While large language models (LLMs) exhibit remarkable capabilities across various tasks, they encounter potential security risks such as jailbreak attacks, which exploit vulnerabilities to bypass security measures and generate harmful outputs. Existing jailbreak strategies mainly focus on maximizing attack success rate (ASR), frequently neglecting other critical factors, including the relevance of the jailbreak response to the query and the level of stealthiness. This narrow focus on single objectives can result in ineffective attacks that either lack contextual relevance or are easily recognizable. In this work, we introduce BlackDAN, an innovative black-box attack framework with multi-objective optimization, aiming to generate high-quality prompts that effectively facilitate jailbreaking while maintaining contextual relevance and minimizing detectability. BlackDAN leverages Multiobjective Evolutionary Algorithms (MOEAs), specifically the NSGA-II algorithm, to optimize jailbreaks across multiple objectives including ASR, stealthiness, and semantic relevance. By integrating mechanisms like mutation, crossover, and Pareto-dominance, BlackDAN provides a transparent and interpretable process for generating jailbreaks. Furthermore, the framework allows customization based on user preferences, enabling the selection of prompts that balance harmfulness, relevance, and other factors. Experimental results demonstrate that BlackDAN outperforms traditional single-objective methods, yielding higher success rates and improved robustness across various LLMs and multimodal LLMs, while ensuring jailbreak responses are both relevant and less detectable.
相關內容
As command-line interfaces remain integral to high-performance computing environments, the risk of exploitation through stealthy and complex command-line abuse grows. Conventional security solutions struggle to detect these anomalies due to their context-specific nature, lack of labeled data, and the prevalence of sophisticated attacks like Living-off-the-Land (LOL). To address this gap, we introduce the Scalable Command-Line Anomaly Detection Engine (SCADE), a framework that combines global statistical models with local context-specific analysis for unsupervised anomaly detection. SCADE leverages novel statistical methods, including BM25 and Log Entropy, alongside dynamic thresholding to adaptively detect rare, malicious command-line patterns in low signal-to-noise ratio (SNR) environments. Experimental results show that SCADE achieves above 98% SNR in identifying anomalous behavior while minimizing false positives. Designed for scalability and precision, SCADE provides an innovative, metadata-enriched approach to anomaly detection, offering a robust solution for cybersecurity in high-computation environments. This work presents SCADE's architecture, detection methodology, and its potential for enhancing anomaly detection in enterprise systems. We argue that SCADE represents a significant advancement in unsupervised anomaly detection, offering a robust, adaptive framework for security analysts and researchers seeking to enhance detection accuracy in high-computation environments.
The use of large language models (LLMs) for qualitative analysis is gaining attention in various fields, including software engineering, where qualitative methods are essential for understanding human and social factors. This study aimed to investigate how LLMs are currently used in qualitative analysis and their potential applications in software engineering research, focusing on the benefits, limitations, and practices associated with their use. A systematic mapping study was conducted, analyzing 21 relevant studies to explore reported uses of LLMs for qualitative analysis. The findings indicate that LLMs are primarily used for tasks such as coding, thematic analysis, and data categorization, offering benefits like increased efficiency and support for new researchers. However, limitations such as output variability, challenges in capturing nuanced perspectives, and ethical concerns related to privacy and transparency were also identified. The study emphasizes the need for structured strategies and guidelines to optimize LLM use in qualitative research within software engineering, enhancing their effectiveness while addressing ethical considerations. While LLMs show promise in supporting qualitative analysis, human expertise remains crucial for interpreting data, and ongoing exploration of best practices will be vital for their successful integration into empirical software engineering research.
