Ax Sam Gunn 24d ago

How to sketch a learning algorithm

Data deletion scheme predicting model behavior after training data exclusion. Fast approximation for understanding data influence on learned models.

Ax Ashmal Vayani, Parth Parag Kulkarni, Joseph Fioresi, Song Wang, Mubarak Shah 24d ago

MedRoute: RL-Based Dynamic Specialist Routing in Multi-Agent Medical Diagnosis

Multi-agent system using RL for dynamic specialist routing in medical diagnosis. LMM agents route diagnostic queries to appropriate specialists for precision diagnosis.

Ax Mar Gonz\`alez I Catal\`a, Haitz S\'aez de Oc\'ariz Borde, George D. Monta\~nez, Pietro Li\`o 24d ago

The Stepwise Informativeness Assumption: Why are Entropy Dynamics and Reasoning Correlated in LLMs?

Theoretical analysis explaining why entropy dynamics in LLM internal representations correlate with reasoning correctness. Proposes stepwise informativeness assumption.

Ax Jaehyeok Lee, Xiaoyuan Yi, Jing Yao, Hyunjin Hwang, Roy Ka-Wei Lee, Xing Xie, JinYeong Bak 24d ago

Distributional Open-Ended Evaluation of LLM Cultural Value Alignment Based on Value Codebook

DOVE benchmark evaluates LLM cultural value alignment through open-ended generation. Addresses limitations of discriminative multiple-choice formats and subcultural heterogeneity.

Ax Elena Villalobos (Tecnol\'ogico de Monterrey, Mexico City, Mexico), Adolfo De Un\'anue T. (Tecnol\'ogico de Monterrey, Mexico City, Mexico), Fernanda Sobrino (Tecnol\'ogico de Monterrey, Mexico City, Mexico), David Ak\'e (Tecnol\'ogico de Monterrey, Mexico City, Mexico), Stephany Cisneros (Tecnol\'ogico de Monterrey, Mexico City, Mexico), Jorge Lecona (Container Terminal Operations, Veracruz, Mexico), Alejandra Matadamaz (Container Terminal Operations, Veracruz, Mexico) 24d ago

Toward Reducing Unproductive Container Moves: Predicting Service Requirements and Dwell Times

ML models predict container service requirements and dwell times at port terminals to reduce unproductive moves. Real-world logistics operations case study.

Ax Jiayuan Liu, Barry Wang, Jiarui Gan, Tonghan Wang, Leon Xie, Mingyu Guo, Vincent Conitzer 24d ago

Incentive-Aware Multi-Fidelity Optimization for Generative Advertising in Large Language Models

Multi-fidelity optimization framework combining VCG incentive mechanisms with efficient sampling to optimize LLM advertising while managing advertiser strategic behavior.

Ax Shoaib Sadiq Salehmohamed, Jinal Prashant Thakkar, Hansika Aredla, Shaik Mohammed Omar, Shalmali Ayachit 24d ago

Weakly Supervised Distillation of Hallucination Signals into Transformer Representations

Framework to distill hallucination detection signals into transformer representations during training, enabling inference-time hallucination detection without external verification systems.

Ax Nibedita Roy, Vishal Halder, Gugan Thoppe, Alexandre Reiffers-Masson, Mihir Dhanakshirur, Naman, Alexandre Azor 24d ago

Tight Convergence Rates for Online Distributed Linear Estimation with Adversarial Measurements

Distributed mean estimation with adversarial measurements and asynchronous worker activation. Theoretical convergence rate analysis for parameter-server setup.

Ax Xaver Fink, Borja Fernandez Adiego, Daniele Mirarchi, Eloise Matheson, Alvaro Garcia Gonzales, Gianmarco Ricci, Joost-Pieter Katoen 24d ago

Adversarial Robustness of Time-Series Classification for Crystal Collimator Alignment

CNN adversarial robustness for particle physics beam-loss monitoring at LHC. Analyzes classifier robustness under adversarial inputs during crystal alignment.

Ax Syed Irfan Ali Meerza, Feiyi Wang, Jian Liu 24d ago

FedSpy-LLM: Towards Scalable and Generalizable Data Reconstruction Attacks from Gradients on LLMs

FedSpy-LLM demonstrates data reconstruction attacks on LLMs in federated learning, highlighting privacy risks in gradient sharing.

