Contextual Online Learning for Multimedia Content Aggregation

Abstract

The last decade has witnessed a tremendous growth in the volume as well as the diversity of multimedia content generated by a multitude of sources (news agencies, social media, etc.). Faced with a variety of content choices, consumers are exhibiting diverse preferences for content; their preferences often depend on the context in which they consume content as well as various exogenous events. To satisfy the consumers' demand for such diverse content, multimedia content aggregators (CAs) have emerged which gather content from numerous multimedia sources. A key challenge for such systems is to accurately predict what type of content each of its consumers prefers in a certain context, and adapt these predictions to the evolving consumers' preferences, contexts, and content characteristics. We propose a novel, distributed, online multimedia content aggregation framework, which gathers content generated by multiple heterogeneous producers to fulfill its consumers' demand for content. Since both the multimedia content characteristics and the consumers' preferences and contexts are unknown, the optimal content aggregation strategy is unknown a priori. Our proposed content aggregation algorithm is able to learn online what content to gather and how to match content and users by exploiting similarities between consumer types. We prove bounds for our proposed learning algorithms that guarantee both the accuracy of the predictions as well as the learning speed. Importantly, our algorithms operate efficiently even when feedback from consumers is missing or content and preferences evolve over time. Illustrative results highlight the merits of the proposed content aggregation system in a variety of settings.Index Terms-Content aggregation, distributed online learning, multi-armed bandits, social multimedia.

Actionable intelligence and online learning for semantic computing

“…Some MAB methods learn independently for each context, 16 while some MAB methods exploit the similarity between classifier accuracies. 22 While in the above formulation the stream distribution is static, numerous works considered time-varying distributions for classification rules, 11 which is also called concept drift. 24,25 A change in accuracy of a classifier can happen when the classifier updates its prediction rule.…”

“…Knowing this, the learner can analytically determine the window of history of past predictions, labels and decisions it should take into account when estimating the current classifier accuracies. 11 Intuitively, as increases, the learner can rely on a larger window of history to estimate the classifier accuracies. MAB methods can also be used for the case when the accuracy drift is abrupt but infrequent, i.e., there can be a large change in a classifier's accuracy between two consecutive time slots, but the frequency of such changes is low so that the changes can be tracked.…”

“…Nonadaptive contexts method has been successfully applied in numerous semantic computing applications including context-driven image stream mining 2 and item recommendation in social networks. 9 Similarly, adaptive contexts methods has been successfully applied in context-driven medical expertise discovery, 4 multimedia content aggregation 11 and recommender systems. 8 In the presence of accuracy drift, the learner needs to choose classifiers to make predictions solely on the recent relevant history.…”

“…These streams are then mined in realtime to assist numerous semantic computing applications (see Fig. 1): patient monitoring, 1 personalized diagnosis, 2,3 personalized treatment recommendation, 4,5 recommender systems, [6][7][8] social networks, 9 network security, 10 multimedia content aggregation, 11 personalized education 12,13 etc. Hence, online data mining systems have emerged that enable such applications to analyze, extract actionable intelligence and make decisions in real-time, based on the correlated, highdimensional and dynamic data captured by multiple heterogeneous data sources.…”

Multi-Perspective Cost-Sensitive Context-Aware Multi-Instance Sparse Coding and Its Application to Sensitive Video Recognition

“…Recognition of sensitive videos is a newly emergent research topic in the multimedia and pattern recognition communities, in the context of multimedia retrieval [55,56], multimedia content understanding [59,60], and multimodal fusion [56,57], etc. In recent years a number of specific attempts have been made to deal with the problem of sensitive video recognition, and most of them focus on adult video recognition [11,12,18,21].…”