Motion forecasting (MF) for autonomous driving aims at anticipating trajectories of surrounding agents in complex urban scenarios. In this work, we investigate a mixed strategy in MF training that first pre-train motion forecasters on pseudo-labeled data, then fine-tune them on annotated data. To obtain pseudo-labeled trajectories, we propose a simple pipeline that leverages off-the-shelf single-frame 3D object detectors and non-learning trackers. The whole pre-training strategy including pseudo-labeling is coined as PPT. Our extensive experiments demonstrate that: (1) combining PPT with supervised fine-tuning on annotated data achieves superior performance on diverse testbeds, especially under annotation-efficient regimes, (2) scaling up to multiple datasets improves the previous state-of-the-art and (3) PPT helps enhance cross-dataset generalization. Our findings showcase PPT as a promising pre-training solution for robust motion forecasting in diverse autonomous driving contexts.
The outstanding capabilities of large language models (LLMs) render them a crucial component in various autonomous agent systems. While traditional methods depend on the inherent knowledge of LLMs without fine-tuning, more recent approaches have shifted toward the reinforcement learning strategy to further enhance agents' ability to solve complex interactive tasks with environments and tools. However, previous approaches are constrained by the sparse reward issue, where existing datasets solely provide a final scalar reward for each multi-step reasoning chain, potentially leading to ineffectiveness and inefficiency in policy learning. In this paper, we introduce StepAgent, which utilizes step-wise reward to optimize the agent's reinforcement learning process. Inheriting the spirit of novice-to-expert theory, we first compare the actions of the expert and the agent to automatically generate intermediate rewards for fine-grained optimization. Additionally, we propose implicit-reward and inverse reinforcement learning techniques to facilitate agent reflection and policy adjustment. Further theoretical analysis demonstrates that the action distribution of the agent can converge toward the expert action distribution over multiple training cycles. Experimental results across various datasets indicate that StepAgent outperforms existing baseline methods.
Proof of Stake (PoS) blockchains offer promising alternatives to traditional Proof of Work (PoW) systems, providing scalability and energy efficiency. However, blockchains operate in a decentralized manner and the network is composed of diverse users. This openness creates the potential for malicious nodes to disrupt the network in various ways. Therefore, it is crucial to embed a mechanism within the blockchain network to constantly monitor, identify, and eliminate these malicious nodes without involving any central authority. In this paper, we propose MRL-PoS+, a novel consensus algorithm to enhance the security of PoS blockchains by leveraging Multi-agent Reinforcement Learning (MRL) techniques. Our proposed consensus algorithm introduces a penalty-reward scheme for detecting and eliminating malicious nodes. This approach involves the detection of behaviors that can lead to potential attacks in a blockchain network and hence penalizes the malicious nodes, restricting them from performing certain actions. Our developed Proof of Concept demonstrates effectiveness in eliminating malicious nodes for six types of major attacks. Experimental results demonstrate that MRL-PoS+ significantly improves the attack resilience of PoS blockchains compared to the traditional schemes without incurring additional computation overhead.
We develop a unifying framework for information-theoretic lower bound in statistical estimation and interactive decision making. Classical lower bound techniques -- such as Fano's method, Le Cam's method, and Assouad's lemma -- are central to the study of minimax risk in statistical estimation, yet are insufficient to provide tight lower bounds for \emph{interactive decision making} algorithms that collect data interactively (e.g., algorithms for bandits and reinforcement learning). Recent work of Foster et al. (2021, 2023) provides minimax lower bounds for interactive decision making using seemingly different analysis techniques from the classical methods. These results -- which are proven using a complexity measure known as the \emph{Decision-Estimation Coefficient} (DEC) -- capture difficulties unique to interactive learning, yet do not recover the tightest known lower bounds for passive estimation. We propose a unified view of these distinct methodologies through a new lower bound approach called \emph{interactive Fano method}. As an application, we introduce a novel complexity measure, the \emph{Fractional Covering Number}, which facilitates the new lower bounds for interactive decision making that extend the DEC methodology by incorporating the complexity of estimation. Using the fractional covering number, we (i) provide a unified characterization of learnability for \emph{any} stochastic bandit problem, (ii) close the remaining gap between the upper and lower bounds in Foster et al. (2021, 2023) (up to polynomial factors) for any interactive decision making problem in which the underlying model class is convex.
