Bid Shading has become increasingly important in Online Advertising, with a large amount of commercial [4,12,13,29] and research work [11,20,28] recently published. Most approaches for solving the bid shading problem involve estimating the probability of win distribution, and then maximizing surplus [28]. These generally use parametric assumptions for the distribution, and there has been some discussion as to whether Log-Normal, Gamma, Beta, or other distributions are most effective [8,38,41,44]. In this paper, we show evidence that online auctions generally diverge in interesting ways from classic distributions. In particular, real auctions generally exhibit significant structure, due to the way that humans set up campaigns and inventory floor prices [16,26]. Using these insights, we present a nonparametric method for Bid Shading which enables the exploitation of this deep structure. The algorithm has low time and space complexity, and is designed to operate within the challenging millisecond Service Level Agreements of Real-Time Bid Servers. We deploy it in one of the largest Demand Side Platforms in the United States, and show that it reliably out-performs best in class Parametric benchmarks. We conclude by suggesting some ways that the best aspects of parametric and nonparametric approaches could be combined.

Since 2019, most ad exchanges and sell-side platforms (SSPs), in the online advertising industry, shifted from second to first price auctions. Due to the fundamental difference between these auctions, demand-side platforms (DSPs) have had to update their bidding strategies to avoid bidding unnecessarily high and hence overpaying. Bid shading was proposed to adjust the bid price intended for second-price auctions, in order to balance cost and winning probability in a first-price auction setup. In this study, we introduce a novel deep distribution network for optimal bidding in both open (non-censored) and closed (censored) online first-price auctions. Offline and online A/B testing results show that our algorithm outperforms previous state-of-art algorithms in terms of both surplus and effective cost per action (eCPX) metrics. Furthermore, the algorithm is optimized in run-time and has been deployed into VerizonMedia DSP as production algorithm, serving hundreds of billions of bid requests per day. Online A/B test shows that advertiser's ROI are improved by +2.4%, +2.4%, and +8.6% for impression based (CPM), click based (CPC), and conversion based (CPA) campaigns respectively.

Online auctions play a central role in online advertising, and are one of the main reasons for the industry's scalability and growth. With great changes in how auctions are being organized, such as changing the second- to first-price auction type, advertisers and demand platforms are compelled to adapt to a new volatile environment. Bid shading is a known technique for preventing overpaying in auction systems that can help maintain the strategy equilibrium in first-price auctions, tackling one of its greatest drawbacks. In this study, we propose a machine learning approach of modeling optimal bid shading for non-censored online first-price ad auctions. We clearly motivate the approach and extensively evaluate it in both offline and online settings on a major demand side platform. The results demonstrate the superiority and robustness of the new approach as compared to the existing approaches across a range of performance metrics.

Challenge Response is one of the cornerstones of online security. The simplest form of Challenge-Response is asking for a password. Much cryptographic work has focused on developing strong forms of encryption, however some networks require transmission over networks which might be monitored. We discuss this problem in the context of a particular kind of open network used by 30,000 users, and which is an important medium supporting emergency services. The current challenge-response implementation on this network relies upon sending information about the password. We calculate the number of observations needed to capture password using brute force attack, replay attack, and version spaces. We show that even strong passwords (completely random set of characters) are at significant risk of discovery in as few as 16 login attempts. We next present an algorithm that adds adversarial “chaff” to the password information designed to minimize relative information gain during challenge-response. We show that, with enough adversarial chaff, unambiguous password recovery from passive data capture may not be possible, although passwords can still be recovered by an attacker actively probing the system. Despite this, better protection of passwords is useful, and would be immediately helpful to people using these services.

Online advertising has exceeded television, to became the largest revenue advertising category in the United States [146], [147]. Some of the reasons for this success are online advertising's unique capabilities for measurement and personalized ad delivery. However, this success also presents new challenges for society. The same technology used for selling products can be employed as a tool for mass persuasion. This paper examines three aspects of this problem: data privacy, information bubbles, and false information. The paper offers some thoughts on addressing each problem.

