Abstract:Although AdaBoost has achieved great success, it still suffers from following problems: (1) the training process could be unmanageable when the number of features is extremely large; (2) the same weak classifier may be learned multiple times from a weak classifier pool, which does not provide additional information for updating the model; (3) there is an imbalance between the amount of the positive samples and that of the negative samples for multi-class classification problems. In this paper, we propose a two-stage AdaBoost learning framework to select and fuse the discriminative feature effectively. Moreover, an improved AdaBoost algorithm is developed to select weak classifiers. Instead of boosting in the original feature space, whose dimensionality is usually very high, multiple feature subspaces with lower dimensionality are generated. In the first stage, boosting is carried out in each subspace. Then the trained classifiers are further combined with simple fusion method in the second stage. Experimental results on facial expression recognition data demonstrate that our proposed algorithms not only reduce the computational cost for training, but also achieve comparable classification performance.

Speech Characteristic Signal Recognition Based on BP_AdaBoost

AdaBoost-based Multi-attribute Classification Technology and Its Application

Multi-class AdaBoost ELM

Multiclass AdaBoost ELM and Its Application in LBP Based Face Recognition

Vibration Forecasting Using BP Neural Network for Image Stabilization and an Improving Method

Two-Stage Multi-Class Adaboost for Facial Expression Recognition

An improved algorithm of BP-Adaboost and application of radar multi-target classification

Prediction of Improved BP Neural Network by Adaboost Algorithm

A medical diagnosis model based on BP-AsymBoost algorithm

Improved AdaBoost Method to Increase Recognition Rate of Traditional Lip-Reading System Based on HMM Method

Analysis of NIR Using Boosting Algorithm

Improved AdaBoost Algorithm Using VQMAP for Speaker Identification

A new method for general work piece recognition based on Neural Network

Human action recognition based on multi-feature fusion and hierarchical BP-AdaBoost algorithm

Improvements and applications of BP neural network algorithm in radar target recognition

Boosting Tail Neural Network for Realtime Custom Keyword Spotting

Research on Speech Recognition Methods Based on Spectral Subtraction and Improved BP Neural Network Algorithm

Improved Pso_adaboost Ensemble Algorithm For Imbalanced Data

Application of Improved BP Neural Network Based on Matlab

Failure Supervising Method for Electricity Meter Based on BP-AdaBoost Algorithm

Fault Signal Classification Using Adaptive Boosting Algorithm