Data Classification Methods Based On Analysis Dictionary Learning

In the rapid development of information technology and gradual expanding of data capacity,"information flooding" and "data explosion" often appear.In order to quickly obtain valuable information from massive data,it is necessary to study efficient and applicable classification algorithms.In recent years,scholars have started to study dictionary learning method and achieved promising performance.However,it is found that most of these algorithms are based on the synthetic dictionary,which is inefficient and difficult to understand.Therefore,the analysis dictionary begins to attract the attention of scholars.The analysis dictionary can provide a more intuitive explanation for the encoding process and improve the running efficiency,but it has a poor classification performance.In this paper,we present two novel discriminative analysis dictionary learning frameworks and validate their classification performance.Main contributions of this paper lie in two parts:(1)A novel discriminative analysis dictionary learning frame,named Synthesis Linear Classifier based Analysis Dictionary Learning(SLC-ADL),is presented.Firstly,we incorporate a synthesis-linear-classifier-based error term into the basic analysis dictionary learning model,which makes full use of the label information.Then,we develop an alternating iterative algorithm to solve the new model and obtain closed-form solutions leading to pretty competitive running efficiency.What is more,we design three classification schemes by fully exploiting the synthesis linear classifier.Finally,extensive comparison experiments on scene categorization,object classification,action recognition and face recognition clearly verify the classification performance of the proposed algorithm.(2)The Class-aware Analysis Dictionary Learning(CADL)model is proposed to improve the classification performance of conventional ADL.The objective function of CADL mainly includes two parts to promote the discriminability.The first part aims to learn a discriminative analysis sub-dictionary for each class instead of a global dictionary for all classes.The second part aims to enhance the discrimination of coding coefficients by integrating a max-margin regularization term into our proposed framework.By performing comparative experimental analysis on four pattern classification datasets,we demonstrate the superiority of our CADL method to the state-of-the-art DL methods.