CIF: Small: Theory and Algorithms for Statistical Content Identification

CIF：小：统计内容识别的理论和算法

• 批准号: 1219145
• 负责人: Pierre Moulin
• 金额: $ 44.79万
• 依托单位: University of Illinois at Urbana-Champaign
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-08-01 至 2017-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1219145&HistoricalAwards=false
• 关键词: CIF; Small; Theory; Algorithms; Statistical

Automatic Content Identification is an emerging technology that has found applications to broadcast monitoring, connected audio, content tracking, digital asset management, near-duplicate identification, contextual advertising, and as a filtering technology for file sharing. Content identification algorithms must be robust to common signal degradations. They operate on highly compressed data (robust hashes, aka content fingerprints) to meet storage, communication, and computing constraints.The goal of this project is to develop an analytical framework for content identification based upon fundamental principles and modern methods of statistical inference and information theory and to develop novel content identification algorithms. The project focuses on the following four research topics:1. Hash-Based Inference.2. Information-Theoretic Analysis: Content identification is formulated as a communication problem with storage constraints and its fundamental performance limits are investigated.3. Code Design: A learning-theoretic approach is developed for statistical modeling of content fingerprints and degradation channels from training data, and for designing hashing codes and decoding metrics that are optimally matched to these statistics.4. Applications: to audio, images, and video are explored, as well as forensic analysis and security.The project is synergistic and trains graduate students for leadership roles in information technology. The project benefits other areas, including but not limited to content retrieval, clustering, database indexing, pattern recognition, biometrics, and human/computer interaction.

自动内容识别是一项新兴技术，已应用于广播监控、连接音频、内容跟踪、数字资产管理、近似重复识别、上下文广告以及作为文件共享的过滤技术。内容识别算法必须对常见的信号降级具有鲁棒性。他们操作高度压缩的数据（鲁棒散列，又名内容指纹），以满足存储，通信和计算的约束，该项目的目标是开发一个分析框架的基础上的基本原则和现代方法的统计推断和信息论的内容识别和开发新的内容识别算法。本项目主要围绕以下四个方面展开研究：1.基于哈希的推理。2.信息理论分析：将内容识别问题转化为一个具有存储约束的通信问题，研究了内容识别的基本性能限制.代码设计：提出了一种学习理论方法，用于从训练数据中对内容指纹和退化通道进行统计建模，并设计与这些特征最佳匹配的哈希码和解码度量.应用领域：音频，图像和视频的探索，以及法医分析和安全。2该项目是协同和培训研究生的领导作用，在信息技术。该项目还惠及其他领域，包括但不限于内容检索、聚类、数据库索引、模式识别、生物识别和人机交互。