Recent years have witnessed growing interest in adopting large language models (LLMs) for Register Transfer Level (RTL) code optimization. While powerful cloud-based LLMs offer superior optimization capabilities, they pose unacceptable intellectual property (IP) leakage risks when processing proprietary hardware designs. In this paper, we propose a new scenario where Verilog code must be optimized for specific attributes without leaking sensitive IP information. We introduce the first IP-preserving edge-cloud collaborative framework that leverages the benefits of both paradigms. Our approach employs local small LLMs (e.g., Qwen-2.5-Coder-7B) to perform secure comparative analysis between paired high-quality target designs and novice draft codes, yielding general design principles that summarize key insights for improvements. These principles are then used to query stronger cloud LLMs (e.g., Deepseek-V3) for targeted code improvement, ensuring that only abstracted and IP-safe guidance reaches external services. Our experimental results demonstrate that the framework achieves significantly higher optimization success rates compared to baseline methods. For example, combining Qwen-2.5-Coder-7B and Deepseek-V3 achieves a 66.67\% optimization success rate for power utilization, outperforming Deepseek-V3 alone (49.81\%) and even commercial models like GPT-4o (55.81\%). Further investigation of local and cloud LLM combinations reveals that different model pairings exhibit varying strengths for specific optimization objectives, with interesting trends emerging when varying the number of comparative code pairs. Our work establishes a new paradigm for secure hardware design optimization that balances performance gains with IP protection.

Recent advances in LLM agentic systems have improved the automation of offensive security tasks, particularly for Capture the Flag (CTF) challenges. We systematically investigate the key factors that drive agent success and provide a detailed recipe for building effective LLM-based offensive security agents. First, we present CTFJudge, a framework leveraging LLM as a judge to analyze agent trajectories and provide granular evaluation across CTF solving steps. Second, we propose a novel metric, CTF Competency Index (CCI) for partial correctness, revealing how closely agent solutions align with human-crafted gold standards. Third, we examine how LLM hyperparameters, namely temperature, top-p, and maximum token length, influence agent performance and automated cybersecurity task planning. For rapid evaluation, we present CTFTiny, a curated benchmark of 50 representative CTF challenges across binary exploitation, web, reverse engineering, forensics, and cryptography. Our findings identify optimal multi-agent coordination settings and lay the groundwork for future LLM agent research in cybersecurity. We make CTFTiny open source to public https://github.com/NYU-LLM-CTF/CTFTiny along with CTFJudge on https://github.com/NYU-LLM-CTF/CTFJudge.

As large language models (LLMs) and generative AI become increasingly integrated into customer service and moderation applications, adversarial threats emerge from both external manipulations and internal label corruption. In this work, we identify and systematically address these dual adversarial threats by introducing DINA (Dual Defense Against Internal Noise and Adversarial Attacks), a novel unified framework tailored specifically for NLP. Our approach adapts advanced noisy-label learning methods from computer vision and integrates them with adversarial training to simultaneously mitigate internal label sabotage and external adversarial perturbations. Extensive experiments conducted on a real-world dataset from an online gaming service demonstrate that DINA significantly improves model robustness and accuracy compared to baseline models. Our findings not only highlight the critical necessity of dual-threat defenses but also offer practical strategies for safeguarding NLP systems in realistic adversarial scenarios, underscoring broader implications for fair and responsible AI deployment.

Game theory has long served as a foundational tool in cybersecurity to test, predict, and design strategic interactions between attackers and defenders. The recent advent of Large Language Models (LLMs) offers new tools and challenges for the security of computer systems; In this work, we investigate whether classical game-theoretic frameworks can effectively capture the behaviours of LLM-driven actors and bots. Using a reproducible framework for game-theoretic LLM agents, we investigate two canonical scenarios -- the one-shot zero-sum game and the dynamic Prisoner's Dilemma -- and we test whether LLMs converge to expected outcomes or exhibit deviations due to embedded biases. Our experiments involve four state-of-the-art LLMs and span five natural languages, English, French, Arabic, Vietnamese, and Mandarin Chinese, to assess linguistic sensitivity. For both games, we observe that the final payoffs are influenced by agents characteristics such as personality traits or knowledge of repeated rounds. Moreover, we uncover an unexpected sensitivity of the final payoffs to the choice of languages, which should warn against indiscriminate application of LLMs in cybersecurity applications and call for in-depth studies, as LLMs may behave differently when deployed in different countries. We also employ quantitative metrics to evaluate the internal consistency and cross-language stability of LLM agents, to help guide the selection of the most stable LLMs and optimising models for secure applications.

