arXiv subjects

Wenke Lee

Publications and source records attributed to Wenke Lee.

At least 37 records · Page 2Linked to original sources

arXiv2025

Bridging Unsupervised and Semi-Supervised Anomaly Detection: A Theoretically-Grounded and Practical Framework with Synthetic Anomalies

Anomaly detection (AD) is a critical task across domains such as cybersecurity and healthcare. In the unsupervised setting, an effective and theoretically-grounded principle is to train classifiers to distinguish normal data from (synthetic) anomalies. We extend this principle to semi-supervised AD, where training data also include a limited labeled subset of anomalies possibly present in test time. We propose a theoretically-grounded and empirically effective framework for semi-supervised AD that combines known and synthetic anomalies during training. To analyze semi-supervised AD, we introduce the first mathematical formulation of semi-supervised AD, which generalizes unsupervised AD. Here, we show that synthetic anomalies enable (i) better anomaly modeling in low-density regions and (ii) optimal convergence guarantees for neural network classifiers -- the first theoretical result for semi-supervised AD. We empirically validate our framework on five diverse benchmarks, observing consistent performance gains. These improvements also extend beyond our theoretical framework to other classification-based AD methods, validating the generalizability of the synthetic anomaly principle in AD.

stat.ML↗

arXiv2025

RenderBender: A Survey on Adversarial Attacks Using Differentiable Rendering

Differentiable rendering techniques like Gaussian Splatting and Neural Radiance Fields have become powerful tools for generating high-fidelity models of 3D objects and scenes. Their ability to produce both physically plausible and differentiable models of scenes are key ingredient needed to produce physically plausible adversarial attacks on DNNs. However, the adversarial machine learning community has yet to fully explore these capabilities, partly due to differing attack goals (e.g., misclassification, misdetection) and a wide range of possible scene manipulations used to achieve them (e.g., alter texture, mesh). This survey contributes the first framework that unifies diverse goals and tasks, facilitating easy comparison of existing work, identifying research gaps, and highlighting future directions - ranging from expanding attack goals and tasks to account for new modalities, state-of-the-art models, tools, and pipelines, to underscoring the importance of studying real-world threats in complex scenes.

cs.LG↗

arXiv2025

3D Gaussian Splat Vulnerabilities

With 3D Gaussian Splatting (3DGS) being increasingly used in safety-critical applications, how can an adversary manipulate the scene to cause harm? We introduce CLOAK, the first attack that leverages view-dependent Gaussian appearances - colors and textures that change with viewing angle - to embed adversarial content visible only from specific viewpoints. We further demonstrate DAGGER, a targeted adversarial attack directly perturbing 3D Gaussians without access to underlying training data, deceiving multi-stage object detectors e.g., Faster R-CNN, through established methods such as projected gradient descent. These attacks highlight underexplored vulnerabilities in 3DGS, introducing a new potential threat to robotic learning for autonomous navigation and other safety-critical 3DGS applications.

cs.CR↗

arXiv2025

Can Transformers Reason Logically? A Study in SAT Solving

We formally study the logical reasoning capabilities of decoder-only Transformers in the context of the boolean satisfiability (SAT) problem. First, we prove by construction that decoder-only Transformers can decide 3-SAT, in a non-uniform model of computation, using backtracking and deduction via Chain-of-Thought (CoT). %We prove its correctness by showing trace equivalence to the well-known DPLL SAT-solving algorithm. Second, we implement our construction as a PyTorch model with a tool (PARAT) that we designed to empirically demonstrate its correctness and investigate its properties. Third, rather than \textit{programming} a transformer to reason, we evaluate empirically whether it can be \textit{trained} to do so by learning directly from algorithmic traces (``reasoning paths'') from our theoretical construction. The trained models demonstrate strong out-of-distribution generalization on problem sizes seen during training but has limited length generalization, which is consistent with the implications of our theoretical result