Ax Parker Ewen, Dmitriy Rivkin, Mario Bijelic, Felix Heide 24d ago

Telescope: Learnable Hyperbolic Foveation for Ultra-Long-Range Object Detection

Telescope uses learnable hyperbolic foveation for detecting objects at ultra-long range (500m+) in autonomous driving scenarios.

Ax Guruprasad Viswanathan Ramesh, Asmit Nayak, Basieem Siddique, Kassem Fawaz 24d ago

WebSP-Eval: Evaluating Web Agents on Website Security and Privacy Tasks

WebSP-Eval benchmarks web agents on website security and privacy task execution, filling gap in agent evaluation frameworks.

Ax Shao Wang, Rui Ren, Lin Gui 24d ago

ForkKV: Scaling Multi-LoRA Agent Serving via Copy-on-Write Disaggregated KV Cache

ForkKV is a system for efficient multi-LoRA agent serving using copy-on-write KV cache disaggregation to reduce memory overhead.

Ax Michael Rizvi-Martel, Guillaume Rabusseau, Marius Mosbach 24d ago

The Illusion of Superposition? A Principled Analysis of Latent Thinking in Language Models

Research analyzing whether latent chain-of-thought reasoning in LLMs actually enables superposition of multiple solutions.

Ax Kranthi Kommuru, Kunal Khanvilkar, Gaurav Parekh 24d ago

ProofSketcher: Hybrid LLM + Lightweight Proof Checker for Reliable Math/Logic Reasoning

ProofSketcher combines LLMs with formal proof verification to improve mathematical and logical reasoning accuracy and reliability.

Ax Yasamin Fayyaz, Li Yang, Khalil El-Khatib 24d ago

Towards Resilient Intrusion Detection in CubeSats: Challenges, TinyML Solutions, and Future Directions

Proposes TinyML-based intrusion detection for CubeSats addressing cybersecurity vulnerabilities from COTS components and open-source software.

Ax Rebecca M. M. Hicke, Sil Hamilton, David Mimno, Ross Deans Kristensen-McLachlan 24d ago

Attention Flows: Tracing LLM Conceptual Engagement via Story Summaries

Evaluates LLM ability to integrate long-form text information through novel summarization task, comparing human and model-authored novel summaries.

Ax Youssef Abdou 24d ago

The Unreasonable Effectiveness of Data for Recommender Systems

Studies how offline recommendation system performance scales with training dataset size and identifies saturation points in data effectiveness.

Ax Shukai Cai, Sourav Dutta, Mark Loveland, Eirik Valseth, Peter Rivera-Casillas, Corey Trahan, Clint Dawson 24d ago

Operator Learning for Surrogate Modeling of Wave-Induced Forces from Sea Surface Waves

Operator learning surrogate model for wave-induced forces as alternative to expensive numerical wave models in storm surge prediction.

Ax Harrison Katz 24d ago

Learning Debt and Cost-Sensitive Bayesian Retraining: A Forecasting Operations Framework

Defines learning debt and actionable staleness metrics, derives Bayes retraining rule for optimal forecasting model retraining schedules.

Ax Yihua Zhang, Hongkang Li, Yuguang Yao, Aochuan Chen, Shuai Zhang, Pin-Yu Chen, Meng Wang, Sijia Liu 24d ago

Visual prompting reimagined: The power of the Activation Prompts

Activation Prompts improve visual prompting for vision model adaptation, closing performance gap between prompting and conventional fine-tuning.

Ax Smita Deb, Zheng-Meng Zhai, Mulugeta Haile, Ying-Cheng Lai 24d ago

Anticipating tipping in spatiotemporal systems with machine learning

Applies reservoir computing to anticipate critical tipping points in complex spatiotemporal dynamical systems via machine learning.

Ax Jiacheng Xie, Hua-Chieh Shao, Can Wu, Ricardo Otazo, Jie Deng, Mu-Han Lin, Tsuicheng Chiu, Jacob Buatti, Viktor Iakovenko, You Zhang 24d ago

Spatiotemporal Gaussian representation-based dynamic reconstruction and motion estimation framework for time-resolved volumetric MR imaging (DREME-GSMR)

DREME-GSMR framework using 3D Gaussian representations for time-resolved dynamic MRI reconstruction without prior anatomical models.

Ax Basil Kyriacou, Mo Kordzanganeh, Maniraman Periyasamy, Alexey Melnikov 24d ago

Soft-Quantum Algorithms

Soft-quantum algorithms combining quantum operations with classical simulation for variational quantum circuits on few-qubit problems.

