An Ensemble Based Evolutionary Approach to the Class Imbalance Problem with Applications in CBIR

Abstract

Abstract:In order to lower the dependence on textual annotations for image searches, the content based image retrieval (CBIR) has become a popular topic in computer vision. A wide range of CBIR applications consider classification techniques, such as artificial neural networks (ANN), support vector machines (SVM), etc. to understand the query image content to retrieve relevant output. However, in multi-class search environments, the retrieval results are far from optimal due to overlapping semantics amongst subjects of various classes. The classification through multiple classifiers generate better results, but as the number of negative examples increases due to highly correlated semantic classes, classification bias occurs towards the negative class, hence, the combination of the classifiers become even more unstable particularly in one-against-all classification scenarios. In order to resolve this issue, a genetic algorithm (GA) based classifier comity learning (GCCL) method is presented in this paper to generate stable classifiers by combining ANN with SVMs through asymmetric and symmetric bagging. The proposed approach resolves the classification disagreement amongst different classifiers and also resolves the class imbalance problem in CBIR. Once the stable classifiers are generated, the query image is presented to the trained model to understand the underlying semantic content of the query image for association with the precise semantic class. Afterwards, the feature similarity is computed within the obtained class to generate the semantic response of the system. The experiments reveal that the proposed method outperforms various state-of-the-art methods and significantly improves the image retrieval performance.

Deeply Learned Pose Invariant Image Analysis with Applications in 3D Face Recognition

“…We can use the recognition accuracies in test phase as a given PFI is first converted into MVAHF images oriented at 0 ∘ , 10 ∘ , 20 ∘ , and 30 ∘ . Then, each of the mentioned MVAHF images is classified against the gallery and leads to four recognition accuracies which are subsequently used to compute the weights in equation (5). This procedure is similar as employed for each of the training images in the training phase.…”

“…The matching score matrix was again normalized as using the min-max rule as given in equation (3). (5) The normalized matching scores obtained from were utilized in the Softmax layer of the AlexNet to compute the final recognition accuracies. (6) The whole process was repeated to classify MVWF, MVLHF, and MVRHF images.…”

“…It has become a fertile field for researchers due to various applications in computer vision and pattern recognition [2]. It has found numerous real-time applications in access control, security, surveillance, criminal identification, fraud detection, and even in human computer interaction [3,4] based on machine learning algorithms [5]. The advantage of face recognition over other biometric modalities is noninvasive acquisition, social acceptance, and appropriateness for noncooperative scenarios [6].…”

Machine Learning Paradigm towards Content Based Image Retrieval on High Resolution Satellite Images

“…Recent researches reported that classification accuracy can be enhanced through ensemble method. This method combines a set of weak learner models into a strong model [8] [9]. In machine learning paradigm, most popular ensemble strategies are Bagging, random subspace and boosting.…”

“…This technique exploit linear and polynomial kernel function. Aun Irtaza et al [9] offered a genetic algorithm (GA) primarily based classifier comity learning (GCCL) scheme to discover fixed classifiers because of combining ANN with SVMs by asymmetric and symmetric bagging algorithm.…”