Latest papers
Distributed discrete-choice optimization in decentralized settings is often hard to explore and navigate: disentangling what other agents choose, how their choices are interdependent, and how they collectively reach a global objective…
Sequence models must decide what to write into memory and what to retain. In quantum and quantum-inspired sequence learning, nonlinear recurrent updates often require repeated circuit evaluations and sequential backpropagation through time,…
While large language models (LLMs) have advanced ID-based recommendation through Semantic ID (SID) modeling, existing SID generation frameworks largely follow a single-representation-then-quantization paradigm. This design faces two…
Spatially-varying BRDFs (SVBRDFs) are central to material representation in computer graphics, but their high-resolution, multi-channel, mipmapped textures impose a substantial storage burden. Existing compression methods face a fundamental…
LLM-based multi-agent systems have the potential to enable collective intelligence and scale toward solving highly complex tasks through coordinated ensembles of specialized agents. However, despite their theoretical potential, the…
The advent of telescopes with an integrated deformable mirror (DM) presents new challenges for adaptive optics (AO) systems. The alignment between the DM and wavefront sensor (WFS) is expected to regularly evolve during operations due to…
Federated fine-tuning of large language models (LLMs) enables collaborative training without exposing raw data. However, a recent attack, NeuroImprint [1] (arXiv:2606.20553), demonstrates that a malicious parameter server can corrupt a PEFT…
The objective of miniaturizing doped areas in silicon, with the ultimate goal of achieving atomic-precision doping, requires a fundamental understanding of the dopant incorporation process at the atomic level. We present a combined scanning…
Composite visualizations integrate multiple visualizations to represent complex datasets effectively, but their intrinsic composite designs often impose a high initial cognitive load on novice users. Existing visualization onboarding…
Recent work on LLM agents is shifting from external capability elicitation to capability internalization, enabling agents to retain useful skills without retrieval at inference time. On-policy self-distillation (OPSD) offers a promising…
Existing backdoor attacks often become effective immediately after backdoor implantation and may therefore be exposed before exploitation. Machine unlearning activated dormant backdoors mitigate such behavioral exposure by remaining…
We introduce Kazakov-Migdal (KM)-type gauge theories on graphs via the Artin-Ihara $L$-function, providing a unified description of the models proposed in prior works. Using harmonic analysis on the group manifold, we reformulate the…
SHACL shapes enable data graph validation, making automatic shape learning essential for knowledge graph applications. We investigate the well-known fitting approach to this task: given sets P and N of positive and negative example nodes…
Flow matching (FM) has become a popular action head paradigm for modern embodied models. However, as a conditional generative model, it does not explicitly expose its inherent uncertainty, producing faulty action chunks even when it…
The present work investigates the asymptotic behaviours at the zero-noise limit of the first near collision-time and first near collision-location between a pair of independent $d$-dimensional Brownian-driven self-stabilizing (McKean-Vlasov…
This article analyses whether gender shapes Members of the European Parliament's communication on X/Twitter and whether equality-related and LGBTQ+ issues influence patterns of visibility, tone and interaction. Drawing on automated content,…
In the pursuit of fault-tolerant quantum computing, low-latency quantum error correction (QEC) is essential to prevent rapid error accumulation within the syndrome measurement cycle. In this work, we propose a microarchitecture that…
Training terminal agents at scale requires diverse, verifiable terminal tasks and high-quality interaction trajectories, yet acquiring such data remains a significant challenge. Existing synthesis methods face two key limitations: (1) weak…
The performance of Large Language Models (LLMs) is fundamentally influenced by the distributional composition of multi-domain pre-training data. While manual heuristics were prevalent in early models, they increasingly fail to capture the…
Reflection symmetry detection remains challenging due to interference from asymmetric regions and arbitrary orientations of symmetric patterns. Asymmetric regions introduce background clutter that disrupts symmetric pattern matching,…