Ax Asmaa Eldesoukey, Yongxin Chen, Abhishek Halder 24d ago

A Generalized Sinkhorn Algorithm for Mean-Field Schr\"odinger Bridge

Generalized Sinkhorn algorithm for solving mean-field Schrödinger bridge problem in multi-agent systems with nonlocal interactions.

Ax Emre Gurkanli, Michael Spannowsky 24d ago

Quantum-Inspired Tensor Network Autoencoders for Anomaly Detection: A MERA-Based Approach

Tensor-network autoencoder using multiscale MERA architecture for reconstruction-based anomaly detection in particle physics collider jets.

Ax Xiangming Gu, Soham De, Michalis Titsias, Larisa Markeeva, Petar Veli\v{c}kovi\'c, Razvan Pascanu 24d ago

The Illusion of Stochasticity in LLMs

Demonstrates LLMs fail at reliable stochastic sampling required for agentic systems, identifying critical failure point in distribution sampling from inferred data.

Ax Anna\"elle Baiget, Jaron Maene, Seongmin Lee, Benjie Wang, Guy Van den Broeck, Miryung Kim 24d ago

ExplainFuzz: Explainable and Constraint-Conditioned Test Generation with Probabilistic Circuits

ExplainFuzz generates test inputs using probabilistic circuits, improving on grammar-based fuzzers and LLM approaches for constraint-conditioned software testing.

Ax Joshua Castillo, Ravi Mukkamala 24d ago

LLM-based Schema-Guided Extraction and Validation of Missing-Person Intelligence from Heterogeneous Data Sources

Guardian Parser Pack uses LLMs for schema-guided extraction and normalization of missing-person intelligence from heterogeneous investigative documents.

Ax S M Shovan, Arindam Khanda, S M Ferdous, Sajal K. Das, Mahantesh Halappanavar 24d ago

DynLP: Parallel Dynamic Batch Update for Label Propagation in Semi-Supervised Learning

DynLP algorithm for efficient parallel dynamic batch updates in graph-based semi-supervised learning label propagation with incremental data arrival.

Ax Xiao Xiao, Fehmi Cirak 24d ago

Neural parametric representations for thin-shell shape optimisation

Neural parametric representation using MLPs with periodic activation functions for shape optimization of thin-shell structures via gradient-based methods.

Ax Hanyang Wang, Mingxuan Zhu 24d ago

The Detection--Extraction Gap: Models Know the Answer Before They Can Say It

Empirical study showing 52-88% of chain-of-thought tokens in LLMs are generated after answer is already recoverable, revealing the detection-extraction gap.

Ax Yaqi Zhao, Haoliang Sun, Yating Wang, Yongshun Gong, Yilong Yin 24d ago

Holistic Optimal Label Selection for Robust Prompt Learning under Partial Labels

Proposes Holistic Optimal Label Selection (HopS) for prompt learning with partial labels in vision-language models using pre-trained feature encoders.

Ax Napoleon Paxton 24d ago

The Theorems of Dr. David Blackwell and Their Contributions to Artificial Intelligence

Survey of David Blackwell's mathematical theorems (Rao-Blackwell, Approachability, Informativeness) and their foundational relevance to AI.

Ax Zinan Guo, Zihan Wang, Chuan Yan, Liuhuo Wan, Ethan Ma, Guangdong Bai 24d ago

Variational Feature Compression for Model-Specific Representations

Feature compression framework for model-specific representations; prevents cross-model transfer and unauthorized data reuse.

Ax Yifan Zhu, Yihan Wang, Xiao-Shan Gao 24d ago

Towards Robust Content Watermarking Against Removal and Forgery Attacks

Watermarking technique for generated content robust against removal/forgery attacks; addresses copyright protection for diffusion models.

Ax Xueshen Liu, Yongji Wu, Yuncheng Yao, Danyang Zhuo, Ion Stoica, Z. Morley Mao 24d ago

Foundry: Template-Based CUDA Graph Context Materialization for Fast LLM Serving Cold Start

Foundry: CUDA graph template system for fast LLM serving cold start; reduces graph capture time from tens of seconds to milliseconds.

Ax Haoyue Liu, Zhichao Wang, Yongxin Guo, Haoran Shou, Xiaoying Tang 24d ago

Adaptive Prompt Structure Factorization: A Framework for Self-Discovering and Optimizing Compositional Prompt Programs

Adaptive Prompt Structure Factorization: API-only framework using architect model to decompose and optimize compositional prompt programs for LLMs.