cs.LG↗

arXiv2024

BACKRUNNER: Mitigating Smart Contract Attacks in the Real World

Billions of dollars have been lost due to vulnerabilities in smart contracts. To counteract this, researchers have proposed attack frontrunning protections designed to preempt malicious transactions by inserting "whitehat" transactions ahead of them to protect the assets. In this paper, we demonstrate that existing frontrunning protections have become ineffective in real-world scenarios. Specifically, we collected 158 recent real-world attack transactions and discovered that 141 of them can bypass state-of-the-art frontrunning protections. We systematically analyze these attacks and show how inherent limitations of existing frontrunning techniques hinder them from protecting valuable assets in the real world. We then propose a new approach involving 1) preemptive hijack, and 2) attack backrunning, which circumvent the existing limitations and can help protect assets before and after an attack. Our approach adapts the exploit used in the attack to the same or similar contracts before and after the attack to safeguard the assets. We conceptualize adapting exploits as a program repair problem and apply established techniques to implement our approach into a full-fledged framework, BACKRUNNER. Running on previous attacks in 2023, BACKRUNNER can successfully rescue more than \$410M. In the real world, it has helped rescue over \$11.2M worth of assets in 28 separate incidents within two months.

cs.CR↗

arXiv2024

Non-Robust Features are Not Always Useful in One-Class Classification

The robustness of machine learning models has been questioned by the existence of adversarial examples. We examine the threat of adversarial examples in practical applications that require lightweight models for one-class classification. Building on Ilyas et al. (2019), we investigate the vulnerability of lightweight one-class classifiers to adversarial attacks and possible reasons for it. Our results show that lightweight one-class classifiers learn features that are not robust (e.g. texture) under stronger attacks. However, unlike in multi-class classification (Ilyas et al., 2019), these non-robust features are not always useful for the one-class task, suggesting that learning these unpredictive and non-robust features is an unwanted consequence of training.

cs.LG↗

arXiv2024

Revisiting Non-separable Binary Classification and its Applications in Anomaly Detection

The inability to linearly classify XOR has motivated much of deep learning. We revisit this age-old problem and show that linear classification of XOR is indeed possible. Instead of separating data between halfspaces, we propose a slightly different paradigm, equality separation, that adapts the SVM objective to distinguish data within or outside the margin. Our classifier can then be integrated into neural network pipelines with a smooth approximation. From its properties, we intuit that equality separation is suitable for anomaly detection. To formalize this notion, we introduce closing numbers, a quantitative measure on the capacity for classifiers to form closed decision regions for anomaly detection. Springboarding from this theoretical connection between binary classification and anomaly detection, we test our hypothesis on supervised anomaly detection experiments, showing that equality separation can detect both seen and unseen anomalies.

cs.LG↗

arXiv2023

Bridging Both Worlds in Semantics and Time: Domain Knowledge Based Analysis and Correlation of Industrial Process Attacks

Modern industrial control systems (ICS) attacks infect supervisory control and data acquisition (SCADA) hosts to stealthily alter industrial processes, causing damage. To detect attacks with low false alarms, recent work detects attacks in both SCADA and process data. Unfortunately, this led to the same problem - disjointed (false) alerts, due to the semantic and time gap in SCADA and process behavior, i.e., SCADA execution does not map to process dynamics nor evolve at similar time scales. We propose BRIDGE to analyze and correlate SCADA and industrial process attacks using domain knowledge to bridge their unique semantic and time evolution. This enables operators to tie malicious SCADA operations to their adverse process effects, which reduces false alarms and improves attack understanding. BRIDGE (i) identifies process constraints violations in SCADA by measuring actuation dependencies in SCADA process-control, and (ii) detects malicious SCADA effects in processes via a physics-informed neural network that embeds generic knowledge of inertial process dynamics. BRIDGE then dynamically aligns both analysis (i and ii) in a time-window that adjusts their time evolution based on process inertial delays. We applied BRIDGE to 11 diverse real-world industrial processes, and adaptive attacks inspired by past events. BRIDGE correlated 98.3% of attacks with 0.8% false positives (FP), compared to 78.3% detection accuracy and 13.7% FP of recent work.

cs.CR↗

arXiv2022

SCAPHY: Detecting Modern ICS Attacks by Correlating Behaviors in SCADA and PHYsical