Ad-servers have to satisfy many different targeting criteria, and the combination can often result in no feasible solution. We hypothesize that advertisers may be defining these metrics to create a kind of "proxy target". We therefore reformulate the standard ad-serving problem to one where we attempt to get as close as possible to the advertiser's multi-dimensional target inclusive of delivery. We use a simple simulation to illustrate the behavior of this algorithm compared to Constraint and Pacing strategies. The system is then deployed in one of the largest video ad-servers in the United States and we show experimental results from live test ads, as well as 6 months of production performance across hundreds of ads. We find that the live ad-server tests match the simulation, and we report significant gains in multi-KPI performance from using the error minimization strategy.

We present a method for targeting ads on television that works on today's TV systems. The method works by mining vast amounts of Set Top Box data, as well as advertiser customer data. From both sources the system builds demographic profiles, and then looks for media that have the highest match per dollar to the customer profile. The method was tested in four live television campaigns, comprising over 22,000 airings, and we present experimental results.

Ad Servers monetize by sending ads to a requesting publisher which renders them on their web page. The Publisher receives a commission if the ad is clicked. Unfortunately, fraudulent publishers may try to request ads and click on them through a variety of fraudulent schemes including robotic traffic, deceptive placement, and distribution across other sites. Armored ads are designed to thwart these attempts and are also designed to serve as a probe into whether traffic is human or robotic.

Click Fraud Bots pose a significant threat to the online economy. To-date efforts to filter bots have been geared towards identifiable useragent strings, as epitomized by the IAB's Robots and Spiders list. However bots designed to perpetrate malicious activity or fraud, are designed to avoid detection with these kinds of lists, and many use very sophisticated schemes for cloaking their activities. In order to combat this emerging threat, we propose the creation of Bot Signatures for training and evaluation of candidate Click Fraud Detection Systems. Bot signatures comprise keyed records connected to case examples. We demonstrate the technique by developing 8 simulated examples of Bots described in the literature including Click Bot A.

Click Fraud remains one of the most durable fraudulent schemes online. With 50 billion dollars being generated per year by Google alone, a fraudulent publisher is able to capture a significant amount of revenue with a small investment. The most well heeled click fraud attacks employ large distributed botnets, deceptive publisher pages, malware infection, and fake conversion “chaff” in an attempt to cloak fraudulent activity. We describe an algorithm that we call Mix Adjustment which corrects for traffic bias differences. The method is scalable and we show a simple implementation that can be applied to current weblog processing systems. We show two case studies of this algorithm on real fraud detection problems: (a) WOW Bot net detection, (b) Advertiser fraud detection.

This paper presents a practical method for measuring the impact of multiple marketing events on sales, including marketing events that are not traditionally trackable. The technique infers which of several competing media events are likely to have caused a given conversion. We test the method using hold-out sets, and also a live media experiment in which we test whether the method can accurately predict television-generated web conversions.

Online Ad Servers attempt to find best ads to serve for a given triggering user event. The performance of ads may be measured in several ways. We suggest a formulation in which the ad network tries to maximize revenue subject to relevance constraints. We describe several algorithms for ad selection and review their complexity. We tested these algorithms using Microsoft ad network from October 1 2006 to February 8 2007. Over 3 billion impressions, 8 million combinations of triggers with ads, and a number of algorithms were tested over this period. We discover curious differences between ad-servers aimed at revenue versus clickthrough rate.

We develop a method for recommending products to customers with applications to both on-line and surface mail promotional offers. Our method differs from previous work in collaborative filtering [8] and imputation [18], in that we assume probabilities are conditionally independent. This assumption, which is also made in Naive Bayes [5], enables us to pre-compute probabilities and store them in main memory, enabling very fast performance on millions of customers. The algorithm supports a variety of tunable parameters so that the method can address different promotional objectives. We tested the algorithm at an on-line hardware retailer, with 17,400 customers divided randomly into control and experimental groups. In the experimental group, clickthrough increased by +40% (p<0.01), revenue by +38% (p<0.07), and units sold by +61% (p<0.01). By changing the algorithm’s parameter settings we found that these results could be improved even further. This work demonstrates the considerable potential of automated data mining for dramatically increasing the profitability of on and off-line retail promotions.