Accurately extracting and representing the structure of tabular data from financial documents remains a critical challenge in document understanding, particularly for regulatory and analytical use cases. This study addresses the complexity of converting financial tables from Malaysian audited financial reports into Markdown format, a task complicated by rotated layouts, multi-level headers, and implicit structural cues. We propose a fine-tuned vision-language model (VLM), based on Qwen2.5-VL-7B, optimized for high-fidelity Markdown generation from document images. Our approach includes a curated dataset of 2,152 image-text pairs with augmentations and a supervised fine-tuning strategy using LoRA. To assess performance, we evaluated our model on 100 out-of-sample tables using a dual framework: a criteria-based LLM-as-a-judge for fine-grained accuracy and our novel Markdown Tree-Edit-Distance-based Similarity (TEDS) metric for holistic structural fidelity. Our model achieves a 92.20% overall accuracy on the criteria-based assessment and a 96.53% Markdown TEDS score. This performance significantly surpasses its Qwen2.5-VL-7B base model, larger-scale VLMs, and specialized reasoning-enabled models. Compared to these self-hosted alternatives, it also significantly reduces inference time. Furthermore, its accuracy exceeds that of widely used proprietary models such as OpenAI's GPT-4o and Gemini 2.5 Flash. These results demonstrate that domain-specific fine-tuning provides an effective and efficient method to bridge the gap between unstructured financial documents and downstream automation, rivalling much larger and more general models without their computational overhead.

We present HySemRAG, a framework that combines Extract, Transform, Load (ETL) pipelines with Retrieval-Augmented Generation (RAG) to automate large-scale literature synthesis and identify methodological research gaps. The system addresses limitations in existing RAG architectures through a multi-layered approach: hybrid retrieval combining semantic search, keyword filtering, and knowledge graph traversal; an agentic self-correction framework with iterative quality assurance; and post-hoc citation verification ensuring complete traceability. Our implementation processes scholarly literature through eight integrated stages: multi-source metadata acquisition, asynchronous PDF retrieval, custom document layout analysis using modified Docling architecture, bibliographic management, LLM-based field extraction, topic modeling, semantic unification, and knowledge graph construction. The system creates dual data products - a Neo4j knowledge graph enabling complex relationship queries and Qdrant vector collections supporting semantic search - serving as foundational infrastructure for verifiable information synthesis. Evaluation across 643 observations from 60 testing sessions demonstrates structured field extraction achieving 35.1% higher semantic similarity scores (0.655 $\pm$ 0.178) compared to PDF chunking approaches (0.485 $\pm$ 0.204, p < 0.000001). The agentic quality assurance mechanism achieves 68.3% single-pass success rates with 99.0% citation accuracy in validated responses. Applied to geospatial epidemiology literature on ozone exposure and cardiovascular disease, the system identifies methodological trends and research gaps, demonstrating broad applicability across scientific domains for accelerating evidence synthesis and discovery.

Many AI customer service systems use standard NLP pipelines or finetuned language models, which often fall short on ambiguous, multi-intent, or detail-specific queries. This case study evaluates recent techniques: query rewriting, RAG Fusion, keyword augmentation, intent recognition, and context reranking, for building a robust customer support system in the electric power domain. We compare vector-store and graph-based RAG frameworks, ultimately selecting the graph-based RAG for its superior performance in handling complex queries. We find that query rewriting improves retrieval for queries using non-standard terminology or requiring precise detail. RAG Fusion boosts performance on vague or multifaceted queries by merging multiple retrievals. Reranking reduces hallucinations by filtering irrelevant contexts. Intent recognition supports the decomposition of complex questions into more targeted sub-queries, increasing both relevance and efficiency. In contrast, keyword augmentation negatively impacts results due to biased keyword selection. Our final system combines intent recognition, RAG Fusion, and reranking to handle disambiguation and multi-source queries. Evaluated on both a GPT-4-generated dataset and a real-world electricity provider FAQ dataset, it achieves 97.9% and 89.6% accuracy respectively, substantially outperforming baseline RAG models.

