Fine-tuning has become the standard practice for adapting pre-trained models to downstream tasks. However, the impact on model robustness is not well understood. In this work, we characterize the robustness-accuracy trade-off in fine-tuning. We evaluate the robustness and accuracy of fine-tuned models over 6 benchmark datasets and 7 different fine-tuning strategies. We observe a consistent trade-off between adversarial robustness and accuracy. Peripheral updates such as BitFit are more effective for simple tasks -- over 75% above the average measured by the area under the Pareto frontiers on CIFAR-10 and CIFAR-100. In contrast, fine-tuning information-heavy layers, such as attention layers via Compacter, achieves a better Pareto frontier on more complex tasks -- 57.5% and 34.6% above the average on Caltech-256 and CUB-200, respectively. Lastly, we observe that the robustness of fine-tuning against out-of-distribution data closely tracks accuracy. These insights emphasize the need for robustness-aware fine-tuning to ensure reliable real-world deployments.

Storage integrity is essential to systems and applications that use untrusted storage (e.g., public clouds, end-user devices). However, known methods for achieving storage integrity either suffer from high (and often prohibitive) overheads or provide weak integrity guarantees. In this work, we demonstrate a hybrid approach to storage integrity that simultaneously reduces overhead while providing strong integrity guarantees. Our system, partially asynchronous integrity checking (PAC), allows disk write commitments to be deferred while still providing guarantees around read integrity. PAC delivers a 5.5 × throughput and latency improvement over the state of the art, and 85% of the throughput achieved by non-integrity-assuring approaches. In this way, we show that untrusted storage can be used for integrity-critical workloads without meaningfully sacrificing performance.

Cloud file systems offer organizations a scalable and reliable file storage solution. However, cloud file systems have become prime targets for adversaries, and traditional designs are not equipped to protect organizations against the myriad of attacks that may be initiated by a malicious cloud provider, co-tenant, or end-client. Recently proposed designs leveraging cryptographic techniques and trusted execution environments (TEEs) still force organizations to make undesirable trade-offs, consequently leading to either security, functional, or performance limitations. In this paper, we introduceBFS, a cloud file system that leverages the security capabilities provided by TEEs to bootstrap new security protocols that deliver strong security guarantees, high-performance, and a transparent POSIX-like interface to clients.BFSdelivers stronger security guarantees and up to a$2.5\times$speedup over a state-of-the-art secure file system. Moreover, compared to the industry standard NFS,BFSachieves up to$2.2\times$speedups across micro-benchmarks and incurs$< 1\times$overhead for most macro-benchmark workloads.BFSdemonstrates a holistic cloud file system design that does not sacrifice an organizations’ security yet can embrace all of the functional and performance advantages of outsourcing.

Public clouds provide scalable and cost-efficient computing through resource sharing. However, moving from traditional on-premises service management to clouds introduces new challenges; failure to correctly provision, maintain, or decommission elastic services can lead to functional failure and vulnerability to attack. In this paper, we explore a broad class of attacks on clouds which we refer to as cloud squatting. In a cloud squatting attack, an adversary allocates resources in the cloud (e.g., IP addresses) and thereafter leverages latent configuration to exploit prior tenants. To measure and categorize cloud squatting we deployed a custom Internet telescope within the Amazon Web Services us-east-1 region. Using this apparatus, we deployed over 3 million servers receiving 1.5 million unique IP addresses ($\approx$ 56% of the available pool) over 101 days beginning in March of 2021. We identified 4 classes of cloud services, 7 classes of third-party services, and DNS as sources of exploitable latent configurations. We discovered that exploitable configurations were both common and in many cases extremely dangerous; we received over 5 million cloud messages, many containing sensitive data such as financial transactions, GPS location, and PII. Within the 7 classes of third-party services, we identified dozens of exploitable software systems spanning hundreds of servers (e.g., databases, caches, mobile applications, and web services). Lastly, we identified 5446 exploitable domains panning 231 eTLDs—including 105 in the top 10000 and 23 in the top 1000 popular domains. Through tenant disclosures we have identified several root causes, including (a) a lack of organizational controls, (b) poor service hygiene, and (c) failure to follow best practices. We conclude with a discussion of the space of possible mitigations and describe the mitigations to be deployed by Amazon in response to this study.

