential for the tractability of the optimization problem arising from the sparse SVM formulation we derived subsequently, as well as the validity of the optimization strategy we employed to solve it. Based on the “label-mean” formulation, we can build sparse SVM models for MI classification and explicitly control their sparsities by enforcing the maximum number of expansions allowed in the prediction function. An effective optimization strategy is adopted to solve the formulated sparse learning problem which involves the learning of both the classifier and the expansion vectors. Experimental results on benchmark data sets have demonstrated that the proposed approach is effective in building very sparse SVM models while achieving comparable performance to the state-of-the-art MI classifiers.
Music classification via the bag-of-features approach
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2011
- Type: Text , Journal article
- Relation: Pattern Recognition Letters Vol. 32, no. 14 (2011), p. 1768-1777
- Full Text: false
- Reviewed:
- Description: A central problem in music information retrieval is audio-based music classification. Current music classification systems follow a frame-based analysis model. A whole song is split into frames, where a feature vector is extracted from each local frame. Each song can then be represented by a set of feature vectors. How to utilize the feature set for global song-level classification is an important problem in music classification. Previous studies have used summary features and probability models which are either overly restrictive in modeling power or numerically too difficult to solve. In this paper, we investigate the bag-of-features approach for music classification which can effectively aggregate the local features for song-level feature representation. Moreover, we have extended the standard bag-of-features approach by proposing a multiple codebook model to exploit the randomness in the generation of codebooks. Experimental results for genre classification and artist identification on benchmark data sets show that the proposed classification system is highly competitive against the standard methods.
On low-rank regularized least squares for scalable nonlinear classification
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2011
- Type: Text , Conference paper
- Relation: International Conference on Neural Information Processing p. 490-499
- Full Text: false
- Reviewed:
- Description: In this paper, we revisited the classical technique of Regularized Least Squares (RLS) for the classification of large-scale nonlinear data. Specifically, we focus on a low-rank formulation of RLS and show that it has linear time complexity in the data size only and does not rely on the number of labels and features for problems with moderate feature dimension. This makes low-rank RLS particularly suitable for classification with large data sets. Moreover, we have proposed a general theorem for the closed-form solutions to the Leave-One-Out Cross Validation (LOOCV) estimation problem in empirical risk minimization which encompasses all types of RLS classifiers as special cases. This eliminates the reliance on cross validation, a computationally expensive process for parameter selection, and greatly accelerate the training process of RLS classifiers. Experimental results on real and synthetic large-scale benchmark data sets have shown that low-rank RLS achieves comparable classification performance while being much more efficient than standard kernel SVM for nonlinear classification. The improvement in efficiency is more evident for data sets with higher dimensions.
Learning naive Bayes classifiers for music classification and retrieval
- Authors: Fu, Zhouyu , Lu, Guojun , Ting, Kaiming , Zhang, Dengsheng
- Date: 2010
- Type: Text , Conference paper
- Relation: Proceedings of the 20th International Conference on Pattern Recognition p. 4589-4592
- Full Text: false
- Reviewed:
- Description: In this paper, we explore the use of naive Bayes classifiers for music classification and retrieval. The motivation is to employ all audio features extracted from local windows for classification instead of just using a single song-level feature vector produced by compressing the local features. Two variants of naive Bayes classifiers are studied based on the extensions of standard nearest neighbor and support vector machine classifiers. Experimental results have demonstrated superior performance achieved by the proposed naive Bayes classifiers for both music classification and retrieval as compared to the alternative methods.