Perceptual hashing is widely used to search or match similar images for digital forensics and cybercrime study. Unfortunately, the robustness of perceptual hashing algorithms is not well understood in these contexts. In this paper, we examine the robustness of perceptual hashing and its dependent security applications both experimentally and empirically. We first develop a series of attack algorithms to subvert perceptual hashing based image search. This is done by generating attack images that effectively enlarge the hash distance to the original image while introducing minimal visual changes. To make the attack practical, we design the attack algorithms under a black-box setting, augmented with novel designs (e.g., grayscale initialization) to improve the attack efficiency and transferability. We then evaluate our attack against the standard pHash as well as its robust variant using three different datasets. After confirming the attack effectiveness experimentally, we then empirically test against real-world reverse image search engines including TinEye, Google, Microsoft Bing, and Yandex. We find that our attack is highly successful on TinEye and Bing, and is moderately successful on Google and Yandex. Based on our findings, we discuss possible countermeasures and recommendations.

Business processes underpin a large number of enterprise operations including processing loan applications, managing invoices, and insurance claims. The business process management (BPM) industry is expected to grow at approximately 16 Billion dollar by 2023. There is a large opportunity for infusing AI to reduce cost or provide better customer experience with a $15.7 trillion “potential contribution to the global economy by 2030”. To this end, the BPM literature is rich in machine learning solutions including unsupervised learning to gain insights on clusters of process traces, classification models to predict the outcomes, duration, or paths of partial process traces, extracting business process from documents, and models to recommend how to optimize a business process or navigate decision points. More recently, deep learning models including those from the NLP domain have been applied to process predictions.Unfortunately, very little of these innovations have been applied and adopted by enterprise companies. We assert that a large reason for the lack of adoption of AI models in BPM is that business users are risk-averse and do not implicitly trust AI models. There has, unfortunately, been little attention paid to explaining model predictions to business users with process context. We challenge the BPM community to build on the AI interpretability literature, and the AI Trust community to understand what it means to take advantage of business process artifacts in order to provide business level explanations.

Machine learning has been widely applied to building security applications. However, many machine learning models require the continuous supply of representative labeled data for training, which limits the models' usefulness in practice. In this paper, we use bot detection as an example to explore the use of data synthesis to address this problem. We collected the network traffic from 3 online services in three different months within a year (23 million network requests). We develop a stream-based feature encoding scheme to support machine learning models for detecting advanced bots. The key novelty is that our model detects bots with extremely limited labeled data. We propose a data synthesis method to synthesize unseen (or future) bot behavior distributions. The synthesis method is distribution-aware, using two different generators in a Generative Adversarial Network to synthesize data for the clustered regions and the outlier regions in the feature space. We evaluate this idea and show our method can train a model that outperforms existing methods with only 1% of the labeled data. We show that data synthesis also improves the model's sustainability over time and speeds up the retraining. Finally, we compare data synthesis and adversarial retraining and show they can work complementary with each other to improve the model generalizability.

While deep learning models have achieved unprecedented success in various domains, there is also a growing concern of adversarial attacks against related applications. Recent results show that by adding a small amount of perturbations to an image (imperceptible to humans), the resulting adversarial examples can force a classifier to make targeted mistakes. So far, most existing works focus on crafting adversarial examples in the digital domain, while limited efforts have been devoted to understanding the physical domain attacks. In this work, we explore the feasibility of generating robust adversarial examples that remain effective in the physical domain. Our core idea is to use an image-to-image translation network to simulate the digital-to-physical transformation process for generating robust adversarial examples. To validate our method, we conduct a large-scale physical-domain experiment, which involves manually taking more than 3000 physical domain photos. The results show that our method outperforms existing ones by a large margin and demonstrates a high level of robustness and transferability.

Leaked passwords from data breaches can pose a serious threat if users reuse or slightly modify the passwords for other services. With more services getting breached today, there is still a lack of a quantitative understanding of this risk. In this paper, we perform the first large-scale empirical analysis of password reuse and modification patterns using a ground-truth dataset of 28.8 million users and their 61.5 million passwords in 107 services over 8 years. We find that password reuse and modification is very common (observed on 52% of the users). Sensitive online services such as shopping websites and email services received the most reused and modified passwords. We also observe that users would still reuse the already-leaked passwords for other online services for years after the initial data breach. Finally, to quantify the security risks, we develop a new training-based guessing algorithm. We show that more than 16 million password pairs (including 30% of the modified passwords) can be cracked within just 10 guesses.

Online question and answer (Q&A) services are facing key challenges to motivate domain experts to provide quick and high-quality answers. Recent systems seek to engage real-world experts by allowing them to set a price on their answers. This leads to a "targeted" Q&A model where users to ask questions to a target expert by paying the price. In this paper, we perform a case study on two emerging targeted Q&A systems Fenda (China) and Whale (US) to understand how monetary incentives affect user behavior. By analyzing a large dataset of 220K questions (worth 1 million USD), we find that payments indeed enable quick answers from experts, but also drive certain users to game the system for profits. In addition, this model requires users (experts) to proactively adjust their price to make profits. People who are unwilling to lower their prices are likely to hurt their income and engagement over time.

Social media is often viewed as a sensor into various societal events such as disease outbreaks, protests, and elections. We describe the use of social media as a crowdsourced sensor to gain insight into ongoing cyber-attacks. Our approach detects a broad range of cyber-attacks (e.g., distributed denial of service (DDoS) attacks, data breaches, and account hijacking) in a weakly supervised manner using just a small set of seed event triggers and requires no training or labeled samples. A new query expansion strategy based on convolution kernels and dependency parses helps model semantic structure and aids in identifying key event characteristics. Through a large-scale analysis over Twitter, we demonstrate that our approach consistently identifies and encodes events, outperforming existing methods.

Given a social network, how to find communities of nodes based on their diffusive characteristics? There exist two important types of nodes, for information propagation: nodes that are influential (“kernel nodes”), and nodes that serve as “bridges” to boost the diffusion (“media nodes”). How to find these nodes and uncover connections between them? In addition, it is also important to discover the hidden community structure of these nodes, which can help study their interactions, predict links and also understand the information flow in such networks. In this paper, we give an intuitive and novel optimization-based formulation for this task, which aims to discover media nodes as well as community structures of kernel nodes. We prove our task is computationally challenging, and develop an effective and practical algorithm MeiKe (pronounced as ‘Mike’). It first obtains media nodes via a new successive summarization based approach, and then finds kernel nodes including their community structures. Experimental results show that MeiKe finds high-quality media and kernel communities which match our expectations and ground-truth (outperforming non-trivial baselines by 40% in F1-score). Our case studies also demonstrate the applicability of MeiKe on a variety of datasets.

Ensembles of graphs arise in several natural applications. Many techniques exist to compute frequent, dense subgraphs in these ensembles. In contrast, in this paper, we propose to discover maximally variable regions of the graphs, i.e., sets of nodes that induce very different subgraphs across the ensemble. We first develop two intuitive and novel definitions of such node sets, which we then show can be efficiently enumerated using a level-wise algorithm. Finally, using extensive experiments on multiple real datasets, we show how these sets capture the main structural variations of the given set of networks and also provide us with interesting and relevant insights about these datasets.