Modern Industrial Control Systems (ICS) attacks evade existing tools by using knowledge of ICS processes to blend their activities with benign Supervisory Control and Data Acquisition (SCADA) operation, causing physical world damages. We present SCAPHY to detect ICS attacks in SCADA by leveraging the unique execution phases of SCADA to identify the limited set of legitimate behaviors to control the physical world in different phases, which differentiates from attackers activities. For example, it is typical for SCADA to setup ICS device objects during initialization, but anomalous during processcontrol. To extract unique behaviors of SCADA execution phases, SCAPHY first leverages open ICS conventions to generate a novel physical process dependency and impact graph (PDIG) to identify disruptive physical states. SCAPHY then uses PDIG to inform a physical process-aware dynamic analysis, whereby code paths of SCADA process-control execution is induced to reveal API call behaviors unique to legitimate process-control phases. Using this established behavior, SCAPHY selectively monitors attackers physical world-targeted activities that violates legitimate processcontrol behaviors. We evaluated SCAPHY at a U.S. national lab ICS testbed environment. Using diverse ICS deployment scenarios and attacks across 4 ICS industries, SCAPHY achieved 95% accuracy & 3.5% false positives (FP), compared to 47.5% accuracy and 25% FP of existing work. We analyze SCAPHYs resilience to futuristic attacks where attacker knows our approach.

cs.CR↗

arXiv2021

The Threat of Offensive AI to Organizations

AI has provided us with the ability to automate tasks, extract information from vast amounts of data, and synthesize media that is nearly indistinguishable from the real thing. However, positive tools can also be used for negative purposes. In particular, cyber adversaries can use AI (such as machine learning) to enhance their attacks and expand their campaigns. Although offensive AI has been discussed in the past, there is a need to analyze and understand the threat in the context of organizations. For example, how does an AI-capable adversary impact the cyber kill chain? Does AI benefit the attacker more than the defender? What are the most significant AI threats facing organizations today and what will be their impact on the future? In this survey, we explore the threat of offensive AI on organizations. First, we present the background and discuss how AI changes the adversary's methods, strategies, goals, and overall attack model. Then, through a literature review, we identify 33 offensive AI capabilities which adversaries can use to enhance their attacks. Finally, through a user study spanning industry and academia, we rank the AI threats and provide insights on the adversaries.

cs.AI↗

arXiv2021

A Reinforcement Learning Approach for Dynamic Information Flow Tracking Games for Detecting Advanced Persistent Threats

Advanced Persistent Threats (APTs) are stealthy attacks that threaten the security and privacy of sensitive information. Interactions of APTs with victim system introduce information flows that are recorded in the system logs. Dynamic Information Flow Tracking (DIFT) is a promising detection mechanism for detecting APTs. DIFT taints information flows originating at system entities that are susceptible to an attack, tracks the propagation of the tainted flows, and authenticates the tainted flows at certain system components according to a pre-defined security policy. Deployment of DIFT to defend against APTs in cyber systems is limited by the heavy resource and performance overhead associated with DIFT. In this paper, we propose a resource-efficient model for DIFT by incorporating the security costs, false-positives, and false-negatives associated with DIFT. Specifically, we develop a game-theoretic framework and provide an analytical model of DIFT that enables the study of trade-off between resource efficiency and the effectiveness of detection. Our game model is a nonzero-sum, infinite-horizon, average reward stochastic game. Our model incorporates the information asymmetry between players that arises from DIFT's inability to distinguish malicious flows from benign flows and APT's inability to know the locations where DIFT performs a security analysis. Additionally, the game has incomplete information as the transition probabilities (false-positive and false-negative rates) are unknown. We propose a multiple-time scale stochastic approximation algorithm to learn an equilibrium solution of the game. We prove that our algorithm converges to an average reward Nash equilibrium. We evaluate our proposed model and algorithm on a real-world ransomware dataset and validate the effectiveness of the proposed approach.

math.OC↗

arXiv2021

Stochastic Dynamic Information Flow Tracking Game using Supervised Learning for Detecting Advanced Persistent Threats