Large language models (LLMs) have shown limitations in tasks requiring complex logical reasoning and multi-step problem-solving. To address these challenges, researchers have employed carefully designed prompts and flowcharts, simulating human cognitive processes to enhance LLM performance, such as the Chain of Thought approach. In this paper, we introduce MTMT (Multi-thinking Modes Tree), a novel method that interacts with LLMs to construct a thought tree, simulating various advanced cognitive processes, including but not limited to association, counterfactual thinking, task decomposition, and comparison. By breaking down the original complex task into simpler sub-questions, MTMT facilitates easier problem-solving for LLMs, enabling more effective utilization of the latent knowledge within LLMs. We evaluate the performance of MTMT under different parameter configurations, using GPT-4o mini as the base model. Our results demonstrate that integrating multiple modes of thinking significantly enhances the ability of LLMs to handle complex tasks.
Training and serving long-context large language models (LLMs) incurs substantial overhead. To address this, two critical steps are often required: a pretrained LLM typically undergoes a separate stage for context length extension by training on long-context data, followed by architectural modifications to reduce the overhead of KV cache during serving. This paper argues that integrating length extension with a GPU-friendly KV cache reduction architecture not only reduces training overhead during length extension, but also achieves better long-context performance. This leads to our proposed LongGen, which finetunes a pretrained LLM into an efficient architecture during length extension. LongGen builds on three key insights: (1) Sparse attention patterns, such as window attention (attending to recent tokens), attention sink (initial ones), and blockwise sparse attention (strided token blocks) are well-suited for building efficient long-context models, primarily due to their GPU-friendly memory access patterns, enabling efficiency gains not just theoretically but in practice as well. (2) It is essential for the model to have direct access to all tokens. A hybrid architecture with 1/3 full attention layers and 2/3 efficient ones achieves a balanced trade-off between efficiency and long-context performance. (3) Lightweight training on 5B long-context data is sufficient to extend the hybrid model's context length from 4K to 128K. We evaluate LongGen on both Llama-2 7B and Llama-2 70B, demonstrating its effectiveness across different scales. During training with 128K-long contexts, LongGen achieves 1.55x training speedup and reduces wall-clock time by 36%, compared to a full-attention baseline. During inference, LongGen reduces KV cache memory by 62%, achieving 1.67x prefilling speedup and 1.41x decoding speedup.
While large language models (LLMs) have demonstrated remarkable capabilities across a range of downstream tasks, a significant concern revolves around their propensity to exhibit hallucinations: LLMs occasionally generate content that diverges from the user input, contradicts previously generated context, or misaligns with established world knowledge. This phenomenon poses a substantial challenge to the reliability of LLMs in real-world scenarios. In this paper, we survey recent efforts on the detection, explanation, and mitigation of hallucination, with an emphasis on the unique challenges posed by LLMs. We present taxonomies of the LLM hallucination phenomena and evaluation benchmarks, analyze existing approaches aiming at mitigating LLM hallucination, and discuss potential directions for future research.
The emergence of large language models (LLMs) has substantially influenced natural language processing, demonstrating exceptional results across various tasks. In this study, we employ ``Introspective Tips" to facilitate LLMs in self-optimizing their decision-making. By introspectively examining trajectories, LLM refines its policy by generating succinct and valuable tips. Our method enhances the agent's performance in both few-shot and zero-shot learning situations by considering three essential scenarios: learning from the agent's past experiences, integrating expert demonstrations, and generalizing across diverse games. Importantly, we accomplish these improvements without fine-tuning the LLM parameters; rather, we adjust the prompt to generalize insights from the three aforementioned situations. Our framework not only supports but also emphasizes the advantage of employing LLM in in-contxt decision-making. Experiments involving over 100 games in TextWorld illustrate the superior performance of our approach.
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