Ax Jianhong Pang, Ruoxi Cheng, Ziyi Ye, Xingjun Ma, Zuxuan Wu, Xuanjing Huang, Yu-Gang Jiang 24d ago

Steering the Verifiability of Multimodal AI Hallucinations

Studies multimodal LLM hallucinations, distinguishing obvious from elusive types; proposes steering hallucination verifiability.

Ax Yanan Cao, Ashish Ranjan, Sinduja Subramaniam, Evren Korpeoglu, Kaushiki Nag, Kannan Achan 24d ago

CASE: Cadence-Aware Set Encoding for Large-Scale Next Basket Repurchase Recommendation

CASE: recommendation system using cadence-aware encoding for next-basket repurchase prediction in retail.

Ax Yuheng Zhang, Claire Chen, Nan Jiang 24d ago

Beyond Pessimism: Offline Learning in KL-regularized Games

Pessimism-free algorithm for offline learning in KL-regularized two-player zero-sum games with improved statistical rates.

Ax Kanta Yoshioka, Soshi Hirayae, Yuichiro Tanaka, Yuichi Katori, Takashi Morie, Hakaru Tamukoh 24d ago

CBM-Dual: A 65-nm Fully Connected Chaotic Boltzmann Machine Processor for Dual Function Simulated Annealing and Reservoir Computing

CBM-Dual: silicon processor implementing chaotic Boltzmann machines for simulated annealing and reservoir computing at edge.

Ax Shunan Zhu, Jiawei Chen, Yonghao Yu, Hideya Ochiai 24d ago

FedDetox: Robust Federated SLM Alignment via On-Device Data Sanitization

FedDetox: federated learning framework for small language model alignment with on-device data sanitization against toxic/poisoned data.

Ax Daniele Fossem\`o, Filippo Mignosi, Giuseppe Placidi, Luca Raggioli, Matteo Spezialetti, Fabio Aurelio D'Asaro 24d ago

Explaining Neural Networks in Preference Learning: a Post-hoc Inductive Logic Programming Approach

Uses inductive logic programming (ILASP) to approximate neural networks for preference learning; creates dataset of recipe preferences.

Ax Tom Devynck Bilal Faye Djamel Bouchaffra Nadjib Lazaar Hanane Azzag Mustapha Lebbah 24d ago

Energy-Regularized Spatial Masking: A Novel Approach to Enhancing Robustness and Interpretability in Vision Models

Energy-Regularized Spatial Masking (ERSM): feature selection framework improving robustness and interpretability of vision models via energy regularization.

Ax Md Abu Ahammed Babu, Sushant Kumar Pandey, Darko Durisic, Andras Balint, Miroslaw Staron 24d ago

Data Leakage in Automotive Perception: Practitioners' Insights

Empirical study of how automotive industry practitioners perceive, detect, and manage data leakage between training and evaluation datasets.

Ax Yi-Shuai Niu 24d ago

Continuous-Time Dynamics of the Difference-of-Convex Algorithm

Continuous-time analysis of difference-of-convex algorithm showing equivalence to explicit Euler discretization with application to optimization theory.

Ax Simon Calderon, Niklas Johansson, Onur G\"unl\"u 24d ago

Evaluating PQC KEMs, Combiners, and Cascade Encryption via Adaptive IND-CPA Testing Using Deep Learning

Deep learning approach to empirically validate post-quantum cryptography KEMs and hybrid constructions via IND-CPA adaptive testing.

Ax Zhida Jiang, Zhaolong Xing, Huichao Chai, Tianxing Sun, Qiang Peng, Baopeng Yuan, Jiaxing Wang, Hua Du, Zhixin Wu, Xuemiao Li, Yikui Cao, Xinyu Liu, Yongxiang Feng, Zhen Chen, Ke Zhang 24d ago

NestPipe: Large-Scale Recommendation Training on 1,500+ Accelerators via Nested Pipelining

NestPipe: decentralized embedding training framework for trillion-parameter recommendation models at 1,500+ accelerator scale with nested pipelining.

Ax Mohamed Rabie, Chinthana Panagamuwa, Konstantinos G. Kyriakopoulos 24d ago

ELC: Evidential Lifelong Classifier for Uncertainty Aware Radar Pulse Classification

ELC: evidential lifelong classifier combining uncertainty quantification with continual learning for radar pulse classification with confidence estimation.