Papers
Research papers from arXiv and related sources
CAFE: Channel-Autoregressive Factorized Encoding for Robust Biosignal Spatial Super-Resolution
High-density biosignal recordings are critical for neural decoding and clinical monitoring, yet real-world deployments often rely on low-density (LD) montages due to hardware and operational constr...
Hongjun Liu, Leyu Zhou, Zijianghao Yang, Rujun Han, Shitong Duan, Kuanjian Tang, Chao Yao
Central limit theorem for linear eigenvalue statistics of random geometric graphs
Random spatial networks-that is, graphs whose connectivity is governed by geometric proximity-have emerged as fundamental models for systems constrained by an underlying spatial structure. A protot...
Christian Hirsch, Kyeongsik Nam, Moritz Otto
Dynamic Delayed Tree Expansion For Improved Multi-Path Speculative Decoding
Multi-path speculative decoding accelerates lossless sampling from a target model by using a cheaper draft model to generate a draft tree of tokens, and then applies a verification algorithm that a...
Rahul Thomas, Teo Kitanovski, Micah Goldblum, Arka Pal
Radiological mapping and uncertainty quantification by a fast Microcanonical Langevin Monte Carlo sampler
Radiological mapping plays a critical role in nuclear emergency response and environmental management activities. A radiation image, representing the spatial and intensity distribution of the radio...
Lei Pan, Jaewon Lee, Brian J. Quiter, Jakob Robnik, Uroš Seljak, Jayson R. Vavrek
Fundamental Limits of Black-Box Safety Evaluation: Information-Theoretic and Computational Barriers from Latent Context Conditioning
Black-box safety evaluation of AI systems assumes model behavior on test distributions reliably predicts deployment performance. We formalize and challenge this assumption through latent context-co...
Vishal Srivastava
Mason: Type- and Name-Guided Program Synthesis
Object-oriented programs tend to be written using many common coding idioms, such as those captured by design patterns. While design patterns are useful, implementing them is often tedious and repe...
Jasper Geer, Fox Huston, Jeffrey S. Foster
Lies, Labels, and Mechanisms
We test whether lying aversion can steer equilibrium selection in mechanism design. In a principal-worker environment, the direct mechanism admits two dominant-strategy equilibria: the designer's t...
Alex L. Brown, Ethan Park, Rodrigo A. Velez
DDiT: Dynamic Patch Scheduling for Efficient Diffusion Transformers
Diffusion Transformers (DiTs) have achieved state-of-the-art performance in image and video generation, but their success comes at the cost of heavy computation. This inefficiency is largely due to...
Dahye Kim, Deepti Ghadiyaram, Raghudeep Gadde
Greedy Multi-Path Block Verification for Faster Decoding in Speculative Sampling
The goal of $L$-step speculative decoding is to accelerate autoregressive decoding of a target model by using a cheaper draft model to generate a candidate path of $L$ tokens. Based on a verificati...
Rahul Thomas, Arka Pal
Neural Proposals, Symbolic Guarantees: Neuro-Symbolic Graph Generation with Hard Constraints
We challenge black-box purely deep neural approaches for molecules and graph generation, which are limited in controllability and lack formal guarantees. We introduce Neuro-Symbolic Graph Generativ...
Chuqin Geng, Li Zhang, Mark Zhang, Haolin Ye, Ziyu Zhao, Xujie Si
LLM4Cov: Execution-Aware Agentic Learning for High-coverage Testbench Generation
Execution-aware LLM agents offer a promising paradigm for learning from tool feedback, but such feedback is often expensive and slow to obtain, making online reinforcement learning (RL) impractical...
Hejia Zhang, Zhongming Yu, Chia-Tung Ho, Haoxing Ren, Brucek Khailany, Jishen Zhao
Exact Certification of Data-Poisoning Attacks Using Mixed-Integer Programming
This work introduces a verification framework that provides both sound and complete guarantees for data poisoning attacks during neural network training. We formulate adversarial data manipulation,...
Philip Sosnin, Jodie Knapp, Fraser Kennedy, Josh Collyer, Calvin Tsay
Mind the GAP: Text Safety Does Not Transfer to Tool-Call Safety in LLM Agents
Large language models deployed as agents increasingly interact with external systems through tool calls--actions with real-world consequences that text outputs alone do not carry. Safety evaluation...
Arnold Cartagena, Ariane Teixeira
Nudging Attention to Workplace Meeting Goals: A Large-Scale, Preregistered Field Experiment
Ineffective meetings are pervasive. Thinking ahead explicitly about meeting goals may improve effectiveness, but current collaboration platforms lack integrated support. We tested a lightweight goa...
Lev Tankelevitch, Ava Elizabeth Scott, Nagaravind Challakere, Payod Panda, Sean Rintel
ConvApparel: A Benchmark Dataset and Validation Framework for User Simulators in Conversational Recommenders
The promise of LLM-based user simulators to improve conversational AI is hindered by a critical "realism gap," leading to systems that are optimized for simulated interactions, but may fail to perf...
Ofer Meshi, Krisztian Balog, Sally Goldman, Avi Caciularu, Guy Tennenholtz, Jihwan Jeong, Amir Gl...
Free Quantum Computing
Quantum computing improves substantially on known classical algorithms for various important problems, but the nature of the relationship between quantum and classical computing is not yet fully un...
Jacques Carette, Chris Heunen, Robin Kaarsgaard, Neil J. Ross, Amr Sabry
Stellar Paternity Tests: Matching High-Latitude B Stars to the Open Clusters of their Birth
OB stars generally form in open clusters within the Milky Way's thin disk, so when they are found at high Galactic latitudes, it is thought that they were ejected from their birth clusters during t...
Brandon Schweers, M. Virginia McSwain
A statistical perspective on transformers for small longitudinal cohort data
Modeling of longitudinal cohort data typically involves complex temporal dependencies between multiple variables. There, the transformer architecture, which has been highly successful in language a...
Kiana Farhadyar, Maren Hackenberg, Kira Ahrens, Charlotte Schenk, Bianca Kollmann, Oliver Tüscher...
AgentLAB: Benchmarking LLM Agents against Long-Horizon Attacks
LLM agents are increasingly deployed in long-horizon, complex environments to solve challenging problems, but this expansion exposes them to long-horizon attacks that exploit multi-turn user-agent-...
Tanqiu Jiang, Yuhui Wang, Jiacheng Liang, Ting Wang
Domain Decomposition for Mean Curvature Flow of Surface Polygonal Meshes
We examine the use of domain decomposition for potentially more efficient mean curvature flow of surface meshes, whose faces are arbitrary simple polygons. We first test traditional domain decompos...
Lenka Ptackova, Michal Outrata