arXiv subjects

Xinbo Wu

Publications and source records attributed to Xinbo Wu.

9 recordsLinked to original sources

arXiv2026

A Graph-Based Stackelberg Security Game for Trustworthy 6G Disaggregated Architecture

Open, cloud-native, and AI-enabled 6G architectures make network functions easier to deploy, observe, and replace, often implemented with separate software modules. However, each explicit interface can also create an entry point, trust transition, and lateral-movement route from a cybersecurity perspective. This paper formulates architectural disaggregation as a graph-based Stackelberg design game for 6G system security. A deployment graph specifies candidate 6G system seams, while an architecture-dependent directed attack graph specifies externally reachable functions and multistage paths. The defender commits to seams and allocates hardening and monitoring; an informed attacker then chooses a path and effort. Because edge traversal and evasion factors compose multiplicatively, the attacker's path response becomes a shortest-path problem after a logarithmic transformation. Convex effort costs admit a conjugate representation, and, for each fixed architecture, the defender's control problem is a convex exponential-epigraph program solvable by path generation. We derive an architecture-comparison criterion, a threshold for AI transducers within 6G systems that jointly add visibility and exposure, and comparative statics for attacker entry and bypass innovation. Experiments with stylized standards-anchored O-RAN/6G systems over RAN, RIC, core, cloud, and edge-AI functions show that neither full integration nor maximal disaggregation is generally best. Uncontrolled seams can increase risk, whereas selectively monitored and hardened seams can improve the Stackelberg objective. We find a boundary is worthwhile only when its detection and containment gains exceed its added attack opportunities and coordination cost.

cs.CR↗

arXiv2026

A Theoretical Game of Attacks via Compositional Skills

As large language models grow increasingly capable, concerns about their safe deployment have intensified. While numerous alignment strategies aim to restrict harmful behavior, these defenses can still be circumvented through carefully designed adversarial prompts. In this work, we introduce a theoretical framework that formalizes a game between an attacker and a defender. Within this framework, we design a theoretical best-response attack strategy and show that it is closely related to many existing adversarial prompting methods. We further analyze the resulting game, characterize its equilibria, and reveal inherent advantages for the attacker. Drawing on our theoretical analysis, we also derive a provably optimal defense strategy. Empirically, we evaluate a practical instantiation of the theoretically optimal attack and observe stronger performance relative to existing adversarial prompting approaches in diverse settings encompassing different LLMs and benchmarks.

cs.CL↗

arXiv2026

DeepInsert: Early Layer Bypass for Efficient and Performant Multimodal Understanding

Hyperscaling of data and parameter count in LLMs is yielding diminishing improvement when weighed against training costs, underlining a growing need for more efficient finetuning and inference without sacrificing performance. This is especially so for multimodal language models (MLMs), where the overhead of processing multimodal tokens can limit their practical viability. Parallely, recent work has uncovered implicit cross-modal alignment in the deeper layers of large MLMs, deepening our understanding of how MLMs process and encode information. Motivated by this, and our observation that MLMs naturally defer most cross-modal token interactions to deeper layers of the model, we propose a simple modification. Instead of concatenation with the language prompt at the start, we insert multimodal tokens directly into the middle, allowing them to entirely bypass the early layers. Our results with diverse modalities, (i) LLaVA \& BLIP for vision, (ii) LTU for audio, and (iii) MoLCA for molecular data, and model sizes, starting from 350M to 13B parameters, indicate that our method reduces both training and inference costs, while at least preserving, if not surpassing the performance of existing baselines.

cs.CV↗

arXiv2025

Concealment of Intent: A Game-Theoretic Analysis

As large language models (LLMs) grow more capable, concerns about their safe deployment have also grown. Although alignment mechanisms have been introduced to deter misuse, they remain vulnerable to carefully designed adversarial prompts. In this work, we present a scalable attack strategy: intent-hiding adversarial prompting, which conceals malicious intent through the composition of skills. We develop a game-theoretic framework to model the interaction between such attacks and defense systems that apply both prompt and response filtering. Our analysis identifies equilibrium points and reveals structural advantages for the attacker. To counter these threats, we propose and analyze a defense mechanism tailored to intent-hiding attacks. Empirically, we validate the attack's effectiveness on multiple real-world LLMs across a range of malicious behaviors, demonstrating clear advantages over existing adversarial prompting techniques.