Advanced persistent threats (APTs) are organized prolonged cyberattacks by sophisticated attackers. Although APT activities are stealthy, they interact with the system components and these interactions lead to information flows. Dynamic Information Flow Tracking (DIFT) has been proposed as one of the effective ways to detect APTs using the information flows. However, wide range security analysis using DIFT results in a significant increase in performance overhead and high rates of false-positives and false-negatives generated by DIFT. In this paper, we model the strategic interaction between APT and DIFT as a non-cooperative stochastic game. The game unfolds on a state space constructed from an information flow graph (IFG) that is extracted from the system log. The objective of the APT in the game is to choose transitions in the IFG to find an optimal path in the IFG from an entry point of the attack to an attack target. On the other hand, the objective of DIFT is to dynamically select nodes in the IFG to perform security analysis for detecting APT. Our game model has imperfect information as the players do not have information about the actions of the opponent. We consider two scenarios of the game (i) when the false-positive and false-negative rates are known to both players and (ii) when the false-positive and false-negative rates are unknown to both players. Case (i) translates to a game model with complete information and we propose a value iteration-based algorithm and prove the convergence. Case (ii) translates to a game with unknown transition probabilities. In this case, we propose Hierarchical Supervised Learning (HSL) algorithm that integrates a neural network, to predict the value vector of the game, with a policy iteration algorithm to compute an approximate equilibrium. We implemented our algorithms on real attack datasets and validated the performance of our approach.

cs.GT↗

arXiv2021

Dynamic Information Flow Tracking for Detection of Advanced Persistent Threats: A Stochastic Game Approach

Advanced Persistent Threats (APTs) are stealthy customized attacks by intelligent adversaries. This paper deals with the detection of APTs that infiltrate cyber systems and compromise specifically targeted data and/or infrastructures. Dynamic information flow tracking is an information trace-based detection mechanism against APTs that taints suspicious information flows in the system and generates security analysis for unauthorized use of tainted data. In this paper, we develop an analytical model for resource-efficient detection of APTs using an information flow tracking game. The game is a nonzero-sum, turn-based, stochastic game with asymmetric information as the defender cannot distinguish whether an incoming flow is malicious or benign and hence has only partial state observation. We analyze equilibrium of the game and prove that a Nash equilibrium is given by a solution to the minimum capacity cut set problem on a flow-network derived from the system, where the edge capacities are obtained from the cost of performing security analysis. Finally, we implement our algorithm on the real-world dataset for a data exfiltration attack augmented with false-negative and false-positive rates and compute an optimal defender strategy.

cs.GT↗

arXiv2021

SEPAL: Towards a Large-scale Analysis of SEAndroid Policy Customization

To investigate the status quo of SEAndroid policy customization, we propose SEPAL, a universal tool to automatically retrieve and examine the customized policy rules. SEPAL applies the NLP technique and employs and trains a wide&deep model to quickly and precisely predict whether one rule is unregulated or not.Our evaluation shows SEPAL is effective, practical and scalable. We verify SEPAL outperforms the state of the art approach (i.e., EASEAndroid) by 15% accuracy rate on average. In our experiments, SEPAL successfully identifies 7,111 unregulated policy rules with a low false positive rate from 595,236 customized rules (extracted from 774 Android firmware images of 72 manufacturers). We further discover the policy customization problem is getting worse in newer Android versions (e.g., around 8% for Android 7 and nearly 20% for Android 9), even though more and more efforts are made. Then, we conduct a deep study and discuss why the unregulated rules are introduced and how they can compromise user devices. Last, we report some unregulated rules to seven vendors and so far four of them confirm our findings.

cs.CR↗

arXiv2020

Towards Measuring Supply Chain Attacks on Package Managers for Interpreted Languages

