Key words
(English)
|
Large-scale indexing; multimodal information processing; information retrieval; scalability; machine learning
|
Research programs
(English)
|
COST-Action IC1002 - Multilingual and multifaceted interactive information access (MUMIA)
|
Short description
(English)
|
Via this project, we wish to offer scalable effectiveness while preserving exible and cross-modal search functionality. In other words, we will design large-scale collection indexing strategies that match efficient learning-based multimedia retrieval strategies. The objective is to detail and analyse learning algorithms enabling interactive multimodal search to understand their bottlenecks in terms of data access and processing to finally integrate these methods with suitable largescale data indexing. This way, we will construct retrieval engines immune to the increase of data scale and complexity Two complementary approaches will be followed for achieving our objectives. We will first jointly investigate indexing and learning strategies to fit them onto a distributed environment. In order to complete our objective of true scalability with respect to data volume and complexity, we will investigate in the direction of matching approximate search and retrieval algorithms onto a distributed context. On top of producing principled research results, the project will deliver a working scalable multimedia search engine capable of indexing large-scale collections, as an evolution of our current Cross-modal Search Engine.
|
Further information
(English)
|
Full name of research-institution/enterprise: Université de Genève Laboratoire Vision par ordinateur et multimédia Battelle (Bat A)
|
Partners and International Organizations
(English)
|
AT, BG, CH, DE, DK, EL, ES, FI, FR, HR, MK, IE, IT, NL, NO, RS, SI, UK
|
Abstract
(English)
|
Classical challenges in the design of multimedia information systems are now augmented by the growth of multimedia collections in size, complexity and diversity of the data, and complexity of the information networks. It is now crucial to propose solutions to enable indexing and retrieval of multimedia data for large-scale collections and in a distributed context of data processing and access. We identify two main complementary directions: Large-scale multi-modal information indexing: While large-scale indexing structures exist, the main challenge here is to adapt or exploit these techniques into a distributed multimodal environment, i.e., where concurrent access to the same data must be performed following its various facets. Scalability imposes the use of approximate access structures such as embedding structures or metric-tree; Large-scale multi-modal information retrieval: Building on the above, retrieval strategies must be constructed and adapted so as to handle multimodality. Effectively distributed learning strategies must be defined, that are coherent with available distributed data access modes. Via this project, we wish to offer scalable effectiveness while preserving flexible and cross-modal search functionality. In other words, we will design large-scale collection indexing strategies that match efficient learning-based multimedia retrieval strategies. The objective is to detail and analyze learning algorithms enabling interactive multimodal search to understand their bottlenecks in terms of data access and processing to finally integrate these methods with suitable large-scale data indexing. This way, we will construct retrieval engines immune to the increase of data scale and complexity. Two complementary approaches will be followed for achieving our objectives. We will first jointly investigate indexing and learning strategies to fit them onto a distributed environment. In order to complete our objective of true scalability with respect to data volume and complexity, we will investigate in the direction of matching approximate search and retrieval algorithms onto a distributed context. On top of producing principled research results, the project will deliver a working scalable multimedia search engine capable of indexing large-scale collections, as an evolution of our current Cross-modal Search Engine.
|
References in databases
(English)
|
Swiss Database: COST-DB of the State Secretariat for Education and Research Hallwylstrasse 4 CH-3003 Berne, Switzerland Tel. +41 31 322 74 82 Swiss Project-Number: C11.0043
|
