Download presentation
Presentation is loading. Please wait.
Published byMyles Jacobs Modified over 9 years ago
1
Wei Zhang wei.zhang22@hp.com (zhangwe@eecs.oregonstate.edu) Akshat Surve survea@eecs.oregonstate.edu Xiaoli Fern xfern@eecs.oregonstate.edu Thomas Dietterich tgd@eecs.oregonstate.edu Learning Non-Redundant Codebooks for Classifying Complex Objects
2
Contents Learning codebooks for object classification Learning non-redundant codebooks Framework Boost-Resampling algorithm Boost-Reweighting algorithm Experiments Conclusions and future work 2
3
Contents Learning codebooks for object classification Learning non-redundant codebooks Framework Boost-Resampling algorithm Boost-Reweighting algorithm Experiments Conclusions and future work 3
4
Problem 1: Stonefly Recognition Cal Dor Hes Iso Mos Pte Swe Yor Zap 4
5
Visual Codebook for Object Recognition Interest Region Detector Region Descriptors Visual Codebook 20 17 3 18 2 Image Attribute Vector (Term Frequency) Classifier 6 Training image Testing image 5
6
Problem 2: Document Classification Through the first half of the 20th century, most of the scientific community believed dinosaurs to have been slow, unintelligent cold-blooded animals. Most research conducted since the 1970s, however, has supported the view that dinosaurs were active animals with elevated metabolisms and numerous adaptations for social interaction. The resulting transformation in the scientific understanding of dinosaurs has gradually filtered … 6 Variable-length Document … absent: 0 … active: 1 … animal: 2 … believe: 1 … dinosaur: 3 … social:1 … Fixed-length Bag-of-words
7
Codebook for Document Classification Cluster the words to form code-words Through the first half of the 20th century, most of the scientific community believed dinosaurs to have been slow, unintelligent cold-blooded animals. Most research conducted since the 1970s, however, has supported the view that dinosaurs were active animals with elevated metabolisms and numerous adaptations for social interaction. The resulting transformation in the scientific understanding of dinosaurs has gradually filtered … Training corpus dog, canine, hound,... cluster 1 cluster 2 car, automobile, vehicle, … … Through the first half of the 20th century, most of the scientific community believed dinosaurs to have been slow, unintelligent cold-blooded animals. Most research conducted since the 1970s, however, has supported the view that dinosaurs were active animals with elevated metabolisms and numerous adaptations for social interaction. The resulting transformation in the scientific understanding of dinosaurs has gradually filtered … codebook Input document … cluster K 201 … 02 Classifier 7
8
Contents Learning codebooks for object classification Learning non-redundant codebooks Framework Boost-Resampling algorithm Boost-Reweighting algorithm Experiments Conclusions and future work 8
9
Learning Non-Redundant Codebooks Motivation: Improve the discriminative performance of any codebook and classifier learning approach by encouraging non-redundancy in the learning process. Approach: learn multiple codebooks and classifiers; wrap the codebook and classifier learning process inside a boosting procedure [1]. Codebook Approaches: k-means, Gaussian Mixture Modeling, Information Bottleneck, Vocabulary trees, Spatial pyramid … Non-Redundant Learning [1] Freund, Y. and Schapire, R. (1996). Experiments with a new boosting algorithm. ICML. 9
10
Non-Redundant Codebook and Classifier Learning Framework Classifier C 1 Clustering X based on weights W 1 (B) X X Update boosting weights ………… Codebook D 1 Classifier C t Clustering X based on weights W t (B) X X Codebook D t Classifier C T Clustering X based on weights W T (B) X X Codebook D T W1(B)W1(B) Predictions L 1 Final Predictions L Wt(B)Wt(B) Predictions L t WT(B)WT(B) Predictions L T Update boosting weights ………… 10
11
Instantiations of the Framework Boost-Reweighting (discrete feature space): Supervised clustering features X based on the joint distribution table P t (X, Y) (Y represents the class labels). This table is updated at each iteration based on the new boosting weights. Boost-Resampling (continuous feature space): Generate a non-redundant clustering set by sampling the training examples according to the updated boosting weights. The codebook is constructed by clustering the features in this clustering set. 11
12
Codebook Learning and Classification Algorithms Documents: Codebook Learning: Information Bottleneck (IB) [1]: L = I(X ; X’) − β I(X’ ; Y) Classification: Naïve Bayes Objects: Codebook Learning: K-Means Classification: Bagged Decision Trees [1] Bekkerman, R., El-yaniv, R., Tishby, N., Winter, Y., Guyon, I. and Elisseeff, A. (2003). Distributional word clusters vs. words for text categorization. JMLR. 12
13
Image Attributes: tf − idf Weights [1] Salton, G. and Buckley, C. (1988). Term-weighting approaches in automatic text retrieval. Information Processing & Management. Term-frequency−inverse document frequency (tf−idf) weight [1]: "Document" = Image "Term" = Instance of a visual word Interest Regions Region Descriptors Visual Codebook 20 17 3 18 2 Image Attribute Vector 6 13 Classifier tf-idf
14
Contents Learning codebooks for object classification Learning non-redundant codebooks Framework Boost-Resampling algorithm Boost-Reweighting algorithm Experiments Conclusions and future work 14
15
Experimental Results − Stonefly Recognition DatasetBoostLarios [1]Opelt [2] STONEFLY297.8579.3770.10 STONEFLY498.2182.42/ [1] Larios, N., Deng, H., Zhang, W., Sarpola, M., Yuen, J., Paasch, R., Moldenke, A., Lytle, D., Ruiz Correa, S., Mortensen, E., Shapiro, L. and Dietterich, T. (2008). Automated insect identification through concatenated histograms of local appearance features. Machine Vision and Applications. [2] Opelt, A., Pinz, A., Fussenegger, M. and Auer, P. (2006). Generic object recognition with boosting. PAMI. 3-fold cross validation experiments The size of each codebook K = 100 The number of boosting iterations T = 50 15
16
Experimental Results − Stonefly Recognition (cont.) DatasetBoostSingleRandom STONEFLY297.8585.8489.16 STONEFLY498.2167.2090.42 STONEFLY995.0978.3389.07 Single: learns only a single codebook of size K×T = 5000. Random: weighted sampling is replaced with uniform random sampling that neglects the boosting weights. Boost achieves 77% error reduction comparing with Single on STONEFLY9. 16
17
Experimental Results − Stonefly Recognition (cont.) 17
18
Experimental Results − Document Classification S1000: learns a single codebook of size 1000. S100: learns a single codebook of size 100. Random: 10 bagged samples of the original training corpus are used to estimate the joint distribution table P t (X, Y). DatasetBoostRandomS1000S100 NG1090.2485.4384.3179.88 ENRON1084.4481.0980.9074.23 18
19
Experimental Results − Document Classification (cont.) [TODO]: add Figure 5 in a similar format as Figure 4 19
20
Contents Learning codebooks for object classification Learning non-redundant codebooks Framework Boost-Resampling algorithm Boost-Reweighting algorithm Experiments Conclusions and future work 20
21
Conclusions and Future Work Conclusions: Non-redundant learning is a simple and general framework to effectively improve the performance of codebooks. Future work: Explore the underlying reasons for the effectiveness of non- redundant codebooks – discriminative analysis, non-redundancy tests; More comparison experiments on well-established datasets. 21
22
Acknowledgements Supported by Oregon State University insect ID project: http://web.engr.oregonstate.edu/~tgd/bugid http://web.engr.oregonstate.edu/~tgd/bugid Supported by NSF under grant number IIS-0705765. Thank you ! 22
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.