Package managers have become a vital part of the modern software development process. They allow developers to reuse third-party code, share their own code, minimize their codebase, and simplify the build process. However, recent reports showed that package managers have been abused by attackers to distribute malware, posing significant security risks to developers and end-users. For example, eslint-scope, a package with millions of weekly downloads in Npm, was compromised to steal credentials from developers. To understand the security gaps and the misplaced trust that make recent supply chain attacks possible, we propose a comparative framework to qualitatively assess the functional and security features of package managers for interpreted languages. Based on qualitative assessment, we apply well-known program analysis techniques such as metadata, static, and dynamic analysis to study registry abuse. Our initial efforts found 339 new malicious packages that we reported to the registries for removal. The package manager maintainers confirmed 278 (82%) from the 339 reported packages where three of them had more than 100,000 downloads. For these packages we were issued official CVE numbers to help expedite the removal of these packages from infected victims. We outline the challenges of tailoring program analysis tools to interpreted languages and release our pipeline as a reference point for the community to build on and help in securing the software supply chain.

cs.CR↗

arXiv2020

The Creation and Detection of Deepfakes: A Survey

Generative deep learning algorithms have progressed to a point where it is difficult to tell the difference between what is real and what is fake. In 2018, it was discovered how easy it is to use this technology for unethical and malicious applications, such as the spread of misinformation, impersonation of political leaders, and the defamation of innocent individuals. Since then, these `deepfakes' have advanced significantly. In this paper, we explore the creation and detection of deepfakes and provide an in-depth view of how these architectures work. The purpose of this survey is to provide the reader with a deeper understanding of (1) how deepfakes are created and detected, (2) the current trends and advancements in this domain, (3) the shortcomings of the current defense solutions, and (4) the areas which require further research and attention.

cs.CV↗

arXiv2020

SQUIRREL: Testing Database Management Systems with Language Validity and Coverage Feedback

Fuzzing is an increasingly popular technique for verifying software functionalities and finding security vulnerabilities. However, current mutation-based fuzzers cannot effectively test database management systems (DBMSs), which strictly check inputs for valid syntax and semantics. Generation-based testing can guarantee the syntax correctness of the inputs, but it does not utilize any feedback, like code coverage, to guide the path exploration. In this paper, we develop Squirrel, a novel fuzzing framework that considers both language validity and coverage feedback to test DBMSs. We design an intermediate representation (IR) to maintain SQL queries in a structural and informative manner. To generate syntactically correct queries, we perform type-based mutations on IR, including statement insertion, deletion and replacement. To mitigate semantic errors, we analyze each IR to identify the logical dependencies between arguments, and generate queries that satisfy these dependencies. We evaluated Squirrel on four popular DBMSs: SQLite, MySQL, PostgreSQL and MariaDB. Squirrel found 51 bugs in SQLite, 7 in MySQL and 5 in MariaDB. 52 of the bugs are fixed with 12 CVEs assigned. In our experiment, Squirrel achieves 2.4x-243.9x higher semantic correctness than state-of-the-art fuzzers, and explores 2.0x-10.9x more new edges than mutation-based tools. These results show that Squirrel is effective in finding memory errors of database management systems.

cs.CR↗

arXiv2018

A Game Theoretic Approach for Dynamic Information Flow Tracking to Detect Multi-Stage Advanced Persistent Threats

Advanced Persistent Threats (APTs) infiltrate cyber systems and compromise specifically targeted data and/or resources through a sequence of stealthy attacks consisting of multiple stages. Dynamic information flow tracking has been proposed to detect APTs. In this paper, we develop a dynamic information flow tracking game for resource-efficient detection of APTs via multi-stage dynamic games. The game evolves on an information flow graph, whose nodes are processes and objects (e.g. file, network endpoints) in the system and the edges capture the interaction between different processes and objects. Each stage of the game has pre-specified targets which are characterized by a set of nodes of the graph and the goal of the APT is to evade detection and reach a target node of that stage. The goal of the defender is to maximize the detection probability while minimizing performance overhead on the system. The resource costs of the players are different and the information structure is asymmetric resulting in a nonzero-sum imperfect information game. We first calculate the best responses of the players and characterize the set of Nash equilibria for single stage attacks. Subsequently, we provide a polynomial-time algorithm to compute a correlated equilibrium for the multi-stage attack case. Finally, we experiment our model and algorithms on real-world nation state attack data obtained from Refinable Attack Investigation system.

cs.GT↗