Recent work has demonstrated substantial gains on many NLP tasks and benchmarks by pre-training on a large corpus of text followed by fine-tuning on a specific task. While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples. By contrast, humans can generally perform a new language task from only a few examples or from simple instructions - something which current NLP systems still largely struggle to do. Here we show that scaling up language models greatly improves task-agnostic, few-shot performance, sometimes even reaching competitiveness with prior state-of-the-art fine-tuning approaches. Specifically, we train GPT-3, an autoregressive language model with 175 billion parameters, 10x more than any previous non-sparse language model, and test its performance in the few-shot setting. For all tasks, GPT-3 is applied without any gradient updates or fine-tuning, with tasks and few-shot demonstrations specified purely via text interaction with the model. GPT-3 achieves strong performance on many NLP datasets, including translation, question-answering, and cloze tasks, as well as several tasks that require on-the-fly reasoning or domain adaptation, such as unscrambling words, using a novel word in a sentence, or performing 3-digit arithmetic. At the same time, we also identify some datasets where GPT-3's few-shot learning still struggles, as well as some datasets where GPT-3 faces methodological issues related to training on large web corpora. Finally, we find that GPT-3 can generate samples of news articles which human evaluators have difficulty distinguishing from articles written by humans. We discuss broader societal impacts of this finding and of GPT-3 in general.

/pdf/language-models-are-few-shot-learners-2fg8gvia7m.pdf

Language Models are Few-Shot Learners

We will review some of the major results in random graphs and some of the more challenging open problems. We will cover algorithmic and structural questions. We will touch on newer models, including those related to the WWW.

Random graphs

Recent work has demonstrated that deep neural networks are vulnerable to adversarial examples---inputs that are almost indistinguishable from natural data and yet classified incorrectly by the network. In fact, some of the latest findings suggest that the existence of adversarial attacks may be an inherent weakness of deep learning models. To address this problem, we study the adversarial robustness of neural networks through the lens of robust optimization. This approach provides us with a broad and unifying view on much of the prior work on this topic. Its principled nature also enables us to identify methods for both training and attacking neural networks that are reliable and, in a certain sense, universal. In particular, they specify a concrete security guarantee that would protect against any adversary. These methods let us train networks with significantly improved resistance to a wide range of adversarial attacks. They also suggest the notion of security against a first-order adversary as a natural and broad security guarantee. We believe that robustness against such well-defined classes of adversaries is an important stepping stone towards fully resistant deep learning models. Code and pre-trained models are available at this https URL and this https URL.

Towards Deep Learning Models Resistant to Adversarial Attacks

https://www.science.smith.edu/~jcardell/Courses/EGR328/Readings/WSNSurvey2.pdf

Wireless sensor network survey

The problem of privacy-preserving data analysis has a long history spanning multiple disciplines. As electronic data about individuals becomes increasingly detailed, and as technology enables ever more powerful collection and curation of these data, the need increases for a robust, meaningful, and mathematically rigorous definition of privacy, together with a computationally rich class of algorithms that satisfy this definition. Differential Privacy is such a definition.After motivating and discussing the meaning of differential privacy, the preponderance of this monograph is devoted to fundamental techniques for achieving differential privacy, and application of these techniques in creative combinations, using the query-release problem as an ongoing example. A key point is that, by rethinking the computational goal, one can often obtain far better results than would be achieved by methodically replacing each step of a non-private computation with a differentially private implementation. Despite some astonishingly powerful computational results, there are still fundamental limitations — not just on what can be achieved with differential privacy but on what can be achieved with any method that protects against a complete breakdown in privacy. Virtually all the algorithms discussed herein maintain differential privacy against adversaries of arbitrary computational power. Certain algorithms are computationally intensive, others are efficient. Computational complexity for the adversary and the algorithm are both discussed.We then turn from fundamentals to applications other than queryrelease, discussing differentially private methods for mechanism design and machine learning. The vast majority of the literature on differentially private algorithms considers a single, static, database that is subject to many analyses. Differential privacy in other models, including distributed databases and computations on data streams is discussed.Finally, we note that this work is meant as a thorough introduction to the problems and techniques of differential privacy, but is not intended to be an exhaustive survey — there is by now a vast amount of work in differential privacy, and we can cover only a small portion of it.