cs.CL↗

arXiv2025

ITBench: Evaluating AI Agents across Diverse Real-World IT Automation Tasks

Realizing the vision of using AI agents to automate critical IT tasks depends on the ability to measure and understand effectiveness of proposed solutions. We introduce ITBench, a framework that offers a systematic methodology for benchmarking AI agents to address real-world IT automation tasks. Our initial release targets three key areas: Site Reliability Engineering (SRE), Compliance and Security Operations (CISO), and Financial Operations (FinOps). The design enables AI researchers to understand the challenges and opportunities of AI agents for IT automation with push-button workflows and interpretable metrics. ITBench includes an initial set of 94 real-world scenarios, which can be easily extended by community contributions. Our results show that agents powered by state-of-the-art models resolve only 13.8% of SRE scenarios, 25.2% of CISO scenarios, and 0% of FinOps scenarios. We expect ITBench to be a key enabler of AI-driven IT automation that is correct, safe, and fast.

cs.AI↗

arXiv2024

SwitchCIT: Switching for Continual Instruction Tuning

Large language models (LLMs) and multimodal models (MMs) have exhibited impressive capabilities in various domains, particularly in general language understanding and visual reasoning. However, these models, trained on massive data, may not be finely optimized for specific tasks triggered by instructions. Continual instruction tuning is crucial to adapt a large model to evolving tasks and domains, ensuring their effectiveness and relevance across a wide range of applications. In the context of continual instruction tuning, where models are sequentially trained on different tasks, catastrophic forgetting can occur, leading to performance degradation on previously learned tasks. This work addresses the catastrophic forgetting in continual instruction learning through a switching mechanism for routing computations to parameter-efficient tuned models. We demonstrate the effectiveness of our method through experiments on continual instruction tuning of different natural language generation tasks and vision-language tasks. We also showcase the advantages of our proposed method in terms of efficiency, scalability, portability, and privacy preservation.

cs.CL↗

arXiv2024

Semantically Grounded QFormer for Efficient Vision Language Understanding

General purpose Vision Language Models (VLMs) have received tremendous interest in recent years, owing to their ability to learn rich vision-language correlations as well as their broad zero-shot competencies. One immensely popular line of work utilizes frozen unimodal models, by bridging vision representations to language using a trainable module called the QFormer. However, this method relies heavily on large-scale multimodal pretraining with huge computational overheads. To that end, we propose a more efficient framework for QFormer-based vision-language alignment. Our key idea relies on the observation that QFormer latents correspond more strongly to the frozen LLM's intermediate latent space. Consequently, instead of using QFormer latents as inputs to the LLM, we alter the framework by using the latents to directly condition the LLM latent space for image-to-text generation. We demonstrate the effectiveness of our approach against existing baselines in improving the efficiency of vision-language pretraining.

cs.CV↗

arXiv2024

A Meta-Learning Perspective on Transformers for Causal Language Modeling

The Transformer architecture has become prominent in developing large causal language models. However, mechanisms to explain its capabilities are not well understood. Focused on the training process, here we establish a meta-learning view of the Transformer architecture when trained for the causal language modeling task, by explicating an inner optimization process within the Transformer. Further, within the inner optimization, we discover and theoretically analyze a special characteristic of the norms of learned token representations within Transformer-based causal language models. Our analysis is supported by experiments in various settings.

cs.LG↗

arXiv2024

Transformer-based Causal Language Models Perform Clustering

Even though large language models (LLMs) have demonstrated remarkable capability in solving various natural language tasks, the capability of an LLM to follow human instructions is still a concern. Recent works have shown great improvements in the instruction-following capability via additional training for instruction-following tasks. However, the mechanisms responsible for effective instruction-following capabilities remain inadequately understood. Here, we introduce a simplified instruction-following task and use synthetic datasets to analyze a Transformer-based causal language model. Our findings suggest that the model learns task-specific information by clustering data within its hidden space, with this clustering process evolving dynamically during learning. We also demonstrate how this phenomenon assists the model in handling unseen instances, and validate our results in a more realistic setting. Furthermore, we present inspired applications regarding pre-training and alignment.

cs.CL↗