Machine learning-based network intrusion detection systems have demonstrated state-of-the-art accuracy in flagging malicious traffic. However, machine learning has been shown to be vulnerable to adversarial examples, particularly in domains such as image recognition. In many threat models, the adversary exploits the unconstrained nature of images–the adversary is free to select some arbitrary amount of pixels to perturb. However, it is not clear how these attacks translate to domains such as network intrusion detection as they contain domain constraints, which limit which and how features can be modified by the adversary. In this paper, we explore whether the constrained nature of networks offers additional robustness against adversarial examples versus the unconstrained nature of images. We do this by creating two algorithms: (1) the Adapative-JSMA, an augmented version of the popular JSMA which obeys domain constraints, and (2) the Histogram Sketch Generation which generates adversarial sketches: targeted universal perturbation vectors that encode feature saliency within the envelope of domain constraints. To assess how these algorithms perform, we evaluate them in a constrained network intrusion detection setting and an unconstrained image recognition setting. The results show that our approaches generate misclassification rates in network intrusion detection applications that were comparable to those of image recognition applications (greater than 95%). Our investigation shows that the constrained attack surface exposed by network intrusion detection systems is still sufficiently large to craft successful adversarial examples – and thus, network constraints do not appear to add robustness against adversarial examples. Indeed, even if a defender constrains an adversary to as little as five random features, generating adversarial examples is still possible.

Abstract Millions of consumers depend on smart camera systems to remotely monitor their homes and businesses. However, the architecture and design of popular commercial systems require users to relinquish control of their data to untrusted third parties, such as service providers (e.g., the cloud). Third parties therefore can (and in some instances have) access the video footage without the users’ knowledge or consent—violating the core tenet of user privacy. In this paper, we present CaCTUs, a privacy-preserving smart Camera system Controlled Totally by Users. CaCTUs returns control to the user; the root of trust begins with the user and is maintained through a series of cryptographic protocols, designed to support popular features, such as sharing, deleting, and viewing videos live. We show that the system can support live streaming with a latency of 2 s at a frame rate of 10 fps and a resolution of 480 p. In so doing, we demonstrate that it is feasible to implement a performant smart-camera system that leverages the convenience of a cloud-based model while retaining the ability to control access to (private) data.

Machine learning is vulnerable to adversarial examples--inputs designed to cause models to perform poorly. However, it is unclear if adversarial examples represent realistic inputs in the modeled domains. Diverse domains such as networks and phishing have domain constraints--complex relationships between features that an adversary must satisfy for an attack to be realized (in addition to any adversary-specific goals). In this paper, we explore how domain constraints limit adversarial capabilities and how adversaries can adapt their strategies to create realistic (constraint-compliant) examples. In this, we develop techniques to learn domain constraints from data, and show how the learned constraints can be integrated into the adversarial crafting process. We evaluate the efficacy of our approach in network intrusion and phishing datasets and find: (1) up to 82% of adversarial examples produced by state-of-the-art crafting algorithms violate domain constraints, (2) domain constraints are robust to adversarial examples; enforcing constraints yields an increase in model accuracy by up to 34%. We observe not only that adversaries must alter inputs to satisfy domain constraints, but that these constraints make the generation of valid adversarial examples far more challenging.

Data sharing among partners---users, companies, organizations---is crucial for the advancement of collaborative machine learning in many domains such as healthcare, finance, and security. Sharing through secure computation and other means allow these partners to perform privacy-preserving computations on their private data in controlled ways. However, in reality, there exist complex relationships among members (partners). Politics, regulations, interest, trust, data demands and needs prevent members from sharing their complete data. Thus, there is a need for a mechanism to meet these conflicting relationships on data sharing. This paper presents, an approach to exchange data among members who have complex relationships. A novel policy language, CPL, that allows members to define the specifications of data exchange requirements is introduced. With CPL, members can easily assert who and what to exchange through their local policies and negotiate a global sharing agreement. The agreement is implemented in a distributed privacy-preserving model that guarantees sharing among members will comply with the policy as negotiated. The use of Curie is validated through an example healthcare application built on recently introduced secure multi-party computation and differential privacy frameworks, and policy and performance trade-offs are explored.

For well over a quarter century, detection systems have been driven by models learned from input features collected from real or simulated environments. An artifact (e.g., network event, potential malware sample, suspicious email) is deemed malicious or non-malicious based on its similarity to the learned model at runtime. However, the training of the models has been historically limited to only those features available at runtime. In this paper, we consider an alternate learning approach that trains models using privileged information--features available at training time but not at runtime--to improve the accuracy and resilience of detection systems. In particular, we adapt and extend recent advances in knowledge transfer, model influence, and distillation to enable the use of forensic or other data unavailable at runtime in a range of security domains. An empirical evaluation shows that privileged information increases precision and recall over a system with no privileged information: we observe up to 7.7% relative decrease in detection error for fast-flux bot detection, 8.6% for malware traffic detection, 7.3% for malware classification, and 16.9% for face recognition. We explore the limitations and applications of different privileged information techniques in detection systems. Such techniques provide a new means for detection systems to learn from data that would otherwise not be available at runtime.