Info Theory

The paper characterizes the secure rate-distortion-perception (RDP) trade-off region for neural image compression over various noisy and noiseless channels, demonstrating that randomized distributed f…

The paper proposes viewing national AI development, specifically in France, as a 'national AI learning system' governed by a controlled balance between information injection and entropy dissipation, a…

STaR-KV introduces a novel, training-free KV cache compression framework that adaptively re-weights token importance across spatial, temporal, and distributional axes, significantly reducing GPU memor…

The paper establishes an information-theoretic upper bound on the combined functional capacity and perturbation retention of code LLMs, quantifying the security budget available for code generation.

The paper proposes an efficient and provably secure linguistic steganography method using range coding that achieves high embedding capacity and speed, outperforming existing methods.

The paper proposes a novel information-geometric framework to analyze LLM stability by integrating task utility, external entropy, and internal structural proxies, showing this composite score improve…

The paper establishes a theoretical information-theoretic bound proving that for Vision-Language-Action (VLA) models, capability and robustness cannot both be arbitrarily high, quantifying the trade-o…

This paper provides the first systematic threat analysis of State-Space Models (SSMs) in safety-critical applications, introducing novel attack classes and formal metrics to quantify their security an…

InfoMerge is a novel, training-free method that significantly compresses visual tokens for Video-LLMs by estimating temporal redundancy and allocating tokens based on content richness, achieving high…

The paper proposes a provably secure steganography scheme based on list decoding that significantly increases embedding capacity for Large Language Models (LLMs) compared to existing methods.

The paper proposes a dual-interventional framework to characterize how linguistic structures and contextual cues influence LLMs' spatial reasoning for navigation, finding that topological information…

The paper proposes a comprehensive application-layer reference monitor to detect and mitigate data exfiltration via covert channels embedded in LLM agent egress payloads across text, image, and audio…

The paper introduces Single-stage Sparse Retrieval (SSR), a method that replaces computationally expensive vector clustering with sparse autoencoding to achieve highly efficient multi-vector retrieval…

This paper demonstrates that Simple Power Analysis (SPA) can successfully extract secret session key bits from post-quantum cryptosystems, specifically during the key decapsulation phase, using only a…

The paper introduces covert quantum computing, a framework to ensure privacy in multi-tenant quantum cloud environments, and finds that while local crosstalk is limited, long-range coupling effects cr…

The paper introduces the Asymmetric Collaborative Framework (ACF), a novel method that enables robust covert communication between autonomous agents despite inherent cognitive asymmetry caused by dyna…

The paper introduces a novel third-order, rotation-invariant spherical bispectrum for watermarking panoramic images, enabling reliable watermark embedding and extraction under arbitrary 3D rotations.

MLEvolve is a novel self-evolving multi-agent framework that enables LLM agents to discover and optimize machine learning algorithms for complex, long-horizon tasks.

The paper introduces Score Broadcast and Decorrelation (SBD), a general theoretical framework that unifies broadcast-based credit assignment across various differentiable loss functions by leveraging…

The paper introduces Residualized Sparse Autoencoders (ReSAEs) to improve multi-layer interventions in transformers by training each layer on the residual activation, which better preserves cross-laye…