Multimedia Data Mining

Jelena Tešic Advisor: B.S. Manjunath
Vision Research Laboratory Department of Electrical and Computer Engineering University of California, Santa Barbara

Multimedia Database Managemnet 2 .Data Mining n Data Mining definition: n n A class of database applications that look for hidden patterns in a group of data. Finding rules of the game knowing the moves of the game n Unifying framework for data representation and problem solving in order to learn and discover from large amounts of different types of data.

relational. images. n Relation with conventional data mining term Multimedia Database Managemnet 3 . temporal. audio. graphical. numeric. video. processed. stored and transmitted over network in digital form Multi-lingual text. and categorical data.Multimedia Data Mining n Multimedia data types n n any type of information medium that can be represented.

Definitions n n Subfield of data mining that deals with an extraction of implicit knowledge. multimedia data relationships. or other patterns not explicitly stored in multimedia databases Influence on related interdisciplinary fields n n Databases – extension of the KDD (rule patterns) Information systems – multimedia information analysis and retrieval – content-based image and video search and efficient storage organization Multimedia Database Managemnet 4 .

Case-base reasoning n Case representations n Structured (KDD applications) n n Object-oriented Relational attribute-value case Limited expressive power Collection of case descriptors Links – connect information within case Multimedia Database Managemnet 5 n Unstructured (multimedia) n n n .

represented by different views Domain decomposition n Complex case is represented as multiple cases n n Hierarchy structure supports human reasoning Automated process n n n Structured representation layer Vector of case attributes Identify attributes Multimedia Database Managemnet 6 .Case representation n n Hierarchy of concepts.

Case Library Layern Layern-1 ………………………………………………… Case Case Case Case Layer0 Multimedia Database Managemnet 7 .

Knowledge Discovery in Multimedia Databases n n Find patterns in primarily unstructured data Machine learning where a case library replaces the training set Conventional Knowledge Case Library Data Mining Discovered Knowledge Multimedia Database Managemnet 8 .

Information model n Data segmentation n Multimedia data are divided into logical interconnected segments (objects) Mining and analysis procedures should reveal some relations between objects on the different level Incorporated linked patterns n Pattern extraction n n Knowledge representation n n Information model – dynamic structure Multimedia Database Managemnet 9 .

Multimedia Mining Hierarchy Image Video Audio Multimedia Data Data segmentation Pattern extraction Object-based representation Feature extraction Additional information Case (event) definition Knowledge representation Information modeling Multimedia Database Managemnet 10 .

Importance of Case-base reasoning n n Finding patterns based on the specific interest Previous experience n n n Assist with indexing and adapting cases to improve retrieval Indication when the adaptation lies outside some reasonable experience Dynamic thematic paths in the hierarchy can assist with navigation in the retrieved cases n Learning loop of the case-based reasoning Multimedia Database Managemnet 11 .

Advantages n Generation of indexing schemes. based on n n n the related terms to regularities discovered in other media types (semantic extraction) Structural patterns discovered in multimedia (graph indexing) One case library and its dynamic nature n n Retrieval – flexibility in formulating queries Adaptation of the new case description based on the user’s feedback Multimedia Database Managemnet 12 .

Advantages – cont’d n n Case-based mechanism provides incorporation and management of the discovered knowledge Multimedia data mining can improve the case-based system n Discover of unknown patterns n Modular approach to the case-base reasoning multimedia data mining model Multimedia Database Managemnet 13 .

Modular approach n n Developer n n Implementation of any data segmentation and data mining method Adaptation of the stored knowledge Online processing (photo collection) Automatic classification Real time complex query response Feedback Multimedia Database Managemnet 14 n User n n n n .

System implementation n n Pattern recognition for larger image databases (Toshiba) Content-based retrieval n n Relationship among features User’s feedback (feature weights) System prototype Multimedia Database Managemnet 15 n MultiMedia Miner (Han. SFU. CA) n .

MultiMedia Miner n Multimedia Data Cube n n n n Image Excavator (Extraction of images) Preprocessor .Feature extractor User interface Search engine characterizer. associator Multimedia Database Managemnet 16 n Multimedia Miner n n . comparator classifier.

Sixth ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. August 20-23. CA Multimedia Database Managemnet 17 . AHPCR Center. May 1-2. MN Workshop on Data Mining in the Internet Age. IBM Almaden. Boston. MA Workshop on Mining Scientific Datasets. San Jose. Minneapolis.Related workshops in 2000 n n n Workshop on Multimedia Data Mining. July 20-21.

ca/~zaiane/mdm_kdd2000/ Multimedia Database Managemnet 18 http://www.cs.sfu.Conclusion n Multimedia data mining n n New methodologies Influence on the related fields n n n http://vision.ece.ucsb.cs.

Sign up to vote on this title
UsefulNot useful