It is universally acknowledged that Wi-Fi communications are important to secure. Thus, the Wi-Fi Alliance published WPA3 in 2018 with a distinctive security feature: it leverages a Password-Authenticated Key Exchange (PAKE) protocol to protect users' passwords from offline dictionary attacks. Unfortunately, soon after its release, several attacks were reported against its implementations, in response to which the protocol was updated in a best-effort manner. In this paper, we show that the proposed mitigations are not enough, especially for a complex protocol to implement even for savvy developers. Indeed, we present **Dragondoom**, a collection of side-channel vulnerabilities of varying strength allowing attackers to recover users' passwords in widely deployed Wi-Fi daemons, such as hostap in its default settings. Our findings target both password conversion methods, namely the default probabilistic hunting-and-pecking and its newly standardized deterministic alternative based on SSWU. We successfully exploit our leakage in practice through microarchitectural mechanisms, and overcome the limited spatial resolution of Flush+Reload. Our attacks outperform previous works in terms of required measurements. Then, driven by the need to end the spiral of patch-and-hack in Dragonfly implementations, we propose **Dragonstar**, an implementation of Dragonfly leveraging a formally verified implementation of the underlying mathematical operations, thereby removing all the related leakage vector. Our implementation relies on HACL*, a formally verified crypto library guaranteeing secret-independence. We design Dragonstar, so that its integration within hostap requires minimal modifications to the existing project. Our experiments show that the performance of HACL*-based hostap is comparable to OpenSSL-based, implying that Dragonstar is both efficient and proved to be leakage-free.
相關內容
The use of modern Natural Language Processing (NLP) techniques has shown to be beneficial for software engineering tasks, such as vulnerability detection and type inference. However, training deep NLP models requires significant computational resources. This paper explores techniques that aim at achieving the best usage of resources and available information in these models. We propose a generic approach, EarlyBIRD, to build composite representations of code from the early layers of a pre-trained transformer model. We empirically investigate the viability of this approach on the CodeBERT model by comparing the performance of 12 strategies for creating composite representations with the standard practice of only using the last encoder layer. Our evaluation on four datasets shows that several early layer combinations yield better performance on defect detection, and some combinations improve multi-class classification. More specifically, we obtain a +2 average improvement of detection accuracy on Devign with only 3 out of 12 layers of CodeBERT and a 3.3x speed-up of fine-tuning. These findings show that early layers can be used to obtain better results using the same resources, as well as to reduce resource usage during fine-tuning and inference.
The Mixture of Experts (MoE) is a widely known neural architecture where an ensemble of specialized sub-models optimizes overall performance with a constant computational cost. However, conventional MoEs pose challenges at scale due to the need to store all experts in memory. In this paper, we push MoE to the limit. We propose extremely parameter-efficient MoE by uniquely combining MoE architecture with lightweight experts.Our MoE architecture outperforms standard parameter-efficient fine-tuning (PEFT) methods and is on par with full fine-tuning by only updating the lightweight experts -- less than 1% of an 11B parameters model. Furthermore, our method generalizes to unseen tasks as it does not depend on any prior task knowledge. Our research underscores the versatility of the mixture of experts architecture, showcasing its ability to deliver robust performance even when subjected to rigorous parameter constraints. Our code used in all the experiments is publicly available here: //github.com/for-ai/parameter-efficient-moe.
Jailbreak vulnerabilities in Large Language Models (LLMs), which exploit meticulously crafted prompts to elicit content that violates service guidelines, have captured the attention of research communities. While model owners can defend against individual jailbreak prompts through safety training strategies, this relatively passive approach struggles to handle the broader category of similar jailbreaks. To tackle this issue, we introduce FuzzLLM, an automated fuzzing framework designed to proactively test and discover jailbreak vulnerabilities in LLMs. We utilize templates to capture the structural integrity of a prompt and isolate key features of a jailbreak class as constraints. By integrating different base classes into powerful combo attacks and varying the elements of constraints and prohibited questions, FuzzLLM enables efficient testing with reduced manual effort. Extensive experiments demonstrate FuzzLLM's effectiveness and comprehensiveness in vulnerability discovery across various LLMs.
Diffusion Probabilistic Models (DPMs) have achieved considerable success in generation tasks. As sampling from DPMs is equivalent to solving diffusion SDE or ODE which is time-consuming, numerous fast sampling methods built upon improved differential equation solvers are proposed. The majority of such techniques consider solving the diffusion ODE due to its superior efficiency. However, stochastic sampling could offer additional advantages in generating diverse and high-quality data. In this work, we engage in a comprehensive analysis of stochastic sampling from two aspects: variance-controlled diffusion SDE and linear multi-step SDE solver. Based on our analysis, we propose SA-Solver, which is an improved efficient stochastic Adams method for solving diffusion SDE to generate data with high quality. Our experiments show that SA-Solver achieves: 1) improved or comparable performance compared with the existing state-of-the-art sampling methods for few-step sampling; 2) SOTA FID scores on substantial benchmark datasets under a suitable number of function evaluations (NFEs).
