For Project Enquiry +91 9791 044 044

CROSS-DOMAIN FEATURE LEARNING IN MULTIMEDIA

ABSTRACT:

In this paper, the study about huge number of media data, such as text, image/video, and social interaction information, has been generated on the social media sites. It is difficult to design an effective feature representation to describe these data because they have multi-modal property (e.g., text, image, video, and audio) and multi-domain property. These media data can be effectively adopted for many applications (e.g., image/video annotation, image/video retrieval, and event classification) in multimedia. To deal with these issues, we propose a novel cross-domain feature learning (CDFL) algorithm based on stacked denoising auto-encoders. We propose a cross-domain feature learning (CDFL) method based on stacked denoising auto-encoders. The modal correlation constraint and the cross-domain constraint in conventional auto-encoder, our CDFL can maximize the correlations among different modalities and extract domain invariant semantic features simultaneously.

EXISTING SYSTEM	PROPOSED SYSTEM
EXISTING CONCEPT:- The Existing system in multimedia data contains useful information’s and has been adopted for many applications. Most of the existing applications use the metadata, such as time, location and descriptions, as features.  The number of photos uploaded in fixed time duration is used to predict the election winner. In only text descriptions of the video are used for recommendation. These metadata are easy to be extracted.	PROPOSED CONCEPT:- Proposed system in CDFL algorithm, sentiment classification, spam filtering, and event classifications are special cases of our CDFL. We construct events by crawling images and the corresponding texts from Multimedia.  Proposed in cross-domain correlation knowledge is used for web multimedia object.Transformation method is proposed to indirectly transfer semantic knowledge between text and images.
EXISTING TECHNIQUE:- Multi-Modal Feature Learning.	PROPOSED TECHNIQUE:- Cross-Domain Feature Learning (CDFL)
TECHNIQUE DEFNITION:-   The techniques in reduce the distance across two domains by learning a latent feature space where domain similarity is measured through maximum mean discrepancy.	TECHNIQUE DEFNITION:-   To consider the multi-modal property, in conventional denoising auto-encoders by maximizing the correlations among different modalities of media data. To reduce the domain discrepancy.
DRAWBACKS:-  No Efficient upload images focus on the        targets of a specific event perfectly. Low level features domain discrepancy there are more noises on the images.	ADVANTAGES:- Efficient and Maximize the correlations among different modality of media data. Domain discrepancy , make different domains for denoising encoder