Motivated by applications in automated budget optimization, we consider the Adwords problem of Mehta et al. (2005) with unknown advertiser budgets. In this setting, the budget of an advertiser is revealed to the algorithm only when it is exceeded. An algorithm that is oblivious to budgets gives an Ad platform the flexibility to adjust budgets in real-time which, we argue, has tangible benefits. Prominent online algorithms for the Adwords problem critically rely on knowledge of budgets. We give the first budget oblivious algorithm for Adwords with competitive ratio guarantee of at least $0.522$ (better than greedy) against an offline algorithm that knows bids and budgets.
翻译:在自动预算优化应用程序的推动下,我们考虑了Mehta等人(2005年)的口号问题,其预算不明,在这一背景下,广告商的预算只有在超出算法时才披露给算法;一个忽视预算的算法赋予了Ad平台实时调整预算的灵活性,我们认为,这种灵活性具有实际好处。Adword问题的显著在线算法严重依赖对预算的了解。我们给具有至少0.522美元(比贪婪好)的竞争性比率保证的词提供了第一个不为人知的预算算法,以对抗一个熟悉投标和预算的离线算法。