The rapid development of Blockchain technology and the prosperity of cryptocurrency in the past decade have driven the massive demand for digital assets trading, leading to the emergence of many cryptocurrency exchange platforms. Unlike centralised exchanges (CEXs) where listed tokens and cryptocurrencies are assessed by authorities to make the secured trading environment, decentralized exchanges (DEXs) are introduced to allow users to trade their digital assets without the involvement of any third party, therefore exposing security issues and encouraging the rise of many scams and malicious tokens. In this paper, we investigate an emerging malicious token named Trapdoor, which allows users to buy but prevent them from selling and getting their funds back. The first collection of Trapdoor tokens is constructed in this study by investigating malicious behaviours and maneuvers of these tokens. After manually analysing the tokens' source code, we classify those Trapdoor tokens into different categories according to their malicious code embedding technique. Moreover, we also comprehensively analyse the impact of Trapdoor tokens, the behaviours of scammers, and the characteristics of victims from various perspective. Finally, we also implement and publish our Trapdoor token detection tool and Trapdoor maneuvers analysis reports that help in increasing awareness of investors for this kind of scam.
We present an expanded study of the performance of FLASH when using Linux Kernel Hugepages on Ookami, an HPE Apollo 80 A64FX platform. FLASH is a multi-scale, multi-physics simulation code written principally in modern Fortran and makes use of the PARAMESH library to manage a block-structured adaptive mesh. Our initial study used only the Fujitsu compiler to utilize standard hugepages (hp), but further investigation allowed us to utilize hp for multiple compilers by linking to the Fujitsu library libmpg and transparent hugepages (thp) by enabling it at the node level. By comparing the results of hardware counters and in-code timers, we found that hp and thp do not significantly impact the runtime performance of FLASH. Interestingly, there is a significant reduction in the TLB misses, differences in cache and memory access counters, and strange behavior is observed when using thp.
Inspired by the remarkable success of Latent Diffusion Models (LDMs) for image synthesis, we study LDM for text-to-video generation, which is a formidable challenge due to the computational and memory constraints during both model training and inference. A single LDM is usually only capable of generating a very limited number of video frames. Some existing works focus on separate prediction models for generating more video frames, which suffer from additional training cost and frame-level jittering, however. In this paper, we propose a framework called "Reuse and Diffuse" dubbed $\textit{VidRD}$ to produce more frames following the frames already generated by an LDM. Conditioned on an initial video clip with a small number of frames, additional frames are iteratively generated by reusing the original latent features and following the previous diffusion process. Besides, for the autoencoder used for translation between pixel space and latent space, we inject temporal layers into its decoder and fine-tune these layers for higher temporal consistency. We also propose a set of strategies for composing video-text data that involve diverse content from multiple existing datasets including video datasets for action recognition and image-text datasets. Extensive experiments show that our method achieves good results in both quantitative and qualitative evaluations. Our project page is available $\href{//anonymous0x233.github.io/ReuseAndDiffuse/}{here}$.
Autonomous driving has achieved a significant milestone in research and development over the last decade. There is increasing interest in the field as the deployment of self-operating vehicles on roads promises safer and more ecologically friendly transportation systems. With the rise of computationally powerful artificial intelligence (AI) techniques, autonomous vehicles can sense their environment with high precision, make safe real-time decisions, and operate more reliably without human interventions. However, intelligent decision-making in autonomous cars is not generally understandable by humans in the current state of the art, and such deficiency hinders this technology from being socially acceptable. Hence, aside from making safe real-time decisions, the AI systems of autonomous vehicles also need to explain how these decisions are constructed in order to be regulatory compliant across many jurisdictions. Our study sheds a comprehensive light on developing explainable artificial intelligence (XAI) approaches for autonomous vehicles. In particular, we make the following contributions. First, we provide a thorough overview of the present gaps with respect to explanations in the state-of-the-art autonomous vehicle industry. We then show the taxonomy of explanations and explanation receivers in this field. Thirdly, we propose a framework for an architecture of end-to-end autonomous driving systems and justify the role of XAI in both debugging and regulating such systems. Finally, as future research directions, we provide a field guide on XAI approaches for autonomous driving that can improve operational safety and transparency towards achieving public approval by regulators, manufacturers, and all engaged stakeholders.
Graph Neural Networks (GNNs) have been studied from the lens of expressive power and generalization. However, their optimization properties are less well understood. We take the first step towards analyzing GNN training by studying the gradient dynamics of GNNs. First, we analyze linearized GNNs and prove that despite the non-convexity of training, convergence to a global minimum at a linear rate is guaranteed under mild assumptions that we validate on real-world graphs. Second, we study what may affect the GNNs' training speed. Our results show that the training of GNNs is implicitly accelerated by skip connections, more depth, and/or a good label distribution. Empirical results confirm that our theoretical results for linearized GNNs align with the training behavior of nonlinear GNNs. Our results provide the first theoretical support for the success of GNNs with skip connections in terms of optimization, and suggest that deep GNNs with skip connections would be promising in practice.
Deep neural networks (DNNs) are successful in many computer vision tasks. However, the most accurate DNNs require millions of parameters and operations, making them energy, computation and memory intensive. This impedes the deployment of large DNNs in low-power devices with limited compute resources. Recent research improves DNN models by reducing the memory requirement, energy consumption, and number of operations without significantly decreasing the accuracy. This paper surveys the progress of low-power deep learning and computer vision, specifically in regards to inference, and discusses the methods for compacting and accelerating DNN models. The techniques can be divided into four major categories: (1) parameter quantization and pruning, (2) compressed convolutional filters and matrix factorization, (3) network architecture search, and (4) knowledge distillation. We analyze the accuracy, advantages, disadvantages, and potential solutions to the problems with the techniques in each category. We also discuss new evaluation metrics as a guideline for future research.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191