Dynamic streams from news feeds, social media, sensor networks, and financial markets challenge static RAG frameworks. Full-scale indices incur high memory costs; periodic rebuilds introduce latency that undermines data freshness; naive sampling sacrifices semantic coverage. We present Streaming RAG, a unified pipeline that combines multi-vector cosine screening, mini-batch clustering, and a counter-based heavy-hitter filter to maintain a compact prototype set. We further prove an approximation bound \$E\[R(K\_t)] \ge R^\* - L Δ\$ linking retrieval quality to clustering variance. An incremental index upsert mechanism refreshes prototypes without interrupting queries. Experiments on eight real-time streams show statistically significant gains in Recall\@10 (up to 3 points, p < 0.01), end-to-end latency below 15 ms, and throughput above 900 documents per second under a 150 MB budget. Hyperparameter sensitivity analysis over cluster count, admission probability, relevance threshold, and counter capacity validates default settings. In open-domain question answering with GPT-3.5 Turbo, we record 3.2-point gain in Exact Match and 2.8-point gain in F1 on SQuAD; abstractive summarization yields ROUGE-L improvements. Streaming RAG establishes a new Pareto frontier for retrieval augmentation.

Visual search offers an intuitive way for customers to explore diverse product catalogs, particularly in consumer-to-consumer (C2C) marketplaces where listings are often unstructured and visually driven. This paper presents a scalable visual search system deployed in Mercari's C2C marketplace, where end-users act as buyers and sellers. We evaluate recent vision-language models for zero-shot image retrieval and compare their performance with an existing fine-tuned baseline. The system integrates real-time inference and background indexing workflows, supported by a unified embedding pipeline optimized through dimensionality reduction. Offline evaluation using user interaction logs shows that the multilingual SigLIP model outperforms other models across multiple retrieval metrics, achieving a 13.3% increase in nDCG@5 over the baseline. A one-week online A/B test in production further confirms real-world impact, with the treatment group showing substantial gains in engagement and conversion, up to a 40.9% increase in transaction rate via image search. Our findings highlight that recent zero-shot models can serve as a strong and practical baseline for production use, which enables teams to deploy effective visual search systems with minimal overhead, while retaining the flexibility to fine-tune based on future data or domain-specific needs.

The surge in scientific publications challenges traditional review methods, demanding tools that integrate structured metadata with full-text analysis. Hybrid Retrieval Augmented Generation (RAG) systems, combining graph queries with vector search offer promise but are typically static, rely on proprietary tools, and lack uncertainty estimates. We present an agentic approach that encapsulates the hybrid RAG pipeline within an autonomous agent capable of (1) dynamically selecting between GraphRAG and VectorRAG for each query, (2) adapting instruction-tuned generation in real time to researcher needs, and (3) quantifying uncertainty during inference. This dynamic orchestration improves relevance, reduces hallucinations, and promotes reproducibility. Our pipeline ingests bibliometric open-access data from PubMed, arXiv, and Google Scholar APIs, builds a Neo4j citation-based knowledge graph (KG), and embeds full-text PDFs into a FAISS vector store (VS) using the all-MiniLM-L6-v2 model. A Llama-3.3-70B agent selects GraphRAG (translating queries to Cypher for KG) or VectorRAG (combining sparse and dense retrieval with re-ranking). Instruction tuning refines domain-specific generation, and bootstrapped evaluation yields standard deviation for evaluation metrics. On synthetic benchmarks mimicking real-world queries, the Instruction-Tuned Agent with Direct Preference Optimization (DPO) outperforms the baseline, achieving a gain of 0.63 in VS Context Recall and a 0.56 gain in overall Context Precision. Additional gains include 0.24 in VS Faithfulness, 0.12 in both VS Precision and KG Answer Relevance, 0.11 in overall Faithfulness score, 0.05 in KG Context Recall, and 0.04 in both VS Answer Relevance and overall Precision. These results highlight the system's improved reasoning over heterogeneous sources and establish a scalable framework for autonomous, agentic scientific discovery.