/pdf/the-algorithmic-foundations-of-differential-privacy-1n8bb9a7eo.pdf

The Algorithmic Foundations of Differential Privacy

Unsupervised node embedding methods (e.g., DeepWalk, LINE, and node2vec) have attracted growing interests given their simplicity and effectiveness. However, although these methods have been proved effective in a variety of applications, none of the existing work has analyzed the robustness of them. This could be very risky if these methods are attacked by an adversarial party. In this paper, we take the task of link prediction as an example, which is one of the most fundamental problems for graph analysis, and introduce a data positioning attack to node embedding methods. We give a complete characterization of attacker's utilities and present efficient solutions to adversarial attacks for two popular node embedding methods: DeepWalk and LINE. We evaluate our proposed attack model on multiple real-world graphs. Experimental results show that our proposed model can significantly affect the results of link prediction by slightly changing the graph structures (e.g., adding or removing a few edges). We also show that our proposed model is very general and can be transferable across different embedding methods. Finally, we conduct a case study on a coauthor network to better understand our attack method.

/pdf/data-poisoning-attack-against-unsupervised-node-embedding-2pkoc7o5hw.pdf

Data Poisoning Attack against Unsupervised Node Embedding Methods

The standard approach for privilege separation in web applications is to execute application components in different web origins. This limits the practicality of privilege separation since each web origin has financial and administrative cost. In this paper, we propose a new design for achieving effective privilege separation in HTML5 applications that shows how applications can cheaply create arbitrary number of components. Our approach utilizes standardized abstractions already implemented in modern browsers. We do not advocate any changes to the underlying browser or require learning new high-level languages, which contrasts prior approaches. We empirically show that we can retrofit our design to real-world HTML5 applications (browser extensions and rich client-side applications) and achieve reduction of 6x to 10000x in TCB for our case studies. Our mechanism requires less than 13 lines of application-specific code changes and considerably improves auditability.

/pdf/privilege-separation-in-html5-applications-3klp8vunjb.pdf

Privilege separation in HTML5 applications

: In this paper we consider the problem of privately computing the intersection of sets (set-intersection), as well as several variations on this problem: cardinality set-intersection, threshold set-intersection, and over-threshold set-intersection. Cardinality set-intersection is the problem of determining the size of the intersection set, without revealing the actual threshold number t times in the players' private inputs are revealed. Over-threshold set-intersection is a variation on threshold set-intersection in which not only the threshold set is revealed, but also the number of times each element in the threshold set appeared in the private inputs.

/pdf/private-and-threshold-set-intersection-1w76qbcg9d.pdf

Private and threshold set-intersection

The underground malware-based economy is flourishing and it is evident that the classical ad-hoc signature detection methods are becoming insufficient. Malware authors seem to share some source code and malware samples often feature similar behaviors, but such commonalities are difficult to detect with signature-based methods because of an increasing use of numerous freelyavailable randomized obfuscation tools. To address this problem, the security community is actively researching behavioral detection methods that commonly attempt to understand and differentiate how malware behaves, as opposed to just detecting syntactic patterns. We continue that line of research in this paper and explore how formal methods and tools of the verification trade could be used for malware detection and analysis. We propose a new approach to learning and generalizing from observed malware behaviors based on tree automata inference. In particular, we develop an algorithm for inferring k-testable tree automata from system call dataflow dependency graphs and discuss the use of inferred automata in malware recognition and classification.

/pdf/malware-analysis-with-tree-automata-inference-1euaxpj8wf.pdf

Malware analysis with tree automata inference

Training deep neural networks from scratch could be computationally expensive and requires a lot of training data. Recent work has explored different watermarking techniques to protect the pre-trained deep neural networks from potential copyright infringements. However, these techniques could be vulnerable to watermark removal attacks. In this work, we propose REFIT, a unified watermark removal framework based on fine-tuning, which does not rely on the knowledge of the watermarks, and is effective against a wide range of watermarking schemes. In particular, we conduct a comprehensive study of a realistic attack scenario where the adversary has limited training data, which has not been emphasized in prior work on attacks against watermarking schemes. To effectively remove the watermarks without compromising the model functionality under this weak threat model, we propose two techniques that are incorporated into our fine-tuning framework: (1) an adaption of the elastic weight consolidation (EWC) algorithm, which is originally proposed for mitigating the catastrophic forgetting phenomenon; and (2) unlabeled data augmentation (AU), where we leverage auxiliary unlabeled data from other sources. Our extensive evaluation shows the effectiveness of REFIT against diverse watermark embedding schemes. In particular, both EWC and AU significantly decrease the amount of labeled training data needed for effective watermark removal, and the unlabeled data samples used for AU do not necessarily need to be drawn from the same distribution as the benign data for model evaluation. The experimental results demonstrate that our fine-tuning based watermark removal attacks could pose real threats to the copyright of pre-trained models, and thus highlight the importance of further investigating the watermarking problem and proposing more robust watermark embedding schemes against the attacks.

Dawn Song

Papers

Data Poisoning Attack against Unsupervised Node Embedding Methods

Privilege separation in HTML5 applications

Private and threshold set-intersection

Malware analysis with tree automata inference

REFIT: a Unified Watermark Removal Framework for Deep Learning Systems with Limited Data