Abstract We introduce a novel algorithmic approach to content recommendation based on adaptive clustering of exploration-exploitation("bandit") strategies.We provide sharo regret analysis of this algorithm in a standard stochastic noise setting,demo
Algorithms for hyper-parameter optimization.pdf，讲述贝叶斯算法的TPE过程的专业论文The contribution of this work is two novel strategies for approximating f by modeling H: a hier
archical Gaussian Process and a tree-structured parzen estimator. These are described in
感知质量优化和用户招募是移动群智感知的两个重要问题，随着数据量的大幅度增加，感知内容出现冗余，存在感知质量降低的风险。提出了一种感知质量优化的任务分发机制，在保证覆盖率的情况下，提高群体的感知质量。利用聚类算法评估任务真值，量化用户数据质量；基于汤普森抽样算法和贪婪算法设计并实现了一种用户招募策略，在保证任务空间覆盖率的基础上优化感知质量。针对TSUR（Thompson based user recruit）算法的性能进行仿真分析，并与已有的BBTA